2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* Copyright (C) 1991, 1992 Linus Torvalds
|
|
|
|
* Copyright (C) 1994, Karl Keyte: Added support for disk statistics
|
|
|
|
* Elevator latency, (C) 2000 Andrea Arcangeli <andrea@suse.de> SuSE
|
|
|
|
* Queue request tables / lock, selectable elevator, Jens Axboe <axboe@suse.de>
|
|
|
|
* kernel-doc documentation started by NeilBrown <neilb@cse.unsw.edu.au> - July2000
|
|
|
|
* bio rewrite, highmem i/o, etc, Jens Axboe <axboe@suse.de> - may 2001
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This handles all read/write requests to block devices
|
|
|
|
*/
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/backing-dev.h>
|
|
|
|
#include <linux/bio.h>
|
|
|
|
#include <linux/blkdev.h>
|
|
|
|
#include <linux/highmem.h>
|
|
|
|
#include <linux/mm.h>
|
|
|
|
#include <linux/kernel_stat.h>
|
|
|
|
#include <linux/string.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/bootmem.h> /* for max_pfn/max_low_pfn */
|
|
|
|
#include <linux/completion.h>
|
|
|
|
#include <linux/slab.h>
|
|
|
|
#include <linux/swap.h>
|
|
|
|
#include <linux/writeback.h>
|
2006-12-10 11:19:35 +01:00
|
|
|
#include <linux/task_io_accounting_ops.h>
|
2006-01-09 16:02:34 +01:00
|
|
|
#include <linux/interrupt.h>
|
|
|
|
#include <linux/cpu.h>
|
2006-03-23 20:00:26 +01:00
|
|
|
#include <linux/blktrace_api.h>
|
2006-12-08 11:39:46 +01:00
|
|
|
#include <linux/fault-inject.h>
|
2007-09-21 10:44:19 +02:00
|
|
|
#include <linux/scatterlist.h>
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* for max sense size
|
|
|
|
*/
|
|
|
|
#include <scsi/scsi_cmnd.h>
|
|
|
|
|
2006-11-22 15:55:48 +01:00
|
|
|
static void blk_unplug_work(struct work_struct *work);
|
2005-04-17 00:20:36 +02:00
|
|
|
static void blk_unplug_timeout(unsigned long data);
|
2007-10-23 15:05:46 +02:00
|
|
|
static void drive_stat_acct(struct request *rq, int new_io);
|
2006-01-06 09:49:58 +01:00
|
|
|
static void init_request_from_bio(struct request *req, struct bio *bio);
|
2007-07-24 09:28:11 +02:00
|
|
|
static int __make_request(struct request_queue *q, struct bio *bio);
|
2006-07-19 23:39:40 +02:00
|
|
|
static struct io_context *current_io_context(gfp_t gfp_flags, int node);
|
2007-08-16 13:27:52 +02:00
|
|
|
static void blk_recalc_rq_segments(struct request *rq);
|
2007-08-16 13:31:28 +02:00
|
|
|
static void blk_rq_bio_prep(struct request_queue *q, struct request *rq,
|
|
|
|
struct bio *bio);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* For the allocated request tables
|
|
|
|
*/
|
2006-12-07 05:33:20 +01:00
|
|
|
static struct kmem_cache *request_cachep;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* For queue allocation
|
|
|
|
*/
|
2006-12-07 05:33:20 +01:00
|
|
|
static struct kmem_cache *requestq_cachep;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* For io context allocations
|
|
|
|
*/
|
2006-12-07 05:33:20 +01:00
|
|
|
static struct kmem_cache *iocontext_cachep;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Controlling structure to kblockd
|
|
|
|
*/
|
2006-01-09 16:02:34 +01:00
|
|
|
static struct workqueue_struct *kblockd_workqueue;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
unsigned long blk_max_low_pfn, blk_max_pfn;
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_max_low_pfn);
|
|
|
|
EXPORT_SYMBOL(blk_max_pfn);
|
|
|
|
|
2006-01-09 16:02:34 +01:00
|
|
|
static DEFINE_PER_CPU(struct list_head, blk_cpu_done);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/* Amount of time in which a process may batch requests */
|
|
|
|
#define BLK_BATCH_TIME (HZ/50UL)
|
|
|
|
|
|
|
|
/* Number of requests a "batching" process may submit */
|
|
|
|
#define BLK_BATCH_REQ 32
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Return the threshold (number of used requests) at which the queue is
|
|
|
|
* considered to be congested. It include a little hysteresis to keep the
|
|
|
|
* context switch rate down.
|
|
|
|
*/
|
|
|
|
static inline int queue_congestion_on_threshold(struct request_queue *q)
|
|
|
|
{
|
|
|
|
return q->nr_congestion_on;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The threshold at which a queue is considered to be uncongested
|
|
|
|
*/
|
|
|
|
static inline int queue_congestion_off_threshold(struct request_queue *q)
|
|
|
|
{
|
|
|
|
return q->nr_congestion_off;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void blk_queue_congestion_threshold(struct request_queue *q)
|
|
|
|
{
|
|
|
|
int nr;
|
|
|
|
|
|
|
|
nr = q->nr_requests - (q->nr_requests / 8) + 1;
|
|
|
|
if (nr > q->nr_requests)
|
|
|
|
nr = q->nr_requests;
|
|
|
|
q->nr_congestion_on = nr;
|
|
|
|
|
|
|
|
nr = q->nr_requests - (q->nr_requests / 8) - (q->nr_requests / 16) - 1;
|
|
|
|
if (nr < 1)
|
|
|
|
nr = 1;
|
|
|
|
q->nr_congestion_off = nr;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_get_backing_dev_info - get the address of a queue's backing_dev_info
|
|
|
|
* @bdev: device
|
|
|
|
*
|
|
|
|
* Locates the passed device's request queue and returns the address of its
|
|
|
|
* backing_dev_info
|
|
|
|
*
|
|
|
|
* Will return NULL if the request queue cannot be located.
|
|
|
|
*/
|
|
|
|
struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev)
|
|
|
|
{
|
|
|
|
struct backing_dev_info *ret = NULL;
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q = bdev_get_queue(bdev);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (q)
|
|
|
|
ret = &q->backing_dev_info;
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_get_backing_dev_info);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_prep_rq - set a prepare_request function for queue
|
|
|
|
* @q: queue
|
|
|
|
* @pfn: prepare_request function
|
|
|
|
*
|
|
|
|
* It's possible for a queue to register a prepare_request callback which
|
|
|
|
* is invoked before the request is handed to the request_fn. The goal of
|
|
|
|
* the function is to prepare a request for I/O, it can be used to build a
|
|
|
|
* cdb from the request data for instance.
|
|
|
|
*
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_prep_rq(struct request_queue *q, prep_rq_fn *pfn)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
q->prep_rq_fn = pfn;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_prep_rq);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_merge_bvec - set a merge_bvec function for queue
|
|
|
|
* @q: queue
|
|
|
|
* @mbfn: merge_bvec_fn
|
|
|
|
*
|
|
|
|
* Usually queues have static limitations on the max sectors or segments that
|
|
|
|
* we can put in a request. Stacking drivers may have some settings that
|
|
|
|
* are dynamic, and thus we have to query the queue whether it is ok to
|
|
|
|
* add a new bio_vec to a bio at a given offset or not. If the block device
|
|
|
|
* has such limitations, it needs to register a merge_bvec_fn to control
|
|
|
|
* the size of bio's sent to it. Note that a block device *must* allow a
|
|
|
|
* single page to be added to an empty bio. The block device driver may want
|
|
|
|
* to use the bio_split() function to deal with these bio's. By default
|
|
|
|
* no merge_bvec_fn is defined for a queue, and only the fixed limits are
|
|
|
|
* honored.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_merge_bvec(struct request_queue *q, merge_bvec_fn *mbfn)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
q->merge_bvec_fn = mbfn;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_merge_bvec);
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_softirq_done(struct request_queue *q, softirq_done_fn *fn)
|
2006-01-09 16:02:34 +01:00
|
|
|
{
|
|
|
|
q->softirq_done_fn = fn;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_softirq_done);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/**
|
|
|
|
* blk_queue_make_request - define an alternate make_request function for a device
|
|
|
|
* @q: the request queue for the device to be affected
|
|
|
|
* @mfn: the alternate make_request function
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* The normal way for &struct bios to be passed to a device
|
|
|
|
* driver is for them to be collected into requests on a request
|
|
|
|
* queue, and then to allow the device driver to select requests
|
|
|
|
* off that queue when it is ready. This works well for many block
|
|
|
|
* devices. However some block devices (typically virtual devices
|
|
|
|
* such as md or lvm) do not benefit from the processing on the
|
|
|
|
* request queue, and are served best by having the requests passed
|
|
|
|
* directly to them. This can be achieved by providing a function
|
|
|
|
* to blk_queue_make_request().
|
|
|
|
*
|
|
|
|
* Caveat:
|
|
|
|
* The driver that does this *must* be able to deal appropriately
|
|
|
|
* with buffers in "highmemory". This can be accomplished by either calling
|
|
|
|
* __bio_kmap_atomic() to get a temporary kernel mapping, or by calling
|
|
|
|
* blk_queue_bounce() to create a buffer in normal memory.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_make_request(struct request_queue * q, make_request_fn * mfn)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
/*
|
|
|
|
* set defaults
|
|
|
|
*/
|
|
|
|
q->nr_requests = BLKDEV_MAX_RQ;
|
2005-09-07 00:17:47 +02:00
|
|
|
blk_queue_max_phys_segments(q, MAX_PHYS_SEGMENTS);
|
|
|
|
blk_queue_max_hw_segments(q, MAX_HW_SEGMENTS);
|
2005-04-17 00:20:36 +02:00
|
|
|
q->make_request_fn = mfn;
|
|
|
|
q->backing_dev_info.ra_pages = (VM_MAX_READAHEAD * 1024) / PAGE_CACHE_SIZE;
|
|
|
|
q->backing_dev_info.state = 0;
|
|
|
|
q->backing_dev_info.capabilities = BDI_CAP_MAP_COPY;
|
2005-12-05 09:37:06 +01:00
|
|
|
blk_queue_max_sectors(q, SAFE_MAX_SECTORS);
|
2005-04-17 00:20:36 +02:00
|
|
|
blk_queue_hardsect_size(q, 512);
|
|
|
|
blk_queue_dma_alignment(q, 511);
|
|
|
|
blk_queue_congestion_threshold(q);
|
|
|
|
q->nr_batching = BLK_BATCH_REQ;
|
|
|
|
|
|
|
|
q->unplug_thresh = 4; /* hmm */
|
|
|
|
q->unplug_delay = (3 * HZ) / 1000; /* 3 milliseconds */
|
|
|
|
if (q->unplug_delay == 0)
|
|
|
|
q->unplug_delay = 1;
|
|
|
|
|
2006-11-22 15:55:48 +01:00
|
|
|
INIT_WORK(&q->unplug_work, blk_unplug_work);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
q->unplug_timer.function = blk_unplug_timeout;
|
|
|
|
q->unplug_timer.data = (unsigned long)q;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* by default assume old behaviour and bounce for any highmem page
|
|
|
|
*/
|
|
|
|
blk_queue_bounce_limit(q, BLK_BOUNCE_HIGH);
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_make_request);
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static void rq_init(struct request_queue *q, struct request *rq)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
INIT_LIST_HEAD(&rq->queuelist);
|
2006-01-09 16:02:34 +01:00
|
|
|
INIT_LIST_HEAD(&rq->donelist);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
rq->errors = 0;
|
|
|
|
rq->bio = rq->biotail = NULL;
|
2006-07-13 11:55:04 +02:00
|
|
|
INIT_HLIST_NODE(&rq->hash);
|
|
|
|
RB_CLEAR_NODE(&rq->rb_node);
|
2005-06-27 10:55:12 +02:00
|
|
|
rq->ioprio = 0;
|
2005-04-17 00:20:36 +02:00
|
|
|
rq->buffer = NULL;
|
|
|
|
rq->ref_count = 1;
|
|
|
|
rq->q = q;
|
|
|
|
rq->special = NULL;
|
|
|
|
rq->data_len = 0;
|
|
|
|
rq->data = NULL;
|
2005-06-20 14:04:44 +02:00
|
|
|
rq->nr_phys_segments = 0;
|
2005-04-17 00:20:36 +02:00
|
|
|
rq->sense = NULL;
|
|
|
|
rq->end_io = NULL;
|
|
|
|
rq->end_io_data = NULL;
|
2006-01-09 16:02:34 +01:00
|
|
|
rq->completion_data = NULL;
|
2007-07-16 08:52:14 +02:00
|
|
|
rq->next_rq = NULL;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_ordered - does this queue support ordered writes
|
2006-01-06 09:51:03 +01:00
|
|
|
* @q: the request queue
|
|
|
|
* @ordered: one of QUEUE_ORDERED_*
|
2006-01-31 15:24:34 +01:00
|
|
|
* @prepare_flush_fn: rq setup helper for cache flush ordered writes
|
2005-04-17 00:20:36 +02:00
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* For journalled file systems, doing ordered writes on a commit
|
|
|
|
* block instead of explicitly doing wait_on_buffer (which is bad
|
|
|
|
* for performance) can be a big win. Block drivers supporting this
|
|
|
|
* feature should call this function and indicate so.
|
|
|
|
*
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
int blk_queue_ordered(struct request_queue *q, unsigned ordered,
|
2006-01-06 09:51:03 +01:00
|
|
|
prepare_flush_fn *prepare_flush_fn)
|
|
|
|
{
|
|
|
|
if (ordered & (QUEUE_ORDERED_PREFLUSH | QUEUE_ORDERED_POSTFLUSH) &&
|
|
|
|
prepare_flush_fn == NULL) {
|
|
|
|
printk(KERN_ERR "blk_queue_ordered: prepare_flush_fn required\n");
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ordered != QUEUE_ORDERED_NONE &&
|
|
|
|
ordered != QUEUE_ORDERED_DRAIN &&
|
|
|
|
ordered != QUEUE_ORDERED_DRAIN_FLUSH &&
|
|
|
|
ordered != QUEUE_ORDERED_DRAIN_FUA &&
|
|
|
|
ordered != QUEUE_ORDERED_TAG &&
|
|
|
|
ordered != QUEUE_ORDERED_TAG_FLUSH &&
|
|
|
|
ordered != QUEUE_ORDERED_TAG_FUA) {
|
|
|
|
printk(KERN_ERR "blk_queue_ordered: bad value %d\n", ordered);
|
|
|
|
return -EINVAL;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2006-01-06 09:51:03 +01:00
|
|
|
|
2006-01-24 10:34:36 +01:00
|
|
|
q->ordered = ordered;
|
2006-01-06 09:51:03 +01:00
|
|
|
q->next_ordered = ordered;
|
|
|
|
q->prepare_flush_fn = prepare_flush_fn;
|
|
|
|
|
|
|
|
return 0;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_ordered);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Cache flushing for ordered writes handling
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
inline unsigned blk_ordered_cur_seq(struct request_queue *q)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-01-06 09:51:03 +01:00
|
|
|
if (!q->ordseq)
|
|
|
|
return 0;
|
|
|
|
return 1 << ffz(q->ordseq);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
unsigned blk_ordered_req_seq(struct request *rq)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q = rq->q;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
BUG_ON(q->ordseq == 0);
|
2005-10-20 16:23:44 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
if (rq == &q->pre_flush_rq)
|
|
|
|
return QUEUE_ORDSEQ_PREFLUSH;
|
|
|
|
if (rq == &q->bar_rq)
|
|
|
|
return QUEUE_ORDSEQ_BAR;
|
|
|
|
if (rq == &q->post_flush_rq)
|
|
|
|
return QUEUE_ORDSEQ_POSTFLUSH;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-06-15 13:24:28 +02:00
|
|
|
/*
|
|
|
|
* !fs requests don't need to follow barrier ordering. Always
|
|
|
|
* put them at the front. This fixes the following deadlock.
|
|
|
|
*
|
|
|
|
* http://thread.gmane.org/gmane.linux.kernel/537473
|
|
|
|
*/
|
|
|
|
if (!blk_fs_request(rq))
|
|
|
|
return QUEUE_ORDSEQ_DRAIN;
|
|
|
|
|
2006-08-10 08:44:47 +02:00
|
|
|
if ((rq->cmd_flags & REQ_ORDERED_COLOR) ==
|
|
|
|
(q->orig_bar_rq->cmd_flags & REQ_ORDERED_COLOR))
|
2006-01-06 09:51:03 +01:00
|
|
|
return QUEUE_ORDSEQ_DRAIN;
|
|
|
|
else
|
|
|
|
return QUEUE_ORDSEQ_DONE;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_ordered_complete_seq(struct request_queue *q, unsigned seq, int error)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-01-06 09:51:03 +01:00
|
|
|
struct request *rq;
|
|
|
|
int uptodate;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
if (error && !q->orderr)
|
|
|
|
q->orderr = error;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
BUG_ON(q->ordseq & seq);
|
|
|
|
q->ordseq |= seq;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
if (blk_ordered_cur_seq(q) != QUEUE_ORDSEQ_DONE)
|
|
|
|
return;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
2006-01-06 09:51:03 +01:00
|
|
|
* Okay, sequence complete.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
2007-07-18 13:13:10 +02:00
|
|
|
uptodate = 1;
|
|
|
|
if (q->orderr)
|
|
|
|
uptodate = q->orderr;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
q->ordseq = 0;
|
2007-07-18 13:13:10 +02:00
|
|
|
rq = q->orig_bar_rq;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
end_that_request_first(rq, uptodate, rq->hard_nr_sectors);
|
|
|
|
end_that_request_last(rq, uptodate);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
static void pre_flush_end_io(struct request *rq, int error)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-01-06 09:51:03 +01:00
|
|
|
elv_completed_request(rq->q, rq);
|
|
|
|
blk_ordered_complete_seq(rq->q, QUEUE_ORDSEQ_PREFLUSH, error);
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
static void bar_end_io(struct request *rq, int error)
|
|
|
|
{
|
|
|
|
elv_completed_request(rq->q, rq);
|
|
|
|
blk_ordered_complete_seq(rq->q, QUEUE_ORDSEQ_BAR, error);
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
static void post_flush_end_io(struct request *rq, int error)
|
|
|
|
{
|
|
|
|
elv_completed_request(rq->q, rq);
|
|
|
|
blk_ordered_complete_seq(rq->q, QUEUE_ORDSEQ_POSTFLUSH, error);
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static void queue_flush(struct request_queue *q, unsigned which)
|
2006-01-06 09:51:03 +01:00
|
|
|
{
|
|
|
|
struct request *rq;
|
|
|
|
rq_end_io_fn *end_io;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
if (which == QUEUE_ORDERED_PREFLUSH) {
|
|
|
|
rq = &q->pre_flush_rq;
|
|
|
|
end_io = pre_flush_end_io;
|
|
|
|
} else {
|
|
|
|
rq = &q->post_flush_rq;
|
|
|
|
end_io = post_flush_end_io;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2006-01-06 09:51:03 +01:00
|
|
|
|
2006-08-10 08:44:47 +02:00
|
|
|
rq->cmd_flags = REQ_HARDBARRIER;
|
2006-01-06 09:51:03 +01:00
|
|
|
rq_init(q, rq);
|
|
|
|
rq->elevator_private = NULL;
|
2006-09-30 20:29:12 +02:00
|
|
|
rq->elevator_private2 = NULL;
|
2006-01-06 09:51:03 +01:00
|
|
|
rq->rq_disk = q->bar_rq.rq_disk;
|
|
|
|
rq->end_io = end_io;
|
|
|
|
q->prepare_flush_fn(q, rq);
|
|
|
|
|
2006-02-08 10:01:31 +01:00
|
|
|
elv_insert(q, rq, ELEVATOR_INSERT_FRONT);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static inline struct request *start_ordered(struct request_queue *q,
|
2006-01-06 09:51:03 +01:00
|
|
|
struct request *rq)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-01-06 09:51:03 +01:00
|
|
|
q->orderr = 0;
|
|
|
|
q->ordered = q->next_ordered;
|
|
|
|
q->ordseq |= QUEUE_ORDSEQ_STARTED;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Prep proxy barrier request.
|
|
|
|
*/
|
|
|
|
blkdev_dequeue_request(rq);
|
|
|
|
q->orig_bar_rq = rq;
|
|
|
|
rq = &q->bar_rq;
|
2006-08-10 08:44:47 +02:00
|
|
|
rq->cmd_flags = 0;
|
2006-01-06 09:51:03 +01:00
|
|
|
rq_init(q, rq);
|
2006-08-10 08:44:47 +02:00
|
|
|
if (bio_data_dir(q->orig_bar_rq->bio) == WRITE)
|
|
|
|
rq->cmd_flags |= REQ_RW;
|
2007-07-18 13:13:10 +02:00
|
|
|
if (q->ordered & QUEUE_ORDERED_FUA)
|
|
|
|
rq->cmd_flags |= REQ_FUA;
|
2006-01-06 09:51:03 +01:00
|
|
|
rq->elevator_private = NULL;
|
2006-09-30 20:29:12 +02:00
|
|
|
rq->elevator_private2 = NULL;
|
2006-01-06 09:51:03 +01:00
|
|
|
init_request_from_bio(rq, q->orig_bar_rq->bio);
|
|
|
|
rq->end_io = bar_end_io;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Queue ordered sequence. As we stack them at the head, we
|
|
|
|
* need to queue in reverse order. Note that we rely on that
|
|
|
|
* no fs request uses ELEVATOR_INSERT_FRONT and thus no fs
|
2007-09-27 13:01:25 +02:00
|
|
|
* request gets inbetween ordered sequence. If this request is
|
|
|
|
* an empty barrier, we don't need to do a postflush ever since
|
|
|
|
* there will be no data written between the pre and post flush.
|
|
|
|
* Hence a single flush will suffice.
|
2006-01-06 09:51:03 +01:00
|
|
|
*/
|
2007-09-27 13:01:25 +02:00
|
|
|
if ((q->ordered & QUEUE_ORDERED_POSTFLUSH) && !blk_empty_barrier(rq))
|
2006-01-06 09:51:03 +01:00
|
|
|
queue_flush(q, QUEUE_ORDERED_POSTFLUSH);
|
|
|
|
else
|
|
|
|
q->ordseq |= QUEUE_ORDSEQ_POSTFLUSH;
|
|
|
|
|
2006-02-08 10:01:31 +01:00
|
|
|
elv_insert(q, rq, ELEVATOR_INSERT_FRONT);
|
2006-01-06 09:51:03 +01:00
|
|
|
|
|
|
|
if (q->ordered & QUEUE_ORDERED_PREFLUSH) {
|
|
|
|
queue_flush(q, QUEUE_ORDERED_PREFLUSH);
|
|
|
|
rq = &q->pre_flush_rq;
|
|
|
|
} else
|
|
|
|
q->ordseq |= QUEUE_ORDSEQ_PREFLUSH;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
if ((q->ordered & QUEUE_ORDERED_TAG) || q->in_flight == 0)
|
|
|
|
q->ordseq |= QUEUE_ORDSEQ_DRAIN;
|
|
|
|
else
|
|
|
|
rq = NULL;
|
|
|
|
|
|
|
|
return rq;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
int blk_do_ordered(struct request_queue *q, struct request **rqp)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-02-05 08:27:38 +01:00
|
|
|
struct request *rq = *rqp;
|
2007-09-27 13:01:25 +02:00
|
|
|
const int is_barrier = blk_fs_request(rq) && blk_barrier_rq(rq);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
if (!q->ordseq) {
|
|
|
|
if (!is_barrier)
|
|
|
|
return 1;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
if (q->next_ordered != QUEUE_ORDERED_NONE) {
|
|
|
|
*rqp = start_ordered(q, rq);
|
|
|
|
return 1;
|
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* This can happen when the queue switches to
|
|
|
|
* ORDERED_NONE while this request is on it.
|
|
|
|
*/
|
|
|
|
blkdev_dequeue_request(rq);
|
|
|
|
end_that_request_first(rq, -EOPNOTSUPP,
|
|
|
|
rq->hard_nr_sectors);
|
|
|
|
end_that_request_last(rq, -EOPNOTSUPP);
|
|
|
|
*rqp = NULL;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-02-05 08:27:38 +01:00
|
|
|
/*
|
|
|
|
* Ordered sequence in progress
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* Special requests are not subject to ordering rules. */
|
|
|
|
if (!blk_fs_request(rq) &&
|
|
|
|
rq != &q->pre_flush_rq && rq != &q->post_flush_rq)
|
|
|
|
return 1;
|
|
|
|
|
2006-01-06 09:51:03 +01:00
|
|
|
if (q->ordered & QUEUE_ORDERED_TAG) {
|
2006-02-05 08:27:38 +01:00
|
|
|
/* Ordered by tag. Blocking the next barrier is enough. */
|
2006-01-06 09:51:03 +01:00
|
|
|
if (is_barrier && rq != &q->bar_rq)
|
|
|
|
*rqp = NULL;
|
2006-02-05 08:27:38 +01:00
|
|
|
} else {
|
|
|
|
/* Ordered by draining. Wait for turn. */
|
|
|
|
WARN_ON(blk_ordered_req_seq(rq) < blk_ordered_cur_seq(q));
|
|
|
|
if (blk_ordered_req_seq(rq) > blk_ordered_cur_seq(q))
|
|
|
|
*rqp = NULL;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2007-09-27 12:46:13 +02:00
|
|
|
static void req_bio_endio(struct request *rq, struct bio *bio,
|
|
|
|
unsigned int nbytes, int error)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q = rq->q;
|
2006-01-06 09:51:03 +01:00
|
|
|
|
2007-09-27 12:46:13 +02:00
|
|
|
if (&q->bar_rq != rq) {
|
|
|
|
if (error)
|
|
|
|
clear_bit(BIO_UPTODATE, &bio->bi_flags);
|
|
|
|
else if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
|
|
|
|
error = -EIO;
|
2006-01-06 09:51:03 +01:00
|
|
|
|
2007-09-27 12:46:13 +02:00
|
|
|
if (unlikely(nbytes > bio->bi_size)) {
|
|
|
|
printk("%s: want %u bytes done, only %u left\n",
|
|
|
|
__FUNCTION__, nbytes, bio->bi_size);
|
|
|
|
nbytes = bio->bi_size;
|
|
|
|
}
|
2006-01-06 09:51:03 +01:00
|
|
|
|
2007-09-27 12:46:13 +02:00
|
|
|
bio->bi_size -= nbytes;
|
|
|
|
bio->bi_sector += (nbytes >> 9);
|
|
|
|
if (bio->bi_size == 0)
|
2007-09-27 12:47:43 +02:00
|
|
|
bio_endio(bio, error);
|
2007-09-27 12:46:13 +02:00
|
|
|
} else {
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Okay, this is the barrier request in progress, just
|
|
|
|
* record the error;
|
|
|
|
*/
|
|
|
|
if (error && !q->orderr)
|
|
|
|
q->orderr = error;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_bounce_limit - set bounce buffer limit for queue
|
|
|
|
* @q: the request queue for the device
|
|
|
|
* @dma_addr: bus address limit
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Different hardware can have different requirements as to what pages
|
|
|
|
* it can do I/O directly to. A low level driver can call
|
|
|
|
* blk_queue_bounce_limit to have lower memory pages allocated as bounce
|
2006-03-09 02:57:26 +01:00
|
|
|
* buffers for doing I/O to pages residing above @page.
|
2005-04-17 00:20:36 +02:00
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_bounce_limit(struct request_queue *q, u64 dma_addr)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
unsigned long bounce_pfn = dma_addr >> PAGE_SHIFT;
|
2006-03-09 02:57:26 +01:00
|
|
|
int dma = 0;
|
|
|
|
|
|
|
|
q->bounce_gfp = GFP_NOIO;
|
|
|
|
#if BITS_PER_LONG == 64
|
|
|
|
/* Assume anything <= 4GB can be handled by IOMMU.
|
|
|
|
Actually some IOMMUs can handle everything, but I don't
|
|
|
|
know of a way to test this here. */
|
2006-06-21 14:48:09 +02:00
|
|
|
if (bounce_pfn < (min_t(u64,0xffffffff,BLK_BOUNCE_HIGH) >> PAGE_SHIFT))
|
2006-03-09 02:57:26 +01:00
|
|
|
dma = 1;
|
|
|
|
q->bounce_pfn = max_low_pfn;
|
|
|
|
#else
|
|
|
|
if (bounce_pfn < blk_max_low_pfn)
|
|
|
|
dma = 1;
|
|
|
|
q->bounce_pfn = bounce_pfn;
|
|
|
|
#endif
|
|
|
|
if (dma) {
|
2005-04-17 00:20:36 +02:00
|
|
|
init_emergency_isa_pool();
|
|
|
|
q->bounce_gfp = GFP_NOIO | GFP_DMA;
|
2006-03-09 02:57:26 +01:00
|
|
|
q->bounce_pfn = bounce_pfn;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_bounce_limit);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_max_sectors - set max sectors for a request for this queue
|
|
|
|
* @q: the request queue for the device
|
|
|
|
* @max_sectors: max sectors in the usual 512b unit
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Enables a low level driver to set an upper limit on the size of
|
|
|
|
* received requests.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_max_sectors(struct request_queue *q, unsigned int max_sectors)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
if ((max_sectors << 9) < PAGE_CACHE_SIZE) {
|
|
|
|
max_sectors = 1 << (PAGE_CACHE_SHIFT - 9);
|
|
|
|
printk("%s: set to minimum %d\n", __FUNCTION__, max_sectors);
|
|
|
|
}
|
|
|
|
|
2005-12-05 09:37:06 +01:00
|
|
|
if (BLK_DEF_MAX_SECTORS > max_sectors)
|
|
|
|
q->max_hw_sectors = q->max_sectors = max_sectors;
|
|
|
|
else {
|
|
|
|
q->max_sectors = BLK_DEF_MAX_SECTORS;
|
|
|
|
q->max_hw_sectors = max_sectors;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_max_sectors);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_max_phys_segments - set max phys segments for a request for this queue
|
|
|
|
* @q: the request queue for the device
|
|
|
|
* @max_segments: max number of segments
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Enables a low level driver to set an upper limit on the number of
|
|
|
|
* physical data segments in a request. This would be the largest sized
|
|
|
|
* scatter list the driver could handle.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_max_phys_segments(struct request_queue *q,
|
|
|
|
unsigned short max_segments)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
if (!max_segments) {
|
|
|
|
max_segments = 1;
|
|
|
|
printk("%s: set to minimum %d\n", __FUNCTION__, max_segments);
|
|
|
|
}
|
|
|
|
|
|
|
|
q->max_phys_segments = max_segments;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_max_phys_segments);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_max_hw_segments - set max hw segments for a request for this queue
|
|
|
|
* @q: the request queue for the device
|
|
|
|
* @max_segments: max number of segments
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Enables a low level driver to set an upper limit on the number of
|
|
|
|
* hw data segments in a request. This would be the largest number of
|
|
|
|
* address/length pairs the host adapter can actually give as once
|
|
|
|
* to the device.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_max_hw_segments(struct request_queue *q,
|
|
|
|
unsigned short max_segments)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
if (!max_segments) {
|
|
|
|
max_segments = 1;
|
|
|
|
printk("%s: set to minimum %d\n", __FUNCTION__, max_segments);
|
|
|
|
}
|
|
|
|
|
|
|
|
q->max_hw_segments = max_segments;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_max_hw_segments);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_max_segment_size - set max segment size for blk_rq_map_sg
|
|
|
|
* @q: the request queue for the device
|
|
|
|
* @max_size: max size of segment in bytes
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Enables a low level driver to set an upper limit on the size of a
|
|
|
|
* coalesced segment
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_max_segment_size(struct request_queue *q, unsigned int max_size)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
if (max_size < PAGE_CACHE_SIZE) {
|
|
|
|
max_size = PAGE_CACHE_SIZE;
|
|
|
|
printk("%s: set to minimum %d\n", __FUNCTION__, max_size);
|
|
|
|
}
|
|
|
|
|
|
|
|
q->max_segment_size = max_size;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_max_segment_size);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_hardsect_size - set hardware sector size for the queue
|
|
|
|
* @q: the request queue for the device
|
|
|
|
* @size: the hardware sector size, in bytes
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* This should typically be set to the lowest possible sector size
|
|
|
|
* that the hardware can operate on (possible without reverting to
|
|
|
|
* even internal read-modify-write operations). Usually the default
|
|
|
|
* of 512 covers most hardware.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_hardsect_size(struct request_queue *q, unsigned short size)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
q->hardsect_size = size;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_hardsect_size);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Returns the minimum that is _not_ zero, unless both are zero.
|
|
|
|
*/
|
|
|
|
#define min_not_zero(l, r) (l == 0) ? r : ((r == 0) ? l : min(l, r))
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_stack_limits - inherit underlying queue limits for stacked drivers
|
|
|
|
* @t: the stacking driver (top)
|
|
|
|
* @b: the underlying device (bottom)
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
/* zero is "infinity" */
|
2005-12-05 09:37:06 +01:00
|
|
|
t->max_sectors = min_not_zero(t->max_sectors,b->max_sectors);
|
|
|
|
t->max_hw_sectors = min_not_zero(t->max_hw_sectors,b->max_hw_sectors);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
t->max_phys_segments = min(t->max_phys_segments,b->max_phys_segments);
|
|
|
|
t->max_hw_segments = min(t->max_hw_segments,b->max_hw_segments);
|
|
|
|
t->max_segment_size = min(t->max_segment_size,b->max_segment_size);
|
|
|
|
t->hardsect_size = max(t->hardsect_size,b->hardsect_size);
|
2006-03-27 11:18:02 +02:00
|
|
|
if (!test_bit(QUEUE_FLAG_CLUSTER, &b->queue_flags))
|
|
|
|
clear_bit(QUEUE_FLAG_CLUSTER, &t->queue_flags);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_stack_limits);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_segment_boundary - set boundary rules for segment merging
|
|
|
|
* @q: the request queue for the device
|
|
|
|
* @mask: the memory boundary mask
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_segment_boundary(struct request_queue *q, unsigned long mask)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
if (mask < PAGE_CACHE_SIZE - 1) {
|
|
|
|
mask = PAGE_CACHE_SIZE - 1;
|
|
|
|
printk("%s: set to minimum %lx\n", __FUNCTION__, mask);
|
|
|
|
}
|
|
|
|
|
|
|
|
q->seg_boundary_mask = mask;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_segment_boundary);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_dma_alignment - set dma length and memory alignment
|
|
|
|
* @q: the request queue for the device
|
|
|
|
* @mask: alignment mask
|
|
|
|
*
|
|
|
|
* description:
|
|
|
|
* set required memory and length aligment for direct dma transactions.
|
|
|
|
* this is used when buiding direct io requests for the queue.
|
|
|
|
*
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_dma_alignment(struct request_queue *q, int mask)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
q->dma_alignment = mask;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_dma_alignment);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_find_tag - find a request by its tag and queue
|
|
|
|
* @q: The request queue for the device
|
|
|
|
* @tag: The tag of the request
|
|
|
|
*
|
|
|
|
* Notes:
|
|
|
|
* Should be used when a device returns a tag and you want to match
|
|
|
|
* it with a request.
|
|
|
|
*
|
|
|
|
* no locks need be held.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request *blk_queue_find_tag(struct request_queue *q, int tag)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-10-04 08:27:25 +02:00
|
|
|
return blk_map_queue_find_tag(q->queue_tags, tag);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_find_tag);
|
|
|
|
|
|
|
|
/**
|
2006-08-30 21:48:45 +02:00
|
|
|
* __blk_free_tags - release a given set of tag maintenance info
|
|
|
|
* @bqt: the tag map to free
|
2005-04-17 00:20:36 +02:00
|
|
|
*
|
2006-08-30 21:48:45 +02:00
|
|
|
* Tries to free the specified @bqt@. Returns true if it was
|
|
|
|
* actually freed and false if there are still references using it
|
|
|
|
*/
|
|
|
|
static int __blk_free_tags(struct blk_queue_tag *bqt)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-08-30 21:48:45 +02:00
|
|
|
int retval;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-08-30 21:48:45 +02:00
|
|
|
retval = atomic_dec_and_test(&bqt->refcnt);
|
|
|
|
if (retval) {
|
2005-04-17 00:20:36 +02:00
|
|
|
BUG_ON(bqt->busy);
|
|
|
|
|
|
|
|
kfree(bqt->tag_index);
|
|
|
|
bqt->tag_index = NULL;
|
|
|
|
|
|
|
|
kfree(bqt->tag_map);
|
|
|
|
bqt->tag_map = NULL;
|
|
|
|
|
|
|
|
kfree(bqt);
|
2006-08-30 21:48:45 +02:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2006-08-30 21:48:45 +02:00
|
|
|
return retval;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* __blk_queue_free_tags - release tag maintenance info
|
|
|
|
* @q: the request queue for the device
|
|
|
|
*
|
|
|
|
* Notes:
|
|
|
|
* blk_cleanup_queue() will take care of calling this function, if tagging
|
|
|
|
* has been used. So there's no need to call this directly.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
static void __blk_queue_free_tags(struct request_queue *q)
|
2006-08-30 21:48:45 +02:00
|
|
|
{
|
|
|
|
struct blk_queue_tag *bqt = q->queue_tags;
|
|
|
|
|
|
|
|
if (!bqt)
|
|
|
|
return;
|
|
|
|
|
|
|
|
__blk_free_tags(bqt);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
q->queue_tags = NULL;
|
|
|
|
q->queue_flags &= ~(1 << QUEUE_FLAG_QUEUED);
|
|
|
|
}
|
|
|
|
|
2006-08-30 21:48:45 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_free_tags - release a given set of tag maintenance info
|
|
|
|
* @bqt: the tag map to free
|
|
|
|
*
|
|
|
|
* For externally managed @bqt@ frees the map. Callers of this
|
|
|
|
* function must guarantee to have released all the queues that
|
|
|
|
* might have been using this tag map.
|
|
|
|
*/
|
|
|
|
void blk_free_tags(struct blk_queue_tag *bqt)
|
|
|
|
{
|
|
|
|
if (unlikely(!__blk_free_tags(bqt)))
|
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_free_tags);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/**
|
|
|
|
* blk_queue_free_tags - release tag maintenance info
|
|
|
|
* @q: the request queue for the device
|
|
|
|
*
|
|
|
|
* Notes:
|
|
|
|
* This is used to disabled tagged queuing to a device, yet leave
|
|
|
|
* queue in function.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_free_tags(struct request_queue *q)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
clear_bit(QUEUE_FLAG_QUEUED, &q->queue_flags);
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_free_tags);
|
|
|
|
|
|
|
|
static int
|
2007-07-24 09:28:11 +02:00
|
|
|
init_tag_map(struct request_queue *q, struct blk_queue_tag *tags, int depth)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct request **tag_index;
|
|
|
|
unsigned long *tag_map;
|
2005-06-23 09:08:49 +02:00
|
|
|
int nr_ulongs;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-08-30 21:48:45 +02:00
|
|
|
if (q && depth > q->nr_requests * 2) {
|
2005-04-17 00:20:36 +02:00
|
|
|
depth = q->nr_requests * 2;
|
|
|
|
printk(KERN_ERR "%s: adjusted depth to %d\n",
|
|
|
|
__FUNCTION__, depth);
|
|
|
|
}
|
|
|
|
|
2006-03-08 13:31:44 +01:00
|
|
|
tag_index = kzalloc(depth * sizeof(struct request *), GFP_ATOMIC);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (!tag_index)
|
|
|
|
goto fail;
|
|
|
|
|
2005-06-23 09:08:50 +02:00
|
|
|
nr_ulongs = ALIGN(depth, BITS_PER_LONG) / BITS_PER_LONG;
|
2006-03-08 13:31:44 +01:00
|
|
|
tag_map = kzalloc(nr_ulongs * sizeof(unsigned long), GFP_ATOMIC);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (!tag_map)
|
|
|
|
goto fail;
|
|
|
|
|
2005-08-05 22:28:11 +02:00
|
|
|
tags->real_max_depth = depth;
|
2005-04-17 00:20:36 +02:00
|
|
|
tags->max_depth = depth;
|
|
|
|
tags->tag_index = tag_index;
|
|
|
|
tags->tag_map = tag_map;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
fail:
|
|
|
|
kfree(tag_index);
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
2006-08-30 21:48:45 +02:00
|
|
|
static struct blk_queue_tag *__blk_queue_init_tags(struct request_queue *q,
|
|
|
|
int depth)
|
|
|
|
{
|
|
|
|
struct blk_queue_tag *tags;
|
|
|
|
|
|
|
|
tags = kmalloc(sizeof(struct blk_queue_tag), GFP_ATOMIC);
|
|
|
|
if (!tags)
|
|
|
|
goto fail;
|
|
|
|
|
|
|
|
if (init_tag_map(q, tags, depth))
|
|
|
|
goto fail;
|
|
|
|
|
|
|
|
tags->busy = 0;
|
|
|
|
atomic_set(&tags->refcnt, 1);
|
|
|
|
return tags;
|
|
|
|
fail:
|
|
|
|
kfree(tags);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_init_tags - initialize the tag info for an external tag map
|
|
|
|
* @depth: the maximum queue depth supported
|
|
|
|
* @tags: the tag to use
|
|
|
|
**/
|
|
|
|
struct blk_queue_tag *blk_init_tags(int depth)
|
|
|
|
{
|
|
|
|
return __blk_queue_init_tags(NULL, depth);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_init_tags);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/**
|
|
|
|
* blk_queue_init_tags - initialize the queue tag info
|
|
|
|
* @q: the request queue for the device
|
|
|
|
* @depth: the maximum queue depth supported
|
|
|
|
* @tags: the tag to use
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
int blk_queue_init_tags(struct request_queue *q, int depth,
|
2005-04-17 00:20:36 +02:00
|
|
|
struct blk_queue_tag *tags)
|
|
|
|
{
|
|
|
|
int rc;
|
|
|
|
|
|
|
|
BUG_ON(tags && q->queue_tags && tags != q->queue_tags);
|
|
|
|
|
|
|
|
if (!tags && !q->queue_tags) {
|
2006-08-30 21:48:45 +02:00
|
|
|
tags = __blk_queue_init_tags(q, depth);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-08-30 21:48:45 +02:00
|
|
|
if (!tags)
|
2005-04-17 00:20:36 +02:00
|
|
|
goto fail;
|
|
|
|
} else if (q->queue_tags) {
|
|
|
|
if ((rc = blk_queue_resize_tags(q, depth)))
|
|
|
|
return rc;
|
|
|
|
set_bit(QUEUE_FLAG_QUEUED, &q->queue_flags);
|
|
|
|
return 0;
|
|
|
|
} else
|
|
|
|
atomic_inc(&tags->refcnt);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* assign it, all done
|
|
|
|
*/
|
|
|
|
q->queue_tags = tags;
|
|
|
|
q->queue_flags |= (1 << QUEUE_FLAG_QUEUED);
|
2007-10-25 10:14:47 +02:00
|
|
|
INIT_LIST_HEAD(&q->tag_busy_list);
|
2005-04-17 00:20:36 +02:00
|
|
|
return 0;
|
|
|
|
fail:
|
|
|
|
kfree(tags);
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_init_tags);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_resize_tags - change the queueing depth
|
|
|
|
* @q: the request queue for the device
|
|
|
|
* @new_depth: the new max command queueing depth
|
|
|
|
*
|
|
|
|
* Notes:
|
|
|
|
* Must be called with the queue lock held.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
int blk_queue_resize_tags(struct request_queue *q, int new_depth)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct blk_queue_tag *bqt = q->queue_tags;
|
|
|
|
struct request **tag_index;
|
|
|
|
unsigned long *tag_map;
|
2005-06-23 09:08:49 +02:00
|
|
|
int max_depth, nr_ulongs;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (!bqt)
|
|
|
|
return -ENXIO;
|
|
|
|
|
2005-08-05 22:28:11 +02:00
|
|
|
/*
|
|
|
|
* if we already have large enough real_max_depth. just
|
|
|
|
* adjust max_depth. *NOTE* as requests with tag value
|
|
|
|
* between new_depth and real_max_depth can be in-flight, tag
|
|
|
|
* map can not be shrunk blindly here.
|
|
|
|
*/
|
|
|
|
if (new_depth <= bqt->real_max_depth) {
|
|
|
|
bqt->max_depth = new_depth;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2006-08-30 21:48:45 +02:00
|
|
|
/*
|
|
|
|
* Currently cannot replace a shared tag map with a new
|
|
|
|
* one, so error out if this is the case
|
|
|
|
*/
|
|
|
|
if (atomic_read(&bqt->refcnt) != 1)
|
|
|
|
return -EBUSY;
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* save the old state info, so we can copy it back
|
|
|
|
*/
|
|
|
|
tag_index = bqt->tag_index;
|
|
|
|
tag_map = bqt->tag_map;
|
2005-08-05 22:28:11 +02:00
|
|
|
max_depth = bqt->real_max_depth;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (init_tag_map(q, bqt, new_depth))
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
memcpy(bqt->tag_index, tag_index, max_depth * sizeof(struct request *));
|
2005-06-23 09:08:50 +02:00
|
|
|
nr_ulongs = ALIGN(max_depth, BITS_PER_LONG) / BITS_PER_LONG;
|
2005-06-23 09:08:49 +02:00
|
|
|
memcpy(bqt->tag_map, tag_map, nr_ulongs * sizeof(unsigned long));
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
kfree(tag_index);
|
|
|
|
kfree(tag_map);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_resize_tags);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_end_tag - end tag operations for a request
|
|
|
|
* @q: the request queue for the device
|
|
|
|
* @rq: the request that has completed
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Typically called when end_that_request_first() returns 0, meaning
|
|
|
|
* all transfers have been done for a request. It's important to call
|
|
|
|
* this function before end_that_request_last(), as that will put the
|
|
|
|
* request back on the free list thus corrupting the internal tag list.
|
|
|
|
*
|
|
|
|
* Notes:
|
|
|
|
* queue lock must be held.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_end_tag(struct request_queue *q, struct request *rq)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct blk_queue_tag *bqt = q->queue_tags;
|
|
|
|
int tag = rq->tag;
|
|
|
|
|
|
|
|
BUG_ON(tag == -1);
|
|
|
|
|
2005-08-05 22:28:11 +02:00
|
|
|
if (unlikely(tag >= bqt->real_max_depth))
|
2005-06-23 09:08:51 +02:00
|
|
|
/*
|
|
|
|
* This can happen after tag depth has been reduced.
|
|
|
|
* FIXME: how about a warning or info message here?
|
|
|
|
*/
|
2005-04-17 00:20:36 +02:00
|
|
|
return;
|
|
|
|
|
|
|
|
list_del_init(&rq->queuelist);
|
2006-08-10 08:44:47 +02:00
|
|
|
rq->cmd_flags &= ~REQ_QUEUED;
|
2005-04-17 00:20:36 +02:00
|
|
|
rq->tag = -1;
|
|
|
|
|
|
|
|
if (unlikely(bqt->tag_index[tag] == NULL))
|
2005-06-23 09:08:51 +02:00
|
|
|
printk(KERN_ERR "%s: tag %d is missing\n",
|
|
|
|
__FUNCTION__, tag);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
bqt->tag_index[tag] = NULL;
|
2007-09-13 14:26:53 +02:00
|
|
|
|
2007-10-24 10:54:38 +02:00
|
|
|
if (unlikely(!test_bit(tag, bqt->tag_map))) {
|
2007-09-13 14:26:53 +02:00
|
|
|
printk(KERN_ERR "%s: attempt to clear non-busy tag (%d)\n",
|
|
|
|
__FUNCTION__, tag);
|
|
|
|
return;
|
|
|
|
}
|
2007-10-24 10:54:38 +02:00
|
|
|
/*
|
|
|
|
* The tag_map bit acts as a lock for tag_index[bit], so we need
|
|
|
|
* unlock memory barrier semantics.
|
|
|
|
*/
|
|
|
|
clear_bit_unlock(tag, bqt->tag_map);
|
2005-04-17 00:20:36 +02:00
|
|
|
bqt->busy--;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_end_tag);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_start_tag - find a free tag and assign it
|
|
|
|
* @q: the request queue for the device
|
|
|
|
* @rq: the block request that needs tagging
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* This can either be used as a stand-alone helper, or possibly be
|
|
|
|
* assigned as the queue &prep_rq_fn (in which case &struct request
|
|
|
|
* automagically gets a tag assigned). Note that this function
|
|
|
|
* assumes that any type of request can be queued! if this is not
|
|
|
|
* true for your device, you must check the request type before
|
|
|
|
* calling this function. The request will also be removed from
|
|
|
|
* the request queue, so it's the drivers responsibility to readd
|
|
|
|
* it if it should need to be restarted for some reason.
|
|
|
|
*
|
|
|
|
* Notes:
|
|
|
|
* queue lock must be held.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
int blk_queue_start_tag(struct request_queue *q, struct request *rq)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct blk_queue_tag *bqt = q->queue_tags;
|
2005-06-23 09:08:48 +02:00
|
|
|
int tag;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-08-10 08:44:47 +02:00
|
|
|
if (unlikely((rq->cmd_flags & REQ_QUEUED))) {
|
2005-04-17 00:20:36 +02:00
|
|
|
printk(KERN_ERR
|
2005-06-23 09:08:51 +02:00
|
|
|
"%s: request %p for device [%s] already tagged %d",
|
|
|
|
__FUNCTION__, rq,
|
|
|
|
rq->rq_disk ? rq->rq_disk->disk_name : "?", rq->tag);
|
2005-04-17 00:20:36 +02:00
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
|
2006-09-21 20:37:22 +02:00
|
|
|
/*
|
|
|
|
* Protect against shared tag maps, as we may not have exclusive
|
|
|
|
* access to the tag map.
|
|
|
|
*/
|
|
|
|
do {
|
|
|
|
tag = find_first_zero_bit(bqt->tag_map, bqt->max_depth);
|
|
|
|
if (tag >= bqt->max_depth)
|
|
|
|
return 1;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-10-24 10:54:38 +02:00
|
|
|
} while (test_and_set_bit_lock(tag, bqt->tag_map));
|
2007-09-14 08:41:12 +02:00
|
|
|
/*
|
2007-10-24 10:54:38 +02:00
|
|
|
* We need lock ordering semantics given by test_and_set_bit_lock.
|
|
|
|
* See blk_queue_end_tag for details.
|
2007-09-14 08:41:12 +02:00
|
|
|
*/
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-08-10 08:44:47 +02:00
|
|
|
rq->cmd_flags |= REQ_QUEUED;
|
2005-04-17 00:20:36 +02:00
|
|
|
rq->tag = tag;
|
|
|
|
bqt->tag_index[tag] = rq;
|
|
|
|
blkdev_dequeue_request(rq);
|
2007-10-25 10:14:47 +02:00
|
|
|
list_add(&rq->queuelist, &q->tag_busy_list);
|
2005-04-17 00:20:36 +02:00
|
|
|
bqt->busy++;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_start_tag);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_queue_invalidate_tags - invalidate all pending tags
|
|
|
|
* @q: the request queue for the device
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Hardware conditions may dictate a need to stop all pending requests.
|
|
|
|
* In this case, we will safely clear the block side of the tag queue and
|
|
|
|
* readd all requests to the request queue in the right order.
|
|
|
|
*
|
|
|
|
* Notes:
|
|
|
|
* queue lock must be held.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_queue_invalidate_tags(struct request_queue *q)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct list_head *tmp, *n;
|
|
|
|
|
2007-11-09 12:52:45 +01:00
|
|
|
list_for_each_safe(tmp, n, &q->tag_busy_list)
|
|
|
|
blk_requeue_request(q, list_entry_rq(tmp));
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_queue_invalidate_tags);
|
|
|
|
|
|
|
|
void blk_dump_rq_flags(struct request *rq, char *msg)
|
|
|
|
{
|
|
|
|
int bit;
|
|
|
|
|
2006-08-10 08:44:47 +02:00
|
|
|
printk("%s: dev %s: type=%x, flags=%x\n", msg,
|
|
|
|
rq->rq_disk ? rq->rq_disk->disk_name : "?", rq->cmd_type,
|
|
|
|
rq->cmd_flags);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
printk("\nsector %llu, nr/cnr %lu/%u\n", (unsigned long long)rq->sector,
|
|
|
|
rq->nr_sectors,
|
|
|
|
rq->current_nr_sectors);
|
|
|
|
printk("bio %p, biotail %p, buffer %p, data %p, len %u\n", rq->bio, rq->biotail, rq->buffer, rq->data, rq->data_len);
|
|
|
|
|
2006-08-10 08:44:47 +02:00
|
|
|
if (blk_pc_request(rq)) {
|
2005-04-17 00:20:36 +02:00
|
|
|
printk("cdb: ");
|
|
|
|
for (bit = 0; bit < sizeof(rq->cmd); bit++)
|
|
|
|
printk("%02x ", rq->cmd[bit]);
|
|
|
|
printk("\n");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_dump_rq_flags);
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_recount_segments(struct request_queue *q, struct bio *bio)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2007-08-16 13:27:52 +02:00
|
|
|
struct request rq;
|
|
|
|
struct bio *nxt = bio->bi_next;
|
|
|
|
rq.q = q;
|
|
|
|
rq.bio = rq.biotail = bio;
|
|
|
|
bio->bi_next = NULL;
|
|
|
|
blk_recalc_rq_segments(&rq);
|
|
|
|
bio->bi_next = nxt;
|
|
|
|
bio->bi_phys_segments = rq.nr_phys_segments;
|
|
|
|
bio->bi_hw_segments = rq.nr_hw_segments;
|
|
|
|
bio->bi_flags |= (1 << BIO_SEG_VALID);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_recount_segments);
|
|
|
|
|
|
|
|
static void blk_recalc_rq_segments(struct request *rq)
|
|
|
|
{
|
|
|
|
int nr_phys_segs;
|
|
|
|
int nr_hw_segs;
|
|
|
|
unsigned int phys_size;
|
|
|
|
unsigned int hw_size;
|
2005-04-17 00:20:36 +02:00
|
|
|
struct bio_vec *bv, *bvprv = NULL;
|
2007-08-16 13:27:52 +02:00
|
|
|
int seg_size;
|
|
|
|
int hw_seg_size;
|
|
|
|
int cluster;
|
2007-09-25 12:35:59 +02:00
|
|
|
struct req_iterator iter;
|
2005-04-17 00:20:36 +02:00
|
|
|
int high, highprv = 1;
|
2007-08-16 13:27:52 +02:00
|
|
|
struct request_queue *q = rq->q;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-08-16 13:27:52 +02:00
|
|
|
if (!rq->bio)
|
2005-04-17 00:20:36 +02:00
|
|
|
return;
|
|
|
|
|
|
|
|
cluster = q->queue_flags & (1 << QUEUE_FLAG_CLUSTER);
|
2007-08-16 13:27:52 +02:00
|
|
|
hw_seg_size = seg_size = 0;
|
|
|
|
phys_size = hw_size = nr_phys_segs = nr_hw_segs = 0;
|
2007-09-25 12:35:59 +02:00
|
|
|
rq_for_each_segment(bv, rq, iter) {
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* the trick here is making sure that a high page is never
|
|
|
|
* considered part of another segment, since that might
|
|
|
|
* change with the bounce page.
|
|
|
|
*/
|
block: blk_max_pfn is somtimes wrong
There is a small problem in handling page bounce.
At the moment blk_max_pfn equals max_pfn, which is in fact not maximum
possible _number_ of a page frame, but the _amount_ of page frames. For
example for the 32bit x86 node with 4Gb RAM, max_pfn = 0x100000, but not
0xFFFF.
request_queue structure has a member q->bounce_pfn and queue needs bounce
pages for the pages _above_ this limit. This routine is handled by
blk_queue_bounce(), where the following check is produced:
if (q->bounce_pfn >= blk_max_pfn)
return;
Assume, that a driver has set q->bounce_pfn to 0xFFFF, but blk_max_pfn
equals 0x10000. In such situation the check above fails and for each bio
we always fall down for iterating over pages tied to the bio.
I want to notice, that for quite a big range of device drivers (ide, md,
...) such problem doesn't happen because they use BLK_BOUNCE_ANY for
bounce_pfn. BLK_BOUNCE_ANY is defined as blk_max_pfn << PAGE_SHIFT, and
then the check above doesn't fail. But for other drivers, which obtain
reuired value from drivers, it fails. For example sata_nv uses
ATA_DMA_MASK or dev->dma_mask.
I propose to use (max_pfn - 1) for blk_max_pfn. And the same for
blk_max_low_pfn. The patch also cleanses some checks related with
bounce_pfn.
Signed-off-by: Vasily Tarasov <vtaras@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-03-27 08:52:47 +02:00
|
|
|
high = page_to_pfn(bv->bv_page) > q->bounce_pfn;
|
2005-04-17 00:20:36 +02:00
|
|
|
if (high || highprv)
|
|
|
|
goto new_hw_segment;
|
|
|
|
if (cluster) {
|
|
|
|
if (seg_size + bv->bv_len > q->max_segment_size)
|
|
|
|
goto new_segment;
|
|
|
|
if (!BIOVEC_PHYS_MERGEABLE(bvprv, bv))
|
|
|
|
goto new_segment;
|
|
|
|
if (!BIOVEC_SEG_BOUNDARY(q, bvprv, bv))
|
|
|
|
goto new_segment;
|
|
|
|
if (BIOVEC_VIRT_OVERSIZE(hw_seg_size + bv->bv_len))
|
|
|
|
goto new_hw_segment;
|
|
|
|
|
|
|
|
seg_size += bv->bv_len;
|
|
|
|
hw_seg_size += bv->bv_len;
|
|
|
|
bvprv = bv;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
new_segment:
|
|
|
|
if (BIOVEC_VIRT_MERGEABLE(bvprv, bv) &&
|
2007-08-16 13:27:52 +02:00
|
|
|
!BIOVEC_VIRT_OVERSIZE(hw_seg_size + bv->bv_len))
|
2005-04-17 00:20:36 +02:00
|
|
|
hw_seg_size += bv->bv_len;
|
2007-08-16 13:27:52 +02:00
|
|
|
else {
|
2005-04-17 00:20:36 +02:00
|
|
|
new_hw_segment:
|
2007-08-16 13:27:52 +02:00
|
|
|
if (nr_hw_segs == 1 &&
|
|
|
|
hw_seg_size > rq->bio->bi_hw_front_size)
|
|
|
|
rq->bio->bi_hw_front_size = hw_seg_size;
|
2005-04-17 00:20:36 +02:00
|
|
|
hw_seg_size = BIOVEC_VIRT_START_SIZE(bv) + bv->bv_len;
|
|
|
|
nr_hw_segs++;
|
|
|
|
}
|
|
|
|
|
|
|
|
nr_phys_segs++;
|
|
|
|
bvprv = bv;
|
|
|
|
seg_size = bv->bv_len;
|
|
|
|
highprv = high;
|
|
|
|
}
|
2007-08-16 13:27:52 +02:00
|
|
|
|
|
|
|
if (nr_hw_segs == 1 &&
|
|
|
|
hw_seg_size > rq->bio->bi_hw_front_size)
|
|
|
|
rq->bio->bi_hw_front_size = hw_seg_size;
|
|
|
|
if (hw_seg_size > rq->biotail->bi_hw_back_size)
|
|
|
|
rq->biotail->bi_hw_back_size = hw_seg_size;
|
|
|
|
rq->nr_phys_segments = nr_phys_segs;
|
|
|
|
rq->nr_hw_segments = nr_hw_segs;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static int blk_phys_contig_segment(struct request_queue *q, struct bio *bio,
|
2005-04-17 00:20:36 +02:00
|
|
|
struct bio *nxt)
|
|
|
|
{
|
|
|
|
if (!(q->queue_flags & (1 << QUEUE_FLAG_CLUSTER)))
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
if (!BIOVEC_PHYS_MERGEABLE(__BVEC_END(bio), __BVEC_START(nxt)))
|
|
|
|
return 0;
|
|
|
|
if (bio->bi_size + nxt->bi_size > q->max_segment_size)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* bio and nxt are contigous in memory, check if the queue allows
|
|
|
|
* these two to be merged into one
|
|
|
|
*/
|
|
|
|
if (BIO_SEG_BOUNDARY(q, bio, nxt))
|
|
|
|
return 1;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static int blk_hw_contig_segment(struct request_queue *q, struct bio *bio,
|
2005-04-17 00:20:36 +02:00
|
|
|
struct bio *nxt)
|
|
|
|
{
|
|
|
|
if (unlikely(!bio_flagged(bio, BIO_SEG_VALID)))
|
|
|
|
blk_recount_segments(q, bio);
|
|
|
|
if (unlikely(!bio_flagged(nxt, BIO_SEG_VALID)))
|
|
|
|
blk_recount_segments(q, nxt);
|
|
|
|
if (!BIOVEC_VIRT_MERGEABLE(__BVEC_END(bio), __BVEC_START(nxt)) ||
|
2007-06-19 09:09:27 +02:00
|
|
|
BIOVEC_VIRT_OVERSIZE(bio->bi_hw_back_size + nxt->bi_hw_front_size))
|
2005-04-17 00:20:36 +02:00
|
|
|
return 0;
|
2007-06-19 09:09:27 +02:00
|
|
|
if (bio->bi_hw_back_size + nxt->bi_hw_front_size > q->max_segment_size)
|
2005-04-17 00:20:36 +02:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* map a request to scatterlist, return number of sg entries setup. Caller
|
|
|
|
* must make sure sg can hold rq->nr_phys_segments entries
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
int blk_rq_map_sg(struct request_queue *q, struct request *rq,
|
2007-09-21 10:44:19 +02:00
|
|
|
struct scatterlist *sglist)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct bio_vec *bvec, *bvprv;
|
2007-09-25 12:35:59 +02:00
|
|
|
struct req_iterator iter;
|
2007-10-17 19:34:11 +02:00
|
|
|
struct scatterlist *sg;
|
2007-09-25 12:35:59 +02:00
|
|
|
int nsegs, cluster;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
nsegs = 0;
|
|
|
|
cluster = q->queue_flags & (1 << QUEUE_FLAG_CLUSTER);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* for each bio in rq
|
|
|
|
*/
|
|
|
|
bvprv = NULL;
|
2007-10-17 19:34:11 +02:00
|
|
|
sg = NULL;
|
2007-09-25 12:35:59 +02:00
|
|
|
rq_for_each_segment(bvec, rq, iter) {
|
2007-08-16 13:43:12 +02:00
|
|
|
int nbytes = bvec->bv_len;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-08-16 13:43:12 +02:00
|
|
|
if (bvprv && cluster) {
|
2007-09-21 10:44:19 +02:00
|
|
|
if (sg->length + nbytes > q->max_segment_size)
|
2007-08-16 13:43:12 +02:00
|
|
|
goto new_segment;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-08-16 13:43:12 +02:00
|
|
|
if (!BIOVEC_PHYS_MERGEABLE(bvprv, bvec))
|
|
|
|
goto new_segment;
|
|
|
|
if (!BIOVEC_SEG_BOUNDARY(q, bvprv, bvec))
|
|
|
|
goto new_segment;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-09-21 10:44:19 +02:00
|
|
|
sg->length += nbytes;
|
2007-08-16 13:43:12 +02:00
|
|
|
} else {
|
2005-04-17 00:20:36 +02:00
|
|
|
new_segment:
|
2007-10-17 19:34:11 +02:00
|
|
|
if (!sg)
|
|
|
|
sg = sglist;
|
2007-10-23 09:49:25 +02:00
|
|
|
else {
|
|
|
|
/*
|
|
|
|
* If the driver previously mapped a shorter
|
|
|
|
* list, we could see a termination bit
|
|
|
|
* prematurely unless it fully inits the sg
|
|
|
|
* table on each mapping. We KNOW that there
|
|
|
|
* must be more entries here or the driver
|
|
|
|
* would be buggy, so force clear the
|
|
|
|
* termination bit to avoid doing a full
|
|
|
|
* sg_init_table() in drivers for each command.
|
|
|
|
*/
|
|
|
|
sg->page_link &= ~0x02;
|
2007-10-17 19:34:11 +02:00
|
|
|
sg = sg_next(sg);
|
2007-10-23 09:49:25 +02:00
|
|
|
}
|
2007-08-16 13:43:12 +02:00
|
|
|
|
2007-10-24 11:20:47 +02:00
|
|
|
sg_set_page(sg, bvec->bv_page, nbytes, bvec->bv_offset);
|
2007-08-16 13:43:12 +02:00
|
|
|
nsegs++;
|
|
|
|
}
|
|
|
|
bvprv = bvec;
|
2007-09-25 12:35:59 +02:00
|
|
|
} /* segments in rq */
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-10-22 19:39:33 +02:00
|
|
|
if (sg)
|
2007-10-31 12:06:37 +01:00
|
|
|
sg_mark_end(sg);
|
2007-10-22 19:39:33 +02:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
return nsegs;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_rq_map_sg);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* the standard queue merge functions, can be overridden with device
|
|
|
|
* specific ones if so desired
|
|
|
|
*/
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static inline int ll_new_mergeable(struct request_queue *q,
|
2005-04-17 00:20:36 +02:00
|
|
|
struct request *req,
|
|
|
|
struct bio *bio)
|
|
|
|
{
|
|
|
|
int nr_phys_segs = bio_phys_segments(q, bio);
|
|
|
|
|
|
|
|
if (req->nr_phys_segments + nr_phys_segs > q->max_phys_segments) {
|
2006-08-10 08:44:47 +02:00
|
|
|
req->cmd_flags |= REQ_NOMERGE;
|
2005-04-17 00:20:36 +02:00
|
|
|
if (req == q->last_merge)
|
|
|
|
q->last_merge = NULL;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* A hw segment is just getting larger, bump just the phys
|
|
|
|
* counter.
|
|
|
|
*/
|
|
|
|
req->nr_phys_segments += nr_phys_segs;
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static inline int ll_new_hw_segment(struct request_queue *q,
|
2005-04-17 00:20:36 +02:00
|
|
|
struct request *req,
|
|
|
|
struct bio *bio)
|
|
|
|
{
|
|
|
|
int nr_hw_segs = bio_hw_segments(q, bio);
|
|
|
|
int nr_phys_segs = bio_phys_segments(q, bio);
|
|
|
|
|
|
|
|
if (req->nr_hw_segments + nr_hw_segs > q->max_hw_segments
|
|
|
|
|| req->nr_phys_segments + nr_phys_segs > q->max_phys_segments) {
|
2006-08-10 08:44:47 +02:00
|
|
|
req->cmd_flags |= REQ_NOMERGE;
|
2005-04-17 00:20:36 +02:00
|
|
|
if (req == q->last_merge)
|
|
|
|
q->last_merge = NULL;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This will form the start of a new hw segment. Bump both
|
|
|
|
* counters.
|
|
|
|
*/
|
|
|
|
req->nr_hw_segments += nr_hw_segs;
|
|
|
|
req->nr_phys_segments += nr_phys_segs;
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2007-08-16 13:31:27 +02:00
|
|
|
static int ll_back_merge_fn(struct request_queue *q, struct request *req,
|
|
|
|
struct bio *bio)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2005-12-05 09:37:06 +01:00
|
|
|
unsigned short max_sectors;
|
2005-04-17 00:20:36 +02:00
|
|
|
int len;
|
|
|
|
|
2005-12-05 09:37:06 +01:00
|
|
|
if (unlikely(blk_pc_request(req)))
|
|
|
|
max_sectors = q->max_hw_sectors;
|
|
|
|
else
|
|
|
|
max_sectors = q->max_sectors;
|
|
|
|
|
|
|
|
if (req->nr_sectors + bio_sectors(bio) > max_sectors) {
|
2006-08-10 08:44:47 +02:00
|
|
|
req->cmd_flags |= REQ_NOMERGE;
|
2005-04-17 00:20:36 +02:00
|
|
|
if (req == q->last_merge)
|
|
|
|
q->last_merge = NULL;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if (unlikely(!bio_flagged(req->biotail, BIO_SEG_VALID)))
|
|
|
|
blk_recount_segments(q, req->biotail);
|
|
|
|
if (unlikely(!bio_flagged(bio, BIO_SEG_VALID)))
|
|
|
|
blk_recount_segments(q, bio);
|
|
|
|
len = req->biotail->bi_hw_back_size + bio->bi_hw_front_size;
|
|
|
|
if (BIOVEC_VIRT_MERGEABLE(__BVEC_END(req->biotail), __BVEC_START(bio)) &&
|
|
|
|
!BIOVEC_VIRT_OVERSIZE(len)) {
|
|
|
|
int mergeable = ll_new_mergeable(q, req, bio);
|
|
|
|
|
|
|
|
if (mergeable) {
|
|
|
|
if (req->nr_hw_segments == 1)
|
|
|
|
req->bio->bi_hw_front_size = len;
|
|
|
|
if (bio->bi_hw_segments == 1)
|
|
|
|
bio->bi_hw_back_size = len;
|
|
|
|
}
|
|
|
|
return mergeable;
|
|
|
|
}
|
|
|
|
|
|
|
|
return ll_new_hw_segment(q, req, bio);
|
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static int ll_front_merge_fn(struct request_queue *q, struct request *req,
|
2005-04-17 00:20:36 +02:00
|
|
|
struct bio *bio)
|
|
|
|
{
|
2005-12-05 09:37:06 +01:00
|
|
|
unsigned short max_sectors;
|
2005-04-17 00:20:36 +02:00
|
|
|
int len;
|
|
|
|
|
2005-12-05 09:37:06 +01:00
|
|
|
if (unlikely(blk_pc_request(req)))
|
|
|
|
max_sectors = q->max_hw_sectors;
|
|
|
|
else
|
|
|
|
max_sectors = q->max_sectors;
|
|
|
|
|
|
|
|
|
|
|
|
if (req->nr_sectors + bio_sectors(bio) > max_sectors) {
|
2006-08-10 08:44:47 +02:00
|
|
|
req->cmd_flags |= REQ_NOMERGE;
|
2005-04-17 00:20:36 +02:00
|
|
|
if (req == q->last_merge)
|
|
|
|
q->last_merge = NULL;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
len = bio->bi_hw_back_size + req->bio->bi_hw_front_size;
|
|
|
|
if (unlikely(!bio_flagged(bio, BIO_SEG_VALID)))
|
|
|
|
blk_recount_segments(q, bio);
|
|
|
|
if (unlikely(!bio_flagged(req->bio, BIO_SEG_VALID)))
|
|
|
|
blk_recount_segments(q, req->bio);
|
|
|
|
if (BIOVEC_VIRT_MERGEABLE(__BVEC_END(bio), __BVEC_START(req->bio)) &&
|
|
|
|
!BIOVEC_VIRT_OVERSIZE(len)) {
|
|
|
|
int mergeable = ll_new_mergeable(q, req, bio);
|
|
|
|
|
|
|
|
if (mergeable) {
|
|
|
|
if (bio->bi_hw_segments == 1)
|
|
|
|
bio->bi_hw_front_size = len;
|
|
|
|
if (req->nr_hw_segments == 1)
|
|
|
|
req->biotail->bi_hw_back_size = len;
|
|
|
|
}
|
|
|
|
return mergeable;
|
|
|
|
}
|
|
|
|
|
|
|
|
return ll_new_hw_segment(q, req, bio);
|
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static int ll_merge_requests_fn(struct request_queue *q, struct request *req,
|
2005-04-17 00:20:36 +02:00
|
|
|
struct request *next)
|
|
|
|
{
|
2005-06-25 23:59:20 +02:00
|
|
|
int total_phys_segments;
|
|
|
|
int total_hw_segments;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* First check if the either of the requests are re-queued
|
|
|
|
* requests. Can't merge them if they are.
|
|
|
|
*/
|
|
|
|
if (req->special || next->special)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/*
|
2005-06-25 23:59:20 +02:00
|
|
|
* Will it become too large?
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
|
|
|
if ((req->nr_sectors + next->nr_sectors) > q->max_sectors)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
total_phys_segments = req->nr_phys_segments + next->nr_phys_segments;
|
|
|
|
if (blk_phys_contig_segment(q, req->biotail, next->bio))
|
|
|
|
total_phys_segments--;
|
|
|
|
|
|
|
|
if (total_phys_segments > q->max_phys_segments)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
total_hw_segments = req->nr_hw_segments + next->nr_hw_segments;
|
|
|
|
if (blk_hw_contig_segment(q, req->biotail, next->bio)) {
|
|
|
|
int len = req->biotail->bi_hw_back_size + next->bio->bi_hw_front_size;
|
|
|
|
/*
|
|
|
|
* propagate the combined length to the end of the requests
|
|
|
|
*/
|
|
|
|
if (req->nr_hw_segments == 1)
|
|
|
|
req->bio->bi_hw_front_size = len;
|
|
|
|
if (next->nr_hw_segments == 1)
|
|
|
|
next->biotail->bi_hw_back_size = len;
|
|
|
|
total_hw_segments--;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (total_hw_segments > q->max_hw_segments)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/* Merge is OK... */
|
|
|
|
req->nr_phys_segments = total_phys_segments;
|
|
|
|
req->nr_hw_segments = total_hw_segments;
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* "plug" the device if there are no outstanding requests: this will
|
|
|
|
* force the transfer to start only after we have put all the requests
|
|
|
|
* on the list.
|
|
|
|
*
|
|
|
|
* This is called with interrupts off and no requests on the queue and
|
|
|
|
* with the queue lock held.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_plug_device(struct request_queue *q)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
WARN_ON(!irqs_disabled());
|
|
|
|
|
|
|
|
/*
|
|
|
|
* don't plug a stopped queue, it must be paired with blk_start_queue()
|
|
|
|
* which will restart the queueing
|
|
|
|
*/
|
2006-04-19 10:14:49 +02:00
|
|
|
if (blk_queue_stopped(q))
|
2005-04-17 00:20:36 +02:00
|
|
|
return;
|
|
|
|
|
2006-03-23 20:00:26 +01:00
|
|
|
if (!test_and_set_bit(QUEUE_FLAG_PLUGGED, &q->queue_flags)) {
|
2005-04-17 00:20:36 +02:00
|
|
|
mod_timer(&q->unplug_timer, jiffies + q->unplug_delay);
|
2006-03-23 20:00:26 +01:00
|
|
|
blk_add_trace_generic(q, NULL, 0, BLK_TA_PLUG);
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_plug_device);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* remove the queue from the plugged list, if present. called with
|
|
|
|
* queue lock held and interrupts disabled.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
int blk_remove_plug(struct request_queue *q)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
WARN_ON(!irqs_disabled());
|
|
|
|
|
|
|
|
if (!test_and_clear_bit(QUEUE_FLAG_PLUGGED, &q->queue_flags))
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
del_timer(&q->unplug_timer);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_remove_plug);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* remove the plug and let it rip..
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
void __generic_unplug_device(struct request_queue *q)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-04-19 10:14:49 +02:00
|
|
|
if (unlikely(blk_queue_stopped(q)))
|
2005-04-17 00:20:36 +02:00
|
|
|
return;
|
|
|
|
|
|
|
|
if (!blk_remove_plug(q))
|
|
|
|
return;
|
|
|
|
|
2005-06-27 10:55:12 +02:00
|
|
|
q->request_fn(q);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(__generic_unplug_device);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* generic_unplug_device - fire a request queue
|
2007-07-24 09:28:11 +02:00
|
|
|
* @q: The &struct request_queue in question
|
2005-04-17 00:20:36 +02:00
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Linux uses plugging to build bigger requests queues before letting
|
|
|
|
* the device have at them. If a queue is plugged, the I/O scheduler
|
|
|
|
* is still adding and merging requests on the queue. Once the queue
|
|
|
|
* gets unplugged, the request_fn defined for the queue is invoked and
|
|
|
|
* transfers started.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void generic_unplug_device(struct request_queue *q)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
spin_lock_irq(q->queue_lock);
|
|
|
|
__generic_unplug_device(q);
|
|
|
|
spin_unlock_irq(q->queue_lock);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(generic_unplug_device);
|
|
|
|
|
|
|
|
static void blk_backing_dev_unplug(struct backing_dev_info *bdi,
|
|
|
|
struct page *page)
|
|
|
|
{
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q = bdi->unplug_io_data;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-11-07 20:26:56 +01:00
|
|
|
blk_unplug(q);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2006-11-22 15:55:48 +01:00
|
|
|
static void blk_unplug_work(struct work_struct *work)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q =
|
|
|
|
container_of(work, struct request_queue, unplug_work);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-03-23 20:00:26 +01:00
|
|
|
blk_add_trace_pdu_int(q, BLK_TA_UNPLUG_IO, NULL,
|
|
|
|
q->rq.count[READ] + q->rq.count[WRITE]);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
q->unplug_fn(q);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void blk_unplug_timeout(unsigned long data)
|
|
|
|
{
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q = (struct request_queue *)data;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-03-23 20:00:26 +01:00
|
|
|
blk_add_trace_pdu_int(q, BLK_TA_UNPLUG_TIMER, NULL,
|
|
|
|
q->rq.count[READ] + q->rq.count[WRITE]);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
kblockd_schedule_work(&q->unplug_work);
|
|
|
|
}
|
|
|
|
|
2007-11-07 20:26:56 +01:00
|
|
|
void blk_unplug(struct request_queue *q)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* devices don't necessarily have an ->unplug_fn defined
|
|
|
|
*/
|
|
|
|
if (q->unplug_fn) {
|
|
|
|
blk_add_trace_pdu_int(q, BLK_TA_UNPLUG_IO, NULL,
|
|
|
|
q->rq.count[READ] + q->rq.count[WRITE]);
|
|
|
|
|
|
|
|
q->unplug_fn(q);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_unplug);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/**
|
|
|
|
* blk_start_queue - restart a previously stopped queue
|
2007-07-24 09:28:11 +02:00
|
|
|
* @q: The &struct request_queue in question
|
2005-04-17 00:20:36 +02:00
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* blk_start_queue() will clear the stop flag on the queue, and call
|
|
|
|
* the request_fn for the queue if it was in a stopped state when
|
|
|
|
* entered. Also see blk_stop_queue(). Queue lock must be held.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_start_queue(struct request_queue *q)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-06-05 12:09:01 +02:00
|
|
|
WARN_ON(!irqs_disabled());
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
clear_bit(QUEUE_FLAG_STOPPED, &q->queue_flags);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* one level of recursion is ok and is much faster than kicking
|
|
|
|
* the unplug handling
|
|
|
|
*/
|
|
|
|
if (!test_and_set_bit(QUEUE_FLAG_REENTER, &q->queue_flags)) {
|
|
|
|
q->request_fn(q);
|
|
|
|
clear_bit(QUEUE_FLAG_REENTER, &q->queue_flags);
|
|
|
|
} else {
|
|
|
|
blk_plug_device(q);
|
|
|
|
kblockd_schedule_work(&q->unplug_work);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_start_queue);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_stop_queue - stop a queue
|
2007-07-24 09:28:11 +02:00
|
|
|
* @q: The &struct request_queue in question
|
2005-04-17 00:20:36 +02:00
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* The Linux block layer assumes that a block driver will consume all
|
|
|
|
* entries on the request queue when the request_fn strategy is called.
|
|
|
|
* Often this will not happen, because of hardware limitations (queue
|
|
|
|
* depth settings). If a device driver gets a 'queue full' response,
|
|
|
|
* or if it simply chooses not to queue more I/O at one point, it can
|
|
|
|
* call this function to prevent the request_fn from being called until
|
|
|
|
* the driver has signalled it's ready to go again. This happens by calling
|
|
|
|
* blk_start_queue() to restart queue operations. Queue lock must be held.
|
|
|
|
**/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_stop_queue(struct request_queue *q)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
blk_remove_plug(q);
|
|
|
|
set_bit(QUEUE_FLAG_STOPPED, &q->queue_flags);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_stop_queue);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_sync_queue - cancel any pending callbacks on a queue
|
|
|
|
* @q: the queue
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* The block layer may perform asynchronous callback activity
|
|
|
|
* on a queue, such as calling the unplug function after a timeout.
|
|
|
|
* A block device may call blk_sync_queue to ensure that any
|
|
|
|
* such activity is cancelled, thus allowing it to release resources
|
2007-05-09 08:57:56 +02:00
|
|
|
* that the callbacks might use. The caller must already have made sure
|
2005-04-17 00:20:36 +02:00
|
|
|
* that its ->make_request_fn will not re-add plugging prior to calling
|
|
|
|
* this function.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
void blk_sync_queue(struct request_queue *q)
|
|
|
|
{
|
|
|
|
del_timer_sync(&q->unplug_timer);
|
2007-10-23 15:08:19 +02:00
|
|
|
kblockd_flush_work(&q->unplug_work);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_sync_queue);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_run_queue - run a single device queue
|
|
|
|
* @q: The queue to run
|
|
|
|
*/
|
|
|
|
void blk_run_queue(struct request_queue *q)
|
|
|
|
{
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
spin_lock_irqsave(q->queue_lock, flags);
|
|
|
|
blk_remove_plug(q);
|
2006-05-11 08:20:16 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Only recurse once to avoid overrunning the stack, let the unplug
|
|
|
|
* handling reinvoke the handler shortly if we already got there.
|
|
|
|
*/
|
|
|
|
if (!elv_queue_empty(q)) {
|
|
|
|
if (!test_and_set_bit(QUEUE_FLAG_REENTER, &q->queue_flags)) {
|
|
|
|
q->request_fn(q);
|
|
|
|
clear_bit(QUEUE_FLAG_REENTER, &q->queue_flags);
|
|
|
|
} else {
|
|
|
|
blk_plug_device(q);
|
|
|
|
kblockd_schedule_work(&q->unplug_work);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
spin_unlock_irqrestore(q->queue_lock, flags);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_run_queue);
|
|
|
|
|
|
|
|
/**
|
2007-07-24 09:28:11 +02:00
|
|
|
* blk_cleanup_queue: - release a &struct request_queue when it is no longer needed
|
2006-04-02 13:59:55 +02:00
|
|
|
* @kobj: the kobj belonging of the request queue to be released
|
2005-04-17 00:20:36 +02:00
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* blk_cleanup_queue is the pair to blk_init_queue() or
|
|
|
|
* blk_queue_make_request(). It should be called when a request queue is
|
|
|
|
* being released; typically when a block device is being de-registered.
|
|
|
|
* Currently, its primary task it to free all the &struct request
|
|
|
|
* structures that were allocated to the queue and the queue itself.
|
|
|
|
*
|
|
|
|
* Caveat:
|
|
|
|
* Hopefully the low level driver will have finished any
|
|
|
|
* outstanding requests first...
|
|
|
|
**/
|
2006-03-19 00:34:37 +01:00
|
|
|
static void blk_release_queue(struct kobject *kobj)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q =
|
|
|
|
container_of(kobj, struct request_queue, kobj);
|
2005-04-17 00:20:36 +02:00
|
|
|
struct request_list *rl = &q->rq;
|
|
|
|
|
|
|
|
blk_sync_queue(q);
|
|
|
|
|
|
|
|
if (rl->rq_pool)
|
|
|
|
mempool_destroy(rl->rq_pool);
|
|
|
|
|
|
|
|
if (q->queue_tags)
|
|
|
|
__blk_queue_free_tags(q);
|
|
|
|
|
2006-09-29 10:59:40 +02:00
|
|
|
blk_trace_shutdown(q);
|
2006-03-23 20:00:26 +01:00
|
|
|
|
2007-10-17 08:25:46 +02:00
|
|
|
bdi_destroy(&q->backing_dev_info);
|
2005-04-17 00:20:36 +02:00
|
|
|
kmem_cache_free(requestq_cachep, q);
|
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_put_queue(struct request_queue *q)
|
2006-03-19 00:34:37 +01:00
|
|
|
{
|
|
|
|
kobject_put(&q->kobj);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_put_queue);
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_cleanup_queue(struct request_queue * q)
|
2006-03-19 00:34:37 +01:00
|
|
|
{
|
|
|
|
mutex_lock(&q->sysfs_lock);
|
|
|
|
set_bit(QUEUE_FLAG_DEAD, &q->queue_flags);
|
|
|
|
mutex_unlock(&q->sysfs_lock);
|
|
|
|
|
|
|
|
if (q->elevator)
|
|
|
|
elevator_exit(q->elevator);
|
|
|
|
|
|
|
|
blk_put_queue(q);
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
EXPORT_SYMBOL(blk_cleanup_queue);
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static int blk_init_free_list(struct request_queue *q)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct request_list *rl = &q->rq;
|
|
|
|
|
|
|
|
rl->count[READ] = rl->count[WRITE] = 0;
|
|
|
|
rl->starved[READ] = rl->starved[WRITE] = 0;
|
2005-10-28 08:29:39 +02:00
|
|
|
rl->elvpriv = 0;
|
2005-04-17 00:20:36 +02:00
|
|
|
init_waitqueue_head(&rl->wait[READ]);
|
|
|
|
init_waitqueue_head(&rl->wait[WRITE]);
|
|
|
|
|
2005-06-23 09:08:19 +02:00
|
|
|
rl->rq_pool = mempool_create_node(BLKDEV_MIN_RQ, mempool_alloc_slab,
|
|
|
|
mempool_free_slab, request_cachep, q->node);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (!rl->rq_pool)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *blk_alloc_queue(gfp_t gfp_mask)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2005-06-23 09:08:19 +02:00
|
|
|
return blk_alloc_queue_node(gfp_mask, -1);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_alloc_queue);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-03-19 00:34:37 +01:00
|
|
|
static struct kobj_type queue_ktype;
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id)
|
2005-06-23 09:08:19 +02:00
|
|
|
{
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q;
|
2007-10-17 08:25:46 +02:00
|
|
|
int err;
|
2005-06-23 09:08:19 +02:00
|
|
|
|
2007-07-17 13:03:29 +02:00
|
|
|
q = kmem_cache_alloc_node(requestq_cachep,
|
|
|
|
gfp_mask | __GFP_ZERO, node_id);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (!q)
|
|
|
|
return NULL;
|
|
|
|
|
2007-10-17 08:25:46 +02:00
|
|
|
q->backing_dev_info.unplug_io_fn = blk_backing_dev_unplug;
|
|
|
|
q->backing_dev_info.unplug_io_data = q;
|
|
|
|
err = bdi_init(&q->backing_dev_info);
|
|
|
|
if (err) {
|
|
|
|
kmem_cache_free(requestq_cachep, q);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
init_timer(&q->unplug_timer);
|
2006-03-19 00:34:37 +01:00
|
|
|
|
2007-12-18 07:05:35 +01:00
|
|
|
kobject_init_ng(&q->kobj, &queue_ktype);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-03-19 00:34:37 +01:00
|
|
|
mutex_init(&q->sysfs_lock);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
return q;
|
|
|
|
}
|
2005-06-23 09:08:19 +02:00
|
|
|
EXPORT_SYMBOL(blk_alloc_queue_node);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_init_queue - prepare a request queue for use with a block device
|
|
|
|
* @rfn: The function to be called to process requests that have been
|
|
|
|
* placed on the queue.
|
|
|
|
* @lock: Request queue spin lock
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* If a block device wishes to use the standard request handling procedures,
|
|
|
|
* which sorts requests and coalesces adjacent requests, then it must
|
|
|
|
* call blk_init_queue(). The function @rfn will be called when there
|
|
|
|
* are requests on the queue that need to be processed. If the device
|
|
|
|
* supports plugging, then @rfn may not be called immediately when requests
|
|
|
|
* are available on the queue, but may be called at some time later instead.
|
|
|
|
* Plugged queues are generally unplugged when a buffer belonging to one
|
|
|
|
* of the requests on the queue is needed, or due to memory pressure.
|
|
|
|
*
|
|
|
|
* @rfn is not required, or even expected, to remove all requests off the
|
|
|
|
* queue, but only as many as it can handle at a time. If it does leave
|
|
|
|
* requests on the queue, it is responsible for arranging that the requests
|
|
|
|
* get dealt with eventually.
|
|
|
|
*
|
|
|
|
* The queue spin lock must be held while manipulating the requests on the
|
2006-06-05 12:09:01 +02:00
|
|
|
* request queue; this lock will be taken also from interrupt context, so irq
|
|
|
|
* disabling is needed for it.
|
2005-04-17 00:20:36 +02:00
|
|
|
*
|
|
|
|
* Function returns a pointer to the initialized request queue, or NULL if
|
|
|
|
* it didn't succeed.
|
|
|
|
*
|
|
|
|
* Note:
|
|
|
|
* blk_init_queue() must be paired with a blk_cleanup_queue() call
|
|
|
|
* when the block device is deactivated (such as at module unload).
|
|
|
|
**/
|
2005-06-23 09:08:19 +02:00
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *blk_init_queue(request_fn_proc *rfn, spinlock_t *lock)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2005-06-23 09:08:19 +02:00
|
|
|
return blk_init_queue_node(rfn, lock, -1);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_init_queue);
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *
|
2005-06-23 09:08:19 +02:00
|
|
|
blk_init_queue_node(request_fn_proc *rfn, spinlock_t *lock, int node_id)
|
|
|
|
{
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q = blk_alloc_queue_node(GFP_KERNEL, node_id);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (!q)
|
|
|
|
return NULL;
|
|
|
|
|
2005-06-23 09:08:19 +02:00
|
|
|
q->node = node_id;
|
2006-03-18 19:50:00 +01:00
|
|
|
if (blk_init_free_list(q)) {
|
|
|
|
kmem_cache_free(requestq_cachep, q);
|
|
|
|
return NULL;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2005-04-12 23:22:06 +02:00
|
|
|
/*
|
|
|
|
* if caller didn't supply a lock, they get per-queue locking with
|
|
|
|
* our embedded lock
|
|
|
|
*/
|
|
|
|
if (!lock) {
|
|
|
|
spin_lock_init(&q->__queue_lock);
|
|
|
|
lock = &q->__queue_lock;
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
q->request_fn = rfn;
|
|
|
|
q->prep_rq_fn = NULL;
|
|
|
|
q->unplug_fn = generic_unplug_device;
|
|
|
|
q->queue_flags = (1 << QUEUE_FLAG_CLUSTER);
|
|
|
|
q->queue_lock = lock;
|
|
|
|
|
|
|
|
blk_queue_segment_boundary(q, 0xffffffff);
|
|
|
|
|
|
|
|
blk_queue_make_request(q, __make_request);
|
|
|
|
blk_queue_max_segment_size(q, MAX_SEGMENT_SIZE);
|
|
|
|
|
|
|
|
blk_queue_max_hw_segments(q, MAX_HW_SEGMENTS);
|
|
|
|
blk_queue_max_phys_segments(q, MAX_PHYS_SEGMENTS);
|
|
|
|
|
2007-02-20 17:01:57 +01:00
|
|
|
q->sg_reserved_size = INT_MAX;
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* all done
|
|
|
|
*/
|
|
|
|
if (!elevator_init(q, NULL)) {
|
|
|
|
blk_queue_congestion_threshold(q);
|
|
|
|
return q;
|
|
|
|
}
|
|
|
|
|
2006-03-18 19:50:00 +01:00
|
|
|
blk_put_queue(q);
|
2005-04-17 00:20:36 +02:00
|
|
|
return NULL;
|
|
|
|
}
|
2005-06-23 09:08:19 +02:00
|
|
|
EXPORT_SYMBOL(blk_init_queue_node);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
int blk_get_queue(struct request_queue *q)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2005-06-23 09:08:53 +02:00
|
|
|
if (likely(!test_bit(QUEUE_FLAG_DEAD, &q->queue_flags))) {
|
2006-03-19 00:34:37 +01:00
|
|
|
kobject_get(&q->kobj);
|
2005-04-17 00:20:36 +02:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_get_queue);
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static inline void blk_free_request(struct request_queue *q, struct request *rq)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-08-10 08:44:47 +02:00
|
|
|
if (rq->cmd_flags & REQ_ELVPRIV)
|
2005-10-28 08:29:39 +02:00
|
|
|
elv_put_request(q, rq);
|
2005-04-17 00:20:36 +02:00
|
|
|
mempool_free(rq, q->rq.rq_pool);
|
|
|
|
}
|
|
|
|
|
2006-07-18 22:24:11 +02:00
|
|
|
static struct request *
|
2007-07-24 09:28:11 +02:00
|
|
|
blk_alloc_request(struct request_queue *q, int rw, int priv, gfp_t gfp_mask)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct request *rq = mempool_alloc(q->rq.rq_pool, gfp_mask);
|
|
|
|
|
|
|
|
if (!rq)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
/*
|
2006-08-10 08:44:47 +02:00
|
|
|
* first three bits are identical in rq->cmd_flags and bio->bi_rw,
|
2005-04-17 00:20:36 +02:00
|
|
|
* see bio.h and blkdev.h
|
|
|
|
*/
|
2006-08-10 08:59:11 +02:00
|
|
|
rq->cmd_flags = rw | REQ_ALLOCED;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2005-10-28 08:29:39 +02:00
|
|
|
if (priv) {
|
2006-07-28 09:32:57 +02:00
|
|
|
if (unlikely(elv_set_request(q, rq, gfp_mask))) {
|
2005-10-28 08:29:39 +02:00
|
|
|
mempool_free(rq, q->rq.rq_pool);
|
|
|
|
return NULL;
|
|
|
|
}
|
2006-08-10 08:44:47 +02:00
|
|
|
rq->cmd_flags |= REQ_ELVPRIV;
|
2005-10-28 08:29:39 +02:00
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2005-10-28 08:29:39 +02:00
|
|
|
return rq;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* ioc_batching returns true if the ioc is a valid batching request and
|
|
|
|
* should be given priority access to a request.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
static inline int ioc_batching(struct request_queue *q, struct io_context *ioc)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
if (!ioc)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Make sure the process is able to allocate at least 1 request
|
|
|
|
* even if the batch times out, otherwise we could theoretically
|
|
|
|
* lose wakeups.
|
|
|
|
*/
|
|
|
|
return ioc->nr_batch_requests == q->nr_batching ||
|
|
|
|
(ioc->nr_batch_requests > 0
|
|
|
|
&& time_before(jiffies, ioc->last_waited + BLK_BATCH_TIME));
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* ioc_set_batching sets ioc to be a new "batcher" if it is not one. This
|
|
|
|
* will cause the process to be a "batcher" on all queues in the system. This
|
|
|
|
* is the behaviour we want though - once it gets a wakeup it should be given
|
|
|
|
* a nice run.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
static void ioc_set_batching(struct request_queue *q, struct io_context *ioc)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
if (!ioc || ioc_batching(q, ioc))
|
|
|
|
return;
|
|
|
|
|
|
|
|
ioc->nr_batch_requests = q->nr_batching;
|
|
|
|
ioc->last_waited = jiffies;
|
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static void __freed_request(struct request_queue *q, int rw)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct request_list *rl = &q->rq;
|
|
|
|
|
|
|
|
if (rl->count[rw] < queue_congestion_off_threshold(q))
|
2006-10-20 08:28:15 +02:00
|
|
|
blk_clear_queue_congested(q, rw);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (rl->count[rw] + 1 <= q->nr_requests) {
|
|
|
|
if (waitqueue_active(&rl->wait[rw]))
|
|
|
|
wake_up(&rl->wait[rw]);
|
|
|
|
|
|
|
|
blk_clear_queue_full(q, rw);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* A request has just been released. Account for it, update the full and
|
|
|
|
* congestion status, wake up any waiters. Called under q->queue_lock.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
static void freed_request(struct request_queue *q, int rw, int priv)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct request_list *rl = &q->rq;
|
|
|
|
|
|
|
|
rl->count[rw]--;
|
2005-10-28 08:29:39 +02:00
|
|
|
if (priv)
|
|
|
|
rl->elvpriv--;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
__freed_request(q, rw);
|
|
|
|
|
|
|
|
if (unlikely(rl->starved[rw ^ 1]))
|
|
|
|
__freed_request(q, rw ^ 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
#define blkdev_free_rq(list) list_entry((list)->next, struct request, queuelist)
|
|
|
|
/*
|
2005-06-29 05:45:14 +02:00
|
|
|
* Get a free request, queue_lock must be held.
|
|
|
|
* Returns NULL on failure, with queue_lock held.
|
|
|
|
* Returns !NULL on success, with queue_lock *not held*.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
static struct request *get_request(struct request_queue *q, int rw_flags,
|
2006-12-13 13:02:26 +01:00
|
|
|
struct bio *bio, gfp_t gfp_mask)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct request *rq = NULL;
|
|
|
|
struct request_list *rl = &q->rq;
|
2005-11-12 11:09:12 +01:00
|
|
|
struct io_context *ioc = NULL;
|
2006-12-13 13:02:26 +01:00
|
|
|
const int rw = rw_flags & 0x01;
|
2005-11-12 11:09:12 +01:00
|
|
|
int may_queue, priv;
|
|
|
|
|
2006-12-13 13:02:26 +01:00
|
|
|
may_queue = elv_may_queue(q, rw_flags);
|
2005-11-12 11:09:12 +01:00
|
|
|
if (may_queue == ELV_MQUEUE_NO)
|
|
|
|
goto rq_starved;
|
|
|
|
|
|
|
|
if (rl->count[rw]+1 >= queue_congestion_on_threshold(q)) {
|
|
|
|
if (rl->count[rw]+1 >= q->nr_requests) {
|
2006-07-19 23:39:40 +02:00
|
|
|
ioc = current_io_context(GFP_ATOMIC, q->node);
|
2005-11-12 11:09:12 +01:00
|
|
|
/*
|
|
|
|
* The queue will fill after this allocation, so set
|
|
|
|
* it as full, and mark this process as "batching".
|
|
|
|
* This process will be allowed to complete a batch of
|
|
|
|
* requests, others will be blocked.
|
|
|
|
*/
|
|
|
|
if (!blk_queue_full(q, rw)) {
|
|
|
|
ioc_set_batching(q, ioc);
|
|
|
|
blk_set_queue_full(q, rw);
|
|
|
|
} else {
|
|
|
|
if (may_queue != ELV_MQUEUE_MUST
|
|
|
|
&& !ioc_batching(q, ioc)) {
|
|
|
|
/*
|
|
|
|
* The queue is full and the allocating
|
|
|
|
* process is not a "batcher", and not
|
|
|
|
* exempted by the IO scheduler
|
|
|
|
*/
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2006-10-20 08:28:15 +02:00
|
|
|
blk_set_queue_congested(q, rw);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2005-06-28 16:35:11 +02:00
|
|
|
/*
|
|
|
|
* Only allow batching queuers to allocate up to 50% over the defined
|
|
|
|
* limit of requests, otherwise we could have thousands of requests
|
|
|
|
* allocated with any setting of ->nr_requests
|
|
|
|
*/
|
2005-06-29 16:15:40 +02:00
|
|
|
if (rl->count[rw] >= (3 * q->nr_requests / 2))
|
2005-06-28 16:35:11 +02:00
|
|
|
goto out;
|
2005-06-29 16:15:40 +02:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
rl->count[rw]++;
|
|
|
|
rl->starved[rw] = 0;
|
2005-10-28 08:29:39 +02:00
|
|
|
|
2005-10-28 08:30:39 +02:00
|
|
|
priv = !test_bit(QUEUE_FLAG_ELVSWITCH, &q->queue_flags);
|
2005-10-28 08:29:39 +02:00
|
|
|
if (priv)
|
|
|
|
rl->elvpriv++;
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
spin_unlock_irq(q->queue_lock);
|
|
|
|
|
2006-12-13 13:02:26 +01:00
|
|
|
rq = blk_alloc_request(q, rw_flags, priv, gfp_mask);
|
2005-11-12 11:09:12 +01:00
|
|
|
if (unlikely(!rq)) {
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* Allocation failed presumably due to memory. Undo anything
|
|
|
|
* we might have messed up.
|
|
|
|
*
|
|
|
|
* Allocating task should really be put onto the front of the
|
|
|
|
* wait queue, but this is pretty rare.
|
|
|
|
*/
|
|
|
|
spin_lock_irq(q->queue_lock);
|
2005-10-28 08:29:39 +02:00
|
|
|
freed_request(q, rw, priv);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* in the very unlikely event that allocation failed and no
|
|
|
|
* requests for this direction was pending, mark us starved
|
|
|
|
* so that freeing of a request in the other direction will
|
|
|
|
* notice us. another possible fix would be to split the
|
|
|
|
* rq mempool into READ and WRITE
|
|
|
|
*/
|
|
|
|
rq_starved:
|
|
|
|
if (unlikely(rl->count[rw] == 0))
|
|
|
|
rl->starved[rw] = 1;
|
|
|
|
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2005-11-12 11:09:12 +01:00
|
|
|
/*
|
|
|
|
* ioc may be NULL here, and ioc_batching will be false. That's
|
|
|
|
* OK, if the queue is under the request limit then requests need
|
|
|
|
* not count toward the nr_batch_requests limit. There will always
|
|
|
|
* be some limit enforced by BLK_BATCH_TIME.
|
|
|
|
*/
|
2005-04-17 00:20:36 +02:00
|
|
|
if (ioc_batching(q, ioc))
|
|
|
|
ioc->nr_batch_requests--;
|
|
|
|
|
|
|
|
rq_init(q, rq);
|
2006-03-23 20:00:26 +01:00
|
|
|
|
|
|
|
blk_add_trace_generic(q, bio, rw, BLK_TA_GETRQ);
|
2005-04-17 00:20:36 +02:00
|
|
|
out:
|
|
|
|
return rq;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* No available requests for this queue, unplug the device and wait for some
|
|
|
|
* requests to become available.
|
2005-06-29 05:45:14 +02:00
|
|
|
*
|
|
|
|
* Called with q->queue_lock held, and returns with it unlocked.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
static struct request *get_request_wait(struct request_queue *q, int rw_flags,
|
2005-06-27 10:55:12 +02:00
|
|
|
struct bio *bio)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-12-13 13:02:26 +01:00
|
|
|
const int rw = rw_flags & 0x01;
|
2005-04-17 00:20:36 +02:00
|
|
|
struct request *rq;
|
|
|
|
|
2006-12-13 13:02:26 +01:00
|
|
|
rq = get_request(q, rw_flags, bio, GFP_NOIO);
|
2005-06-29 05:45:13 +02:00
|
|
|
while (!rq) {
|
|
|
|
DEFINE_WAIT(wait);
|
2005-04-17 00:20:36 +02:00
|
|
|
struct request_list *rl = &q->rq;
|
|
|
|
|
|
|
|
prepare_to_wait_exclusive(&rl->wait[rw], &wait,
|
|
|
|
TASK_UNINTERRUPTIBLE);
|
|
|
|
|
2006-12-13 13:02:26 +01:00
|
|
|
rq = get_request(q, rw_flags, bio, GFP_NOIO);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (!rq) {
|
|
|
|
struct io_context *ioc;
|
|
|
|
|
2006-03-23 20:00:26 +01:00
|
|
|
blk_add_trace_generic(q, bio, rw, BLK_TA_SLEEPRQ);
|
|
|
|
|
2005-06-29 05:45:14 +02:00
|
|
|
__generic_unplug_device(q);
|
|
|
|
spin_unlock_irq(q->queue_lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
io_schedule();
|
|
|
|
|
|
|
|
/*
|
|
|
|
* After sleeping, we become a "batching" process and
|
|
|
|
* will be able to allocate at least one request, and
|
|
|
|
* up to a big batch of them for a small period time.
|
|
|
|
* See ioc_batching, ioc_set_batching
|
|
|
|
*/
|
2006-07-19 23:39:40 +02:00
|
|
|
ioc = current_io_context(GFP_NOIO, q->node);
|
2005-04-17 00:20:36 +02:00
|
|
|
ioc_set_batching(q, ioc);
|
2005-06-29 05:45:14 +02:00
|
|
|
|
|
|
|
spin_lock_irq(q->queue_lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
finish_wait(&rl->wait[rw], &wait);
|
2005-06-29 05:45:13 +02:00
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
return rq;
|
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request *blk_get_request(struct request_queue *q, int rw, gfp_t gfp_mask)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct request *rq;
|
|
|
|
|
|
|
|
BUG_ON(rw != READ && rw != WRITE);
|
|
|
|
|
2005-06-29 05:45:14 +02:00
|
|
|
spin_lock_irq(q->queue_lock);
|
|
|
|
if (gfp_mask & __GFP_WAIT) {
|
2005-06-27 10:55:12 +02:00
|
|
|
rq = get_request_wait(q, rw, NULL);
|
2005-06-29 05:45:14 +02:00
|
|
|
} else {
|
2005-06-27 10:55:12 +02:00
|
|
|
rq = get_request(q, rw, NULL, gfp_mask);
|
2005-06-29 05:45:14 +02:00
|
|
|
if (!rq)
|
|
|
|
spin_unlock_irq(q->queue_lock);
|
|
|
|
}
|
|
|
|
/* q->queue_lock is unlocked at this point */
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
return rq;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_get_request);
|
|
|
|
|
2006-07-20 14:54:05 +02:00
|
|
|
/**
|
|
|
|
* blk_start_queueing - initiate dispatch of requests to device
|
|
|
|
* @q: request queue to kick into gear
|
|
|
|
*
|
|
|
|
* This is basically a helper to remove the need to know whether a queue
|
|
|
|
* is plugged or not if someone just wants to initiate dispatch of requests
|
|
|
|
* for this queue.
|
|
|
|
*
|
|
|
|
* The queue lock must be held with interrupts disabled.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_start_queueing(struct request_queue *q)
|
2006-07-20 14:54:05 +02:00
|
|
|
{
|
|
|
|
if (!blk_queue_plugged(q))
|
|
|
|
q->request_fn(q);
|
|
|
|
else
|
|
|
|
__generic_unplug_device(q);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_start_queueing);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/**
|
|
|
|
* blk_requeue_request - put a request back on queue
|
|
|
|
* @q: request queue where request should be inserted
|
|
|
|
* @rq: request to be inserted
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Drivers often keep queueing requests until the hardware cannot accept
|
|
|
|
* more, when that condition happens we need to put the request back
|
|
|
|
* on the queue. Must be called with queue lock held.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_requeue_request(struct request_queue *q, struct request *rq)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-03-23 20:00:26 +01:00
|
|
|
blk_add_trace_rq(q, rq, BLK_TA_REQUEUE);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
if (blk_rq_tagged(rq))
|
|
|
|
blk_queue_end_tag(q, rq);
|
|
|
|
|
|
|
|
elv_requeue_request(q, rq);
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_requeue_request);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_insert_request - insert a special request in to a request queue
|
|
|
|
* @q: request queue where request should be inserted
|
|
|
|
* @rq: request to be inserted
|
|
|
|
* @at_head: insert request at head or tail of queue
|
|
|
|
* @data: private data
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Many block devices need to execute commands asynchronously, so they don't
|
|
|
|
* block the whole kernel from preemption during request execution. This is
|
|
|
|
* accomplished normally by inserting aritficial requests tagged as
|
|
|
|
* REQ_SPECIAL in to the corresponding request queue, and letting them be
|
|
|
|
* scheduled for actual execution by the request queue.
|
|
|
|
*
|
|
|
|
* We have the option of inserting the head or the tail of the queue.
|
|
|
|
* Typically we use the tail for new ioctls and so forth. We use the head
|
|
|
|
* of the queue for things like a QUEUE_FULL message from a device, or a
|
|
|
|
* host that is unable to accept a particular command.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_insert_request(struct request_queue *q, struct request *rq,
|
2005-04-24 09:06:05 +02:00
|
|
|
int at_head, void *data)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2005-04-24 09:06:05 +02:00
|
|
|
int where = at_head ? ELEVATOR_INSERT_FRONT : ELEVATOR_INSERT_BACK;
|
2005-04-17 00:20:36 +02:00
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* tell I/O scheduler that this isn't a regular read/write (ie it
|
|
|
|
* must not attempt merges on this) and that it acts as a soft
|
|
|
|
* barrier
|
|
|
|
*/
|
2006-08-10 08:44:47 +02:00
|
|
|
rq->cmd_type = REQ_TYPE_SPECIAL;
|
|
|
|
rq->cmd_flags |= REQ_SOFTBARRIER;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
rq->special = data;
|
|
|
|
|
|
|
|
spin_lock_irqsave(q->queue_lock, flags);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If command is tagged, release the tag
|
|
|
|
*/
|
2005-04-24 09:06:05 +02:00
|
|
|
if (blk_rq_tagged(rq))
|
|
|
|
blk_queue_end_tag(q, rq);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-10-23 15:05:46 +02:00
|
|
|
drive_stat_acct(rq, 1);
|
2005-04-24 09:06:05 +02:00
|
|
|
__elv_add_request(q, rq, where, 0);
|
2006-07-20 14:54:05 +02:00
|
|
|
blk_start_queueing(q);
|
2005-04-17 00:20:36 +02:00
|
|
|
spin_unlock_irqrestore(q->queue_lock, flags);
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_insert_request);
|
|
|
|
|
2006-12-01 10:40:55 +01:00
|
|
|
static int __blk_rq_unmap_user(struct bio *bio)
|
|
|
|
{
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
if (bio) {
|
|
|
|
if (bio_flagged(bio, BIO_USER_MAPPED))
|
|
|
|
bio_unmap_user(bio);
|
|
|
|
else
|
|
|
|
ret = bio_uncopy_user(bio);
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2007-08-16 13:31:27 +02:00
|
|
|
int blk_rq_append_bio(struct request_queue *q, struct request *rq,
|
|
|
|
struct bio *bio)
|
|
|
|
{
|
|
|
|
if (!rq->bio)
|
|
|
|
blk_rq_bio_prep(q, rq, bio);
|
|
|
|
else if (!ll_back_merge_fn(q, rq, bio))
|
|
|
|
return -EINVAL;
|
|
|
|
else {
|
|
|
|
rq->biotail->bi_next = bio;
|
|
|
|
rq->biotail = bio;
|
|
|
|
|
|
|
|
rq->data_len += bio->bi_size;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_rq_append_bio);
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static int __blk_rq_map_user(struct request_queue *q, struct request *rq,
|
2006-12-01 10:40:55 +01:00
|
|
|
void __user *ubuf, unsigned int len)
|
|
|
|
{
|
|
|
|
unsigned long uaddr;
|
|
|
|
struct bio *bio, *orig_bio;
|
|
|
|
int reading, ret;
|
|
|
|
|
|
|
|
reading = rq_data_dir(rq) == READ;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* if alignment requirement is satisfied, map in user pages for
|
|
|
|
* direct dma. else, set up kernel bounce buffers
|
|
|
|
*/
|
|
|
|
uaddr = (unsigned long) ubuf;
|
|
|
|
if (!(uaddr & queue_dma_alignment(q)) && !(len & queue_dma_alignment(q)))
|
|
|
|
bio = bio_map_user(q, NULL, uaddr, len, reading);
|
|
|
|
else
|
|
|
|
bio = bio_copy_user(q, uaddr, len, reading);
|
|
|
|
|
2006-12-19 08:27:31 +01:00
|
|
|
if (IS_ERR(bio))
|
2006-12-01 10:40:55 +01:00
|
|
|
return PTR_ERR(bio);
|
|
|
|
|
|
|
|
orig_bio = bio;
|
|
|
|
blk_queue_bounce(q, &bio);
|
2006-12-19 08:27:31 +01:00
|
|
|
|
2006-12-01 10:40:55 +01:00
|
|
|
/*
|
|
|
|
* We link the bounce buffer in and could have to traverse it
|
|
|
|
* later so we have to get a ref to prevent it from being freed
|
|
|
|
*/
|
|
|
|
bio_get(bio);
|
|
|
|
|
2007-08-16 13:31:27 +02:00
|
|
|
ret = blk_rq_append_bio(q, rq, bio);
|
|
|
|
if (!ret)
|
|
|
|
return bio->bi_size;
|
2006-12-01 10:40:55 +01:00
|
|
|
|
|
|
|
/* if it was boucned we must call the end io function */
|
2007-09-27 12:47:43 +02:00
|
|
|
bio_endio(bio, 0);
|
2006-12-01 10:40:55 +01:00
|
|
|
__blk_rq_unmap_user(orig_bio);
|
|
|
|
bio_put(bio);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/**
|
|
|
|
* blk_rq_map_user - map user data to a request, for REQ_BLOCK_PC usage
|
|
|
|
* @q: request queue where request should be inserted
|
2005-06-20 14:21:01 +02:00
|
|
|
* @rq: request structure to fill
|
2005-04-17 00:20:36 +02:00
|
|
|
* @ubuf: the user buffer
|
|
|
|
* @len: length of user data
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Data will be mapped directly for zero copy io, if possible. Otherwise
|
|
|
|
* a kernel bounce buffer is used.
|
|
|
|
*
|
|
|
|
* A matching blk_rq_unmap_user() must be issued at the end of io, while
|
|
|
|
* still in process context.
|
|
|
|
*
|
|
|
|
* Note: The mapped bio may need to be bounced through blk_queue_bounce()
|
|
|
|
* before being submitted to the device, as pages mapped may be out of
|
|
|
|
* reach. It's the callers responsibility to make sure this happens. The
|
|
|
|
* original bio must be passed back in to blk_rq_unmap_user() for proper
|
|
|
|
* unmapping.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
int blk_rq_map_user(struct request_queue *q, struct request *rq,
|
|
|
|
void __user *ubuf, unsigned long len)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-12-01 10:40:55 +01:00
|
|
|
unsigned long bytes_read = 0;
|
2006-12-19 11:12:46 +01:00
|
|
|
struct bio *bio = NULL;
|
2006-12-01 10:40:55 +01:00
|
|
|
int ret;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2005-12-05 09:37:06 +01:00
|
|
|
if (len > (q->max_hw_sectors << 9))
|
2005-06-20 14:06:01 +02:00
|
|
|
return -EINVAL;
|
|
|
|
if (!len || !ubuf)
|
|
|
|
return -EINVAL;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-12-01 10:40:55 +01:00
|
|
|
while (bytes_read != len) {
|
|
|
|
unsigned long map_len, end, start;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-12-01 10:40:55 +01:00
|
|
|
map_len = min_t(unsigned long, len - bytes_read, BIO_MAX_SIZE);
|
|
|
|
end = ((unsigned long)ubuf + map_len + PAGE_SIZE - 1)
|
|
|
|
>> PAGE_SHIFT;
|
|
|
|
start = (unsigned long)ubuf >> PAGE_SHIFT;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-12-01 10:40:55 +01:00
|
|
|
/*
|
|
|
|
* A bad offset could cause us to require BIO_MAX_PAGES + 1
|
|
|
|
* pages. If this happens we just lower the requested
|
|
|
|
* mapping len by a page so that we can fit
|
|
|
|
*/
|
|
|
|
if (end - start > BIO_MAX_PAGES)
|
|
|
|
map_len -= PAGE_SIZE;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-12-01 10:40:55 +01:00
|
|
|
ret = __blk_rq_map_user(q, rq, ubuf, map_len);
|
|
|
|
if (ret < 0)
|
|
|
|
goto unmap_rq;
|
2006-12-19 11:12:46 +01:00
|
|
|
if (!bio)
|
|
|
|
bio = rq->bio;
|
2006-12-01 10:40:55 +01:00
|
|
|
bytes_read += ret;
|
|
|
|
ubuf += ret;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2006-12-01 10:40:55 +01:00
|
|
|
rq->buffer = rq->data = NULL;
|
|
|
|
return 0;
|
|
|
|
unmap_rq:
|
2006-12-19 11:12:46 +01:00
|
|
|
blk_rq_unmap_user(bio);
|
2006-12-01 10:40:55 +01:00
|
|
|
return ret;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_rq_map_user);
|
|
|
|
|
2005-06-20 14:06:52 +02:00
|
|
|
/**
|
|
|
|
* blk_rq_map_user_iov - map user data to a request, for REQ_BLOCK_PC usage
|
|
|
|
* @q: request queue where request should be inserted
|
|
|
|
* @rq: request to map data to
|
|
|
|
* @iov: pointer to the iovec
|
|
|
|
* @iov_count: number of elements in the iovec
|
2006-12-22 10:06:52 +01:00
|
|
|
* @len: I/O byte count
|
2005-06-20 14:06:52 +02:00
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Data will be mapped directly for zero copy io, if possible. Otherwise
|
|
|
|
* a kernel bounce buffer is used.
|
|
|
|
*
|
|
|
|
* A matching blk_rq_unmap_user() must be issued at the end of io, while
|
|
|
|
* still in process context.
|
|
|
|
*
|
|
|
|
* Note: The mapped bio may need to be bounced through blk_queue_bounce()
|
|
|
|
* before being submitted to the device, as pages mapped may be out of
|
|
|
|
* reach. It's the callers responsibility to make sure this happens. The
|
|
|
|
* original bio must be passed back in to blk_rq_unmap_user() for proper
|
|
|
|
* unmapping.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
int blk_rq_map_user_iov(struct request_queue *q, struct request *rq,
|
2006-12-01 10:40:55 +01:00
|
|
|
struct sg_iovec *iov, int iov_count, unsigned int len)
|
2005-06-20 14:06:52 +02:00
|
|
|
{
|
|
|
|
struct bio *bio;
|
|
|
|
|
|
|
|
if (!iov || iov_count <= 0)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
/* we don't allow misaligned data like bio_map_user() does. If the
|
|
|
|
* user is using sg, they're expected to know the alignment constraints
|
|
|
|
* and respect them accordingly */
|
|
|
|
bio = bio_map_user_iov(q, NULL, iov, iov_count, rq_data_dir(rq)== READ);
|
|
|
|
if (IS_ERR(bio))
|
|
|
|
return PTR_ERR(bio);
|
|
|
|
|
2006-12-01 10:40:55 +01:00
|
|
|
if (bio->bi_size != len) {
|
2007-09-27 12:47:43 +02:00
|
|
|
bio_endio(bio, 0);
|
2006-12-01 10:40:55 +01:00
|
|
|
bio_unmap_user(bio);
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
bio_get(bio);
|
2005-06-20 14:06:52 +02:00
|
|
|
blk_rq_bio_prep(q, rq, bio);
|
|
|
|
rq->buffer = rq->data = NULL;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_rq_map_user_iov);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/**
|
|
|
|
* blk_rq_unmap_user - unmap a request with user data
|
2006-12-19 11:12:46 +01:00
|
|
|
* @bio: start of bio list
|
2005-04-17 00:20:36 +02:00
|
|
|
*
|
|
|
|
* Description:
|
2006-12-19 11:12:46 +01:00
|
|
|
* Unmap a rq previously mapped by blk_rq_map_user(). The caller must
|
|
|
|
* supply the original rq->bio from the blk_rq_map_user() return, since
|
|
|
|
* the io completion may have changed rq->bio.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
2006-12-19 11:12:46 +01:00
|
|
|
int blk_rq_unmap_user(struct bio *bio)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-12-19 11:12:46 +01:00
|
|
|
struct bio *mapped_bio;
|
2006-12-19 11:07:59 +01:00
|
|
|
int ret = 0, ret2;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-12-19 11:12:46 +01:00
|
|
|
while (bio) {
|
|
|
|
mapped_bio = bio;
|
|
|
|
if (unlikely(bio_flagged(bio, BIO_BOUNCED)))
|
2006-12-01 10:40:55 +01:00
|
|
|
mapped_bio = bio->bi_private;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-12-19 11:07:59 +01:00
|
|
|
ret2 = __blk_rq_unmap_user(mapped_bio);
|
|
|
|
if (ret2 && !ret)
|
|
|
|
ret = ret2;
|
|
|
|
|
2006-12-19 11:12:46 +01:00
|
|
|
mapped_bio = bio;
|
|
|
|
bio = bio->bi_next;
|
|
|
|
bio_put(mapped_bio);
|
2006-12-01 10:40:55 +01:00
|
|
|
}
|
2006-12-19 11:07:59 +01:00
|
|
|
|
|
|
|
return ret;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_rq_unmap_user);
|
|
|
|
|
2005-06-20 14:04:44 +02:00
|
|
|
/**
|
|
|
|
* blk_rq_map_kern - map kernel data to a request, for REQ_BLOCK_PC usage
|
|
|
|
* @q: request queue where request should be inserted
|
2005-06-20 14:21:01 +02:00
|
|
|
* @rq: request to fill
|
2005-06-20 14:04:44 +02:00
|
|
|
* @kbuf: the kernel buffer
|
|
|
|
* @len: length of user data
|
2005-06-20 14:21:01 +02:00
|
|
|
* @gfp_mask: memory allocation flags
|
2005-06-20 14:04:44 +02:00
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
int blk_rq_map_kern(struct request_queue *q, struct request *rq, void *kbuf,
|
2005-10-21 09:20:53 +02:00
|
|
|
unsigned int len, gfp_t gfp_mask)
|
2005-06-20 14:04:44 +02:00
|
|
|
{
|
|
|
|
struct bio *bio;
|
|
|
|
|
2005-12-05 09:37:06 +01:00
|
|
|
if (len > (q->max_hw_sectors << 9))
|
2005-06-20 14:06:01 +02:00
|
|
|
return -EINVAL;
|
|
|
|
if (!len || !kbuf)
|
|
|
|
return -EINVAL;
|
2005-06-20 14:04:44 +02:00
|
|
|
|
|
|
|
bio = bio_map_kern(q, kbuf, len, gfp_mask);
|
2005-06-20 14:06:01 +02:00
|
|
|
if (IS_ERR(bio))
|
|
|
|
return PTR_ERR(bio);
|
2005-06-20 14:04:44 +02:00
|
|
|
|
2005-06-20 14:06:01 +02:00
|
|
|
if (rq_data_dir(rq) == WRITE)
|
|
|
|
bio->bi_rw |= (1 << BIO_RW);
|
2005-06-20 14:04:44 +02:00
|
|
|
|
2005-06-20 14:06:01 +02:00
|
|
|
blk_rq_bio_prep(q, rq, bio);
|
2007-05-08 19:12:23 +02:00
|
|
|
blk_queue_bounce(q, &rq->bio);
|
2005-06-20 14:06:01 +02:00
|
|
|
rq->buffer = rq->data = NULL;
|
|
|
|
return 0;
|
2005-06-20 14:04:44 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_rq_map_kern);
|
|
|
|
|
2005-06-20 14:21:01 +02:00
|
|
|
/**
|
|
|
|
* blk_execute_rq_nowait - insert a request into queue for execution
|
|
|
|
* @q: queue to insert the request in
|
|
|
|
* @bd_disk: matching gendisk
|
|
|
|
* @rq: request to insert
|
|
|
|
* @at_head: insert request at head or tail of queue
|
|
|
|
* @done: I/O completion handler
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Insert a fully prepared request at the back of the io scheduler queue
|
|
|
|
* for execution. Don't wait for completion.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
void blk_execute_rq_nowait(struct request_queue *q, struct gendisk *bd_disk,
|
2005-06-20 14:06:52 +02:00
|
|
|
struct request *rq, int at_head,
|
2006-01-06 09:49:03 +01:00
|
|
|
rq_end_io_fn *done)
|
2005-06-20 14:06:52 +02:00
|
|
|
{
|
|
|
|
int where = at_head ? ELEVATOR_INSERT_FRONT : ELEVATOR_INSERT_BACK;
|
|
|
|
|
|
|
|
rq->rq_disk = bd_disk;
|
2006-08-10 08:44:47 +02:00
|
|
|
rq->cmd_flags |= REQ_NOMERGE;
|
2005-06-20 14:06:52 +02:00
|
|
|
rq->end_io = done;
|
2006-03-22 08:08:01 +01:00
|
|
|
WARN_ON(irqs_disabled());
|
|
|
|
spin_lock_irq(q->queue_lock);
|
|
|
|
__elv_add_request(q, rq, where, 1);
|
|
|
|
__generic_unplug_device(q);
|
|
|
|
spin_unlock_irq(q->queue_lock);
|
2005-06-20 14:06:52 +02:00
|
|
|
}
|
2005-11-11 12:30:24 +01:00
|
|
|
EXPORT_SYMBOL_GPL(blk_execute_rq_nowait);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/**
|
|
|
|
* blk_execute_rq - insert a request into queue for execution
|
|
|
|
* @q: queue to insert the request in
|
|
|
|
* @bd_disk: matching gendisk
|
|
|
|
* @rq: request to insert
|
2005-06-20 14:11:09 +02:00
|
|
|
* @at_head: insert request at head or tail of queue
|
2005-04-17 00:20:36 +02:00
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Insert a fully prepared request at the back of the io scheduler queue
|
2005-06-20 14:21:01 +02:00
|
|
|
* for execution and wait for completion.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
int blk_execute_rq(struct request_queue *q, struct gendisk *bd_disk,
|
2005-06-20 14:11:09 +02:00
|
|
|
struct request *rq, int at_head)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-07-03 09:25:26 +02:00
|
|
|
DECLARE_COMPLETION_ONSTACK(wait);
|
2005-04-17 00:20:36 +02:00
|
|
|
char sense[SCSI_SENSE_BUFFERSIZE];
|
|
|
|
int err = 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* we need an extra reference to the request, so we can look at
|
|
|
|
* it after io completion
|
|
|
|
*/
|
|
|
|
rq->ref_count++;
|
|
|
|
|
|
|
|
if (!rq->sense) {
|
|
|
|
memset(sense, 0, sizeof(sense));
|
|
|
|
rq->sense = sense;
|
|
|
|
rq->sense_len = 0;
|
|
|
|
}
|
|
|
|
|
2006-09-30 20:29:12 +02:00
|
|
|
rq->end_io_data = &wait;
|
2005-06-20 14:11:09 +02:00
|
|
|
blk_execute_rq_nowait(q, bd_disk, rq, at_head, blk_end_sync_rq);
|
2005-04-17 00:20:36 +02:00
|
|
|
wait_for_completion(&wait);
|
|
|
|
|
|
|
|
if (rq->errors)
|
|
|
|
err = -EIO;
|
|
|
|
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_execute_rq);
|
|
|
|
|
2007-10-16 11:05:02 +02:00
|
|
|
static void bio_end_empty_barrier(struct bio *bio, int err)
|
|
|
|
{
|
|
|
|
if (err)
|
|
|
|
clear_bit(BIO_UPTODATE, &bio->bi_flags);
|
|
|
|
|
|
|
|
complete(bio->bi_private);
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/**
|
|
|
|
* blkdev_issue_flush - queue a flush
|
|
|
|
* @bdev: blockdev to issue flush for
|
|
|
|
* @error_sector: error sector
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Issue a flush for the block device in question. Caller can supply
|
|
|
|
* room for storing the error offset in case of a flush error, if they
|
|
|
|
* wish to. Caller must run wait_for_completion() on its own.
|
|
|
|
*/
|
|
|
|
int blkdev_issue_flush(struct block_device *bdev, sector_t *error_sector)
|
|
|
|
{
|
2007-10-16 11:05:02 +02:00
|
|
|
DECLARE_COMPLETION_ONSTACK(wait);
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q;
|
2007-10-16 11:05:02 +02:00
|
|
|
struct bio *bio;
|
|
|
|
int ret;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (bdev->bd_disk == NULL)
|
|
|
|
return -ENXIO;
|
|
|
|
|
|
|
|
q = bdev_get_queue(bdev);
|
|
|
|
if (!q)
|
|
|
|
return -ENXIO;
|
|
|
|
|
2007-10-16 11:05:02 +02:00
|
|
|
bio = bio_alloc(GFP_KERNEL, 0);
|
|
|
|
if (!bio)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
bio->bi_end_io = bio_end_empty_barrier;
|
|
|
|
bio->bi_private = &wait;
|
|
|
|
bio->bi_bdev = bdev;
|
|
|
|
submit_bio(1 << BIO_RW_BARRIER, bio);
|
|
|
|
|
|
|
|
wait_for_completion(&wait);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The driver must store the error location in ->bi_sector, if
|
|
|
|
* it supports it. For non-stacked drivers, this should be copied
|
|
|
|
* from rq->sector.
|
|
|
|
*/
|
|
|
|
if (error_sector)
|
|
|
|
*error_sector = bio->bi_sector;
|
|
|
|
|
|
|
|
ret = 0;
|
|
|
|
if (!bio_flagged(bio, BIO_UPTODATE))
|
|
|
|
ret = -EIO;
|
|
|
|
|
|
|
|
bio_put(bio);
|
|
|
|
return ret;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blkdev_issue_flush);
|
|
|
|
|
2007-10-23 15:05:46 +02:00
|
|
|
static void drive_stat_acct(struct request *rq, int new_io)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
int rw = rq_data_dir(rq);
|
|
|
|
|
|
|
|
if (!blk_fs_request(rq) || !rq->rq_disk)
|
|
|
|
return;
|
|
|
|
|
2005-11-01 08:35:42 +01:00
|
|
|
if (!new_io) {
|
2005-11-01 09:26:16 +01:00
|
|
|
__disk_stat_inc(rq->rq_disk, merges[rw]);
|
2005-11-01 08:35:42 +01:00
|
|
|
} else {
|
2005-04-17 00:20:36 +02:00
|
|
|
disk_round_stats(rq->rq_disk);
|
|
|
|
rq->rq_disk->in_flight++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* add-request adds a request to the linked list.
|
|
|
|
* queue lock is held and interrupts disabled, as we muck with the
|
|
|
|
* request queue list.
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
static inline void add_request(struct request_queue * q, struct request * req)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2007-10-23 15:05:46 +02:00
|
|
|
drive_stat_acct(req, 1);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* elevator indicated where it wants this request to be
|
|
|
|
* inserted at elevator_merge time
|
|
|
|
*/
|
|
|
|
__elv_add_request(q, req, ELEVATOR_INSERT_SORT, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* disk_round_stats() - Round off the performance stats on a struct
|
|
|
|
* disk_stats.
|
|
|
|
*
|
|
|
|
* The average IO queue length and utilisation statistics are maintained
|
|
|
|
* by observing the current state of the queue length and the amount of
|
|
|
|
* time it has been in this state for.
|
|
|
|
*
|
|
|
|
* Normally, that accounting is done on IO completion, but that can result
|
|
|
|
* in more than a second's worth of IO being accounted for within any one
|
|
|
|
* second, leading to >100% utilisation. To deal with that, we call this
|
|
|
|
* function to do a round-off before returning the results when reading
|
|
|
|
* /proc/diskstats. This accounts immediately for all queue usage up to
|
|
|
|
* the current jiffies and restarts the counters again.
|
|
|
|
*/
|
|
|
|
void disk_round_stats(struct gendisk *disk)
|
|
|
|
{
|
|
|
|
unsigned long now = jiffies;
|
|
|
|
|
2005-10-13 21:49:29 +02:00
|
|
|
if (now == disk->stamp)
|
|
|
|
return;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2005-10-13 21:48:42 +02:00
|
|
|
if (disk->in_flight) {
|
|
|
|
__disk_stat_add(disk, time_in_queue,
|
|
|
|
disk->in_flight * (now - disk->stamp));
|
|
|
|
__disk_stat_add(disk, io_ticks, (now - disk->stamp));
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
disk->stamp = now;
|
|
|
|
}
|
|
|
|
|
2006-02-01 12:04:53 +01:00
|
|
|
EXPORT_SYMBOL_GPL(disk_round_stats);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* queue lock must be held
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
void __blk_put_request(struct request_queue *q, struct request *req)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
if (unlikely(!q))
|
|
|
|
return;
|
|
|
|
if (unlikely(--req->ref_count))
|
|
|
|
return;
|
|
|
|
|
2005-10-20 16:23:44 +02:00
|
|
|
elv_completed_request(q, req);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* Request may not have originated from ll_rw_blk. if not,
|
|
|
|
* it didn't come out of our reserved rq pools
|
|
|
|
*/
|
2006-08-10 08:59:11 +02:00
|
|
|
if (req->cmd_flags & REQ_ALLOCED) {
|
2005-04-17 00:20:36 +02:00
|
|
|
int rw = rq_data_dir(req);
|
2006-08-10 08:44:47 +02:00
|
|
|
int priv = req->cmd_flags & REQ_ELVPRIV;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
BUG_ON(!list_empty(&req->queuelist));
|
2006-07-28 09:23:08 +02:00
|
|
|
BUG_ON(!hlist_unhashed(&req->hash));
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
blk_free_request(q, req);
|
2005-10-28 08:29:39 +02:00
|
|
|
freed_request(q, rw, priv);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2005-11-11 12:30:24 +01:00
|
|
|
EXPORT_SYMBOL_GPL(__blk_put_request);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
void blk_put_request(struct request *req)
|
|
|
|
{
|
2005-10-20 16:23:44 +02:00
|
|
|
unsigned long flags;
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q = req->q;
|
2005-10-20 16:23:44 +02:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
2005-10-20 16:23:44 +02:00
|
|
|
* Gee, IDE calls in w/ NULL q. Fix IDE and remove the
|
|
|
|
* following if (q) test.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
2005-10-20 16:23:44 +02:00
|
|
|
if (q) {
|
2005-04-17 00:20:36 +02:00
|
|
|
spin_lock_irqsave(q->queue_lock, flags);
|
|
|
|
__blk_put_request(q, req);
|
|
|
|
spin_unlock_irqrestore(q->queue_lock, flags);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_put_request);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_end_sync_rq - executes a completion event on a request
|
|
|
|
* @rq: request to complete
|
2006-01-31 15:24:34 +01:00
|
|
|
* @error: end io status of the request
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
2006-01-06 09:49:03 +01:00
|
|
|
void blk_end_sync_rq(struct request *rq, int error)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-09-30 20:29:12 +02:00
|
|
|
struct completion *waiting = rq->end_io_data;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-09-30 20:29:12 +02:00
|
|
|
rq->end_io_data = NULL;
|
2005-04-17 00:20:36 +02:00
|
|
|
__blk_put_request(rq->q, rq);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* complete last, if this is a stack request the process (and thus
|
|
|
|
* the rq pointer) could be invalid right after this complete()
|
|
|
|
*/
|
|
|
|
complete(waiting);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(blk_end_sync_rq);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Has to be called with the request spinlock acquired
|
|
|
|
*/
|
2007-07-24 09:28:11 +02:00
|
|
|
static int attempt_merge(struct request_queue *q, struct request *req,
|
2005-04-17 00:20:36 +02:00
|
|
|
struct request *next)
|
|
|
|
{
|
|
|
|
if (!rq_mergeable(req) || !rq_mergeable(next))
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/*
|
2006-06-26 18:35:02 +02:00
|
|
|
* not contiguous
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
|
|
|
if (req->sector + req->nr_sectors != next->sector)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
if (rq_data_dir(req) != rq_data_dir(next)
|
|
|
|
|| req->rq_disk != next->rq_disk
|
2006-09-30 20:29:12 +02:00
|
|
|
|| next->special)
|
2005-04-17 00:20:36 +02:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If we are allowed to merge, then append bio list
|
|
|
|
* from next to rq and release next. merge_requests_fn
|
|
|
|
* will have updated segment counts, update sector
|
|
|
|
* counts here.
|
|
|
|
*/
|
2006-12-19 08:33:11 +01:00
|
|
|
if (!ll_merge_requests_fn(q, req, next))
|
2005-04-17 00:20:36 +02:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* At this point we have either done a back merge
|
|
|
|
* or front merge. We need the smaller start_time of
|
|
|
|
* the merged requests to be the current request
|
|
|
|
* for accounting purposes.
|
|
|
|
*/
|
|
|
|
if (time_after(req->start_time, next->start_time))
|
|
|
|
req->start_time = next->start_time;
|
|
|
|
|
|
|
|
req->biotail->bi_next = next->bio;
|
|
|
|
req->biotail = next->biotail;
|
|
|
|
|
|
|
|
req->nr_sectors = req->hard_nr_sectors += next->hard_nr_sectors;
|
|
|
|
|
|
|
|
elv_merge_requests(q, req, next);
|
|
|
|
|
|
|
|
if (req->rq_disk) {
|
|
|
|
disk_round_stats(req->rq_disk);
|
|
|
|
req->rq_disk->in_flight--;
|
|
|
|
}
|
|
|
|
|
2005-06-27 10:55:12 +02:00
|
|
|
req->ioprio = ioprio_best(req->ioprio, next->ioprio);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
__blk_put_request(q, next);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static inline int attempt_back_merge(struct request_queue *q,
|
|
|
|
struct request *rq)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct request *next = elv_latter_request(q, rq);
|
|
|
|
|
|
|
|
if (next)
|
|
|
|
return attempt_merge(q, rq, next);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static inline int attempt_front_merge(struct request_queue *q,
|
|
|
|
struct request *rq)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct request *prev = elv_former_request(q, rq);
|
|
|
|
|
|
|
|
if (prev)
|
|
|
|
return attempt_merge(q, prev, rq);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2006-01-06 09:49:58 +01:00
|
|
|
static void init_request_from_bio(struct request *req, struct bio *bio)
|
|
|
|
{
|
2006-08-10 08:44:47 +02:00
|
|
|
req->cmd_type = REQ_TYPE_FS;
|
2006-01-06 09:49:58 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
* inherit FAILFAST from bio (for read-ahead, and explicit FAILFAST)
|
|
|
|
*/
|
|
|
|
if (bio_rw_ahead(bio) || bio_failfast(bio))
|
2006-08-10 08:44:47 +02:00
|
|
|
req->cmd_flags |= REQ_FAILFAST;
|
2006-01-06 09:49:58 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
* REQ_BARRIER implies no merging, but lets make it explicit
|
|
|
|
*/
|
|
|
|
if (unlikely(bio_barrier(bio)))
|
2006-08-10 08:44:47 +02:00
|
|
|
req->cmd_flags |= (REQ_HARDBARRIER | REQ_NOMERGE);
|
2006-01-06 09:49:58 +01:00
|
|
|
|
2006-06-13 08:26:10 +02:00
|
|
|
if (bio_sync(bio))
|
2006-08-10 08:44:47 +02:00
|
|
|
req->cmd_flags |= REQ_RW_SYNC;
|
2006-08-10 09:01:02 +02:00
|
|
|
if (bio_rw_meta(bio))
|
|
|
|
req->cmd_flags |= REQ_RW_META;
|
2006-06-13 08:26:10 +02:00
|
|
|
|
2006-01-06 09:49:58 +01:00
|
|
|
req->errors = 0;
|
|
|
|
req->hard_sector = req->sector = bio->bi_sector;
|
|
|
|
req->ioprio = bio_prio(bio);
|
|
|
|
req->start_time = jiffies;
|
2007-08-16 13:31:30 +02:00
|
|
|
blk_rq_bio_prep(req->q, req, bio);
|
2006-01-06 09:49:58 +01:00
|
|
|
}
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
static int __make_request(struct request_queue *q, struct bio *bio)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2005-06-29 05:45:13 +02:00
|
|
|
struct request *req;
|
2006-07-18 04:14:45 +02:00
|
|
|
int el_ret, nr_sectors, barrier, err;
|
|
|
|
const unsigned short prio = bio_prio(bio);
|
|
|
|
const int sync = bio_sync(bio);
|
2006-12-13 13:02:26 +01:00
|
|
|
int rw_flags;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
nr_sectors = bio_sectors(bio);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* low level driver can indicate that it wants pages above a
|
|
|
|
* certain limit bounced to low memory (ie for highmem, or even
|
|
|
|
* ISA dma in theory)
|
|
|
|
*/
|
|
|
|
blk_queue_bounce(q, &bio);
|
|
|
|
|
|
|
|
barrier = bio_barrier(bio);
|
2006-01-06 09:51:03 +01:00
|
|
|
if (unlikely(barrier) && (q->next_ordered == QUEUE_ORDERED_NONE)) {
|
2005-04-17 00:20:36 +02:00
|
|
|
err = -EOPNOTSUPP;
|
|
|
|
goto end_io;
|
|
|
|
}
|
|
|
|
|
|
|
|
spin_lock_irq(q->queue_lock);
|
|
|
|
|
2005-06-29 05:45:13 +02:00
|
|
|
if (unlikely(barrier) || elv_queue_empty(q))
|
2005-04-17 00:20:36 +02:00
|
|
|
goto get_rq;
|
|
|
|
|
|
|
|
el_ret = elv_merge(q, &req, bio);
|
|
|
|
switch (el_ret) {
|
|
|
|
case ELEVATOR_BACK_MERGE:
|
|
|
|
BUG_ON(!rq_mergeable(req));
|
|
|
|
|
2006-12-19 08:33:11 +01:00
|
|
|
if (!ll_back_merge_fn(q, req, bio))
|
2005-04-17 00:20:36 +02:00
|
|
|
break;
|
|
|
|
|
2006-03-23 20:00:26 +01:00
|
|
|
blk_add_trace_bio(q, bio, BLK_TA_BACKMERGE);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
req->biotail->bi_next = bio;
|
|
|
|
req->biotail = bio;
|
|
|
|
req->nr_sectors = req->hard_nr_sectors += nr_sectors;
|
2005-06-27 10:55:12 +02:00
|
|
|
req->ioprio = ioprio_best(req->ioprio, prio);
|
2007-10-23 15:05:46 +02:00
|
|
|
drive_stat_acct(req, 0);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (!attempt_back_merge(q, req))
|
2006-07-13 11:55:04 +02:00
|
|
|
elv_merged_request(q, req, el_ret);
|
2005-04-17 00:20:36 +02:00
|
|
|
goto out;
|
|
|
|
|
|
|
|
case ELEVATOR_FRONT_MERGE:
|
|
|
|
BUG_ON(!rq_mergeable(req));
|
|
|
|
|
2006-12-19 08:33:11 +01:00
|
|
|
if (!ll_front_merge_fn(q, req, bio))
|
2005-04-17 00:20:36 +02:00
|
|
|
break;
|
|
|
|
|
2006-03-23 20:00:26 +01:00
|
|
|
blk_add_trace_bio(q, bio, BLK_TA_FRONTMERGE);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
bio->bi_next = req->bio;
|
|
|
|
req->bio = bio;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* may not be valid. if the low level driver said
|
|
|
|
* it didn't need a bounce buffer then it better
|
|
|
|
* not touch req->buffer either...
|
|
|
|
*/
|
|
|
|
req->buffer = bio_data(bio);
|
2006-07-18 04:14:45 +02:00
|
|
|
req->current_nr_sectors = bio_cur_sectors(bio);
|
|
|
|
req->hard_cur_sectors = req->current_nr_sectors;
|
|
|
|
req->sector = req->hard_sector = bio->bi_sector;
|
2005-04-17 00:20:36 +02:00
|
|
|
req->nr_sectors = req->hard_nr_sectors += nr_sectors;
|
2005-06-27 10:55:12 +02:00
|
|
|
req->ioprio = ioprio_best(req->ioprio, prio);
|
2007-10-23 15:05:46 +02:00
|
|
|
drive_stat_acct(req, 0);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (!attempt_front_merge(q, req))
|
2006-07-13 11:55:04 +02:00
|
|
|
elv_merged_request(q, req, el_ret);
|
2005-04-17 00:20:36 +02:00
|
|
|
goto out;
|
|
|
|
|
2005-06-29 05:45:13 +02:00
|
|
|
/* ELV_NO_MERGE: elevator says don't/can't merge. */
|
2005-04-17 00:20:36 +02:00
|
|
|
default:
|
2005-06-29 05:45:13 +02:00
|
|
|
;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2005-06-29 05:45:13 +02:00
|
|
|
get_rq:
|
2006-12-13 13:02:26 +01:00
|
|
|
/*
|
|
|
|
* This sync check and mask will be re-done in init_request_from_bio(),
|
|
|
|
* but we need to set it earlier to expose the sync flag to the
|
|
|
|
* rq allocator and io schedulers.
|
|
|
|
*/
|
|
|
|
rw_flags = bio_data_dir(bio);
|
|
|
|
if (sync)
|
|
|
|
rw_flags |= REQ_RW_SYNC;
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
2005-06-29 05:45:13 +02:00
|
|
|
* Grab a free request. This is might sleep but can not fail.
|
2005-06-29 05:45:14 +02:00
|
|
|
* Returns with the queue unlocked.
|
2005-06-29 05:45:13 +02:00
|
|
|
*/
|
2006-12-13 13:02:26 +01:00
|
|
|
req = get_request_wait(q, rw_flags, bio);
|
2005-06-29 05:45:14 +02:00
|
|
|
|
2005-06-29 05:45:13 +02:00
|
|
|
/*
|
|
|
|
* After dropping the lock and possibly sleeping here, our request
|
|
|
|
* may now be mergeable after it had proven unmergeable (above).
|
|
|
|
* We don't worry about that case for efficiency. It won't happen
|
|
|
|
* often, and the elevators are able to handle it.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
2006-01-06 09:49:58 +01:00
|
|
|
init_request_from_bio(req, bio);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2005-06-29 05:45:13 +02:00
|
|
|
spin_lock_irq(q->queue_lock);
|
|
|
|
if (elv_queue_empty(q))
|
|
|
|
blk_plug_device(q);
|
2005-04-17 00:20:36 +02:00
|
|
|
add_request(q, req);
|
|
|
|
out:
|
2005-04-17 00:25:40 +02:00
|
|
|
if (sync)
|
2005-04-17 00:20:36 +02:00
|
|
|
__generic_unplug_device(q);
|
|
|
|
|
|
|
|
spin_unlock_irq(q->queue_lock);
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
end_io:
|
2007-09-27 12:47:43 +02:00
|
|
|
bio_endio(bio, err);
|
2005-04-17 00:20:36 +02:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If bio->bi_dev is a partition, remap the location
|
|
|
|
*/
|
|
|
|
static inline void blk_partition_remap(struct bio *bio)
|
|
|
|
{
|
|
|
|
struct block_device *bdev = bio->bi_bdev;
|
|
|
|
|
2007-09-27 13:01:25 +02:00
|
|
|
if (bio_sectors(bio) && bdev != bdev->bd_contains) {
|
2005-04-17 00:20:36 +02:00
|
|
|
struct hd_struct *p = bdev->bd_part;
|
2005-11-01 09:26:16 +01:00
|
|
|
const int rw = bio_data_dir(bio);
|
|
|
|
|
|
|
|
p->sectors[rw] += bio_sectors(bio);
|
|
|
|
p->ios[rw]++;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
bio->bi_sector += p->start_sect;
|
|
|
|
bio->bi_bdev = bdev->bd_contains;
|
2007-08-07 15:30:23 +02:00
|
|
|
|
|
|
|
blk_add_trace_remap(bdev_get_queue(bio->bi_bdev), bio,
|
|
|
|
bdev->bd_dev, bio->bi_sector,
|
|
|
|
bio->bi_sector - p->start_sect);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void handle_bad_sector(struct bio *bio)
|
|
|
|
{
|
|
|
|
char b[BDEVNAME_SIZE];
|
|
|
|
|
|
|
|
printk(KERN_INFO "attempt to access beyond end of device\n");
|
|
|
|
printk(KERN_INFO "%s: rw=%ld, want=%Lu, limit=%Lu\n",
|
|
|
|
bdevname(bio->bi_bdev, b),
|
|
|
|
bio->bi_rw,
|
|
|
|
(unsigned long long)bio->bi_sector + bio_sectors(bio),
|
|
|
|
(long long)(bio->bi_bdev->bd_inode->i_size >> 9));
|
|
|
|
|
|
|
|
set_bit(BIO_EOF, &bio->bi_flags);
|
|
|
|
}
|
|
|
|
|
2006-12-08 11:39:46 +01:00
|
|
|
#ifdef CONFIG_FAIL_MAKE_REQUEST
|
|
|
|
|
|
|
|
static DECLARE_FAULT_ATTR(fail_make_request);
|
|
|
|
|
|
|
|
static int __init setup_fail_make_request(char *str)
|
|
|
|
{
|
|
|
|
return setup_fault_attr(&fail_make_request, str);
|
|
|
|
}
|
|
|
|
__setup("fail_make_request=", setup_fail_make_request);
|
|
|
|
|
|
|
|
static int should_fail_request(struct bio *bio)
|
|
|
|
{
|
|
|
|
if ((bio->bi_bdev->bd_disk->flags & GENHD_FL_FAIL) ||
|
|
|
|
(bio->bi_bdev->bd_part && bio->bi_bdev->bd_part->make_it_fail))
|
|
|
|
return should_fail(&fail_make_request, bio->bi_size);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int __init fail_make_request_debugfs(void)
|
|
|
|
{
|
|
|
|
return init_fault_attr_dentries(&fail_make_request,
|
|
|
|
"fail_make_request");
|
|
|
|
}
|
|
|
|
|
|
|
|
late_initcall(fail_make_request_debugfs);
|
|
|
|
|
|
|
|
#else /* CONFIG_FAIL_MAKE_REQUEST */
|
|
|
|
|
|
|
|
static inline int should_fail_request(struct bio *bio)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
#endif /* CONFIG_FAIL_MAKE_REQUEST */
|
|
|
|
|
2007-07-18 13:27:58 +02:00
|
|
|
/*
|
|
|
|
* Check whether this bio extends beyond the end of the device.
|
|
|
|
*/
|
|
|
|
static inline int bio_check_eod(struct bio *bio, unsigned int nr_sectors)
|
|
|
|
{
|
|
|
|
sector_t maxsector;
|
|
|
|
|
|
|
|
if (!nr_sectors)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/* Test device or partition size, when known. */
|
|
|
|
maxsector = bio->bi_bdev->bd_inode->i_size >> 9;
|
|
|
|
if (maxsector) {
|
|
|
|
sector_t sector = bio->bi_sector;
|
|
|
|
|
|
|
|
if (maxsector < nr_sectors || maxsector - nr_sectors < sector) {
|
|
|
|
/*
|
|
|
|
* This may well happen - the kernel calls bread()
|
|
|
|
* without checking the size of the device, e.g., when
|
|
|
|
* mounting a device.
|
|
|
|
*/
|
|
|
|
handle_bad_sector(bio);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/**
|
|
|
|
* generic_make_request: hand a buffer to its device driver for I/O
|
|
|
|
* @bio: The bio describing the location in memory and on the device.
|
|
|
|
*
|
|
|
|
* generic_make_request() is used to make I/O requests of block
|
|
|
|
* devices. It is passed a &struct bio, which describes the I/O that needs
|
|
|
|
* to be done.
|
|
|
|
*
|
|
|
|
* generic_make_request() does not return any status. The
|
|
|
|
* success/failure status of the request, along with notification of
|
|
|
|
* completion, is delivered asynchronously through the bio->bi_end_io
|
|
|
|
* function described (one day) else where.
|
|
|
|
*
|
|
|
|
* The caller of generic_make_request must make sure that bi_io_vec
|
|
|
|
* are set to describe the memory buffer, and that bi_dev and bi_sector are
|
|
|
|
* set to describe the device address, and the
|
|
|
|
* bi_end_io and optionally bi_private are set to describe how
|
|
|
|
* completion notification should be signaled.
|
|
|
|
*
|
|
|
|
* generic_make_request and the drivers it calls may use bi_next if this
|
|
|
|
* bio happens to be merged with someone else, and may change bi_dev and
|
|
|
|
* bi_sector for remaps as it sees fit. So the values of these fields
|
|
|
|
* should NOT be depended on after the call to generic_make_request.
|
|
|
|
*/
|
When stacked block devices are in-use (e.g. md or dm), the recursive calls
to generic_make_request can use up a lot of space, and we would rather they
didn't.
As generic_make_request is a void function, and as it is generally not
expected that it will have any effect immediately, it is safe to delay any
call to generic_make_request until there is sufficient stack space
available.
As ->bi_next is reserved for the driver to use, it can have no valid value
when generic_make_request is called, and as __make_request implicitly
assumes it will be NULL (ELEVATOR_BACK_MERGE fork of switch) we can be
certain that all callers set it to NULL. We can therefore safely use
bi_next to link pending requests together, providing we clear it before
making the real call.
So, we choose to allow each thread to only be active in one
generic_make_request at a time. If a subsequent (recursive) call is made,
the bio is linked into a per-thread list, and is handled when the active
call completes.
As the list of pending bios is per-thread, there are no locking issues to
worry about.
I say above that it is "safe to delay any call...". There are, however,
some behaviours of a make_request_fn which would make it unsafe. These
include any behaviour that assumes anything will have changed after a
recursive call to generic_make_request.
These could include:
- waiting for that call to finish and call it's bi_end_io function.
md use to sometimes do this (marking the superblock dirty before
completing a write) but doesn't any more
- inspecting the bio for fields that generic_make_request might
change, such as bi_sector or bi_bdev. It is hard to see a good
reason for this, and I don't think anyone actually does it.
- inspecing the queue to see if, e.g. it is 'full' yet. Again, I
think this is very unlikely to be useful, or to be done.
Signed-off-by: Neil Brown <neilb@suse.de>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: <dm-devel@redhat.com>
Alasdair G Kergon <agk@redhat.com> said:
I can see nothing wrong with this in principle.
For device-mapper at the moment though it's essential that, while the bio
mappings may now get delayed, they still get processed in exactly
the same order as they were passed to generic_make_request().
My main concern is whether the timing changes implicit in this patch
will make the rare data-corrupting races in the existing snapshot code
more likely. (I'm working on a fix for these races, but the unfinished
patch is already several hundred lines long.)
It would be helpful if some people on this mailing list would test
this patch in various scenarios and report back.
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-05-01 09:53:42 +02:00
|
|
|
static inline void __generic_make_request(struct bio *bio)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q;
|
2006-10-31 07:07:21 +01:00
|
|
|
sector_t old_sector;
|
2005-04-17 00:20:36 +02:00
|
|
|
int ret, nr_sectors = bio_sectors(bio);
|
2006-03-23 20:00:26 +01:00
|
|
|
dev_t old_dev;
|
2007-11-02 08:49:08 +01:00
|
|
|
int err = -EIO;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
might_sleep();
|
|
|
|
|
2007-07-18 13:27:58 +02:00
|
|
|
if (bio_check_eod(bio, nr_sectors))
|
|
|
|
goto end_io;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Resolve the mapping until finished. (drivers are
|
|
|
|
* still free to implement/resolve their own stacking
|
|
|
|
* by explicitly returning 0)
|
|
|
|
*
|
|
|
|
* NOTE: we don't repeat the blk_size check for each new device.
|
|
|
|
* Stacking drivers are expected to know what they are doing.
|
|
|
|
*/
|
2006-10-31 07:07:21 +01:00
|
|
|
old_sector = -1;
|
2006-03-23 20:00:26 +01:00
|
|
|
old_dev = 0;
|
2005-04-17 00:20:36 +02:00
|
|
|
do {
|
|
|
|
char b[BDEVNAME_SIZE];
|
|
|
|
|
|
|
|
q = bdev_get_queue(bio->bi_bdev);
|
|
|
|
if (!q) {
|
|
|
|
printk(KERN_ERR
|
|
|
|
"generic_make_request: Trying to access "
|
|
|
|
"nonexistent block-device %s (%Lu)\n",
|
|
|
|
bdevname(bio->bi_bdev, b),
|
|
|
|
(long long) bio->bi_sector);
|
|
|
|
end_io:
|
2007-11-02 08:49:08 +01:00
|
|
|
bio_endio(bio, err);
|
2005-04-17 00:20:36 +02:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2007-07-18 13:13:10 +02:00
|
|
|
if (unlikely(nr_sectors > q->max_hw_sectors)) {
|
2005-04-17 00:20:36 +02:00
|
|
|
printk("bio too big device %s (%u > %u)\n",
|
|
|
|
bdevname(bio->bi_bdev, b),
|
|
|
|
bio_sectors(bio),
|
|
|
|
q->max_hw_sectors);
|
|
|
|
goto end_io;
|
|
|
|
}
|
|
|
|
|
2005-06-23 09:08:53 +02:00
|
|
|
if (unlikely(test_bit(QUEUE_FLAG_DEAD, &q->queue_flags)))
|
2005-04-17 00:20:36 +02:00
|
|
|
goto end_io;
|
|
|
|
|
2006-12-08 11:39:46 +01:00
|
|
|
if (should_fail_request(bio))
|
|
|
|
goto end_io;
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* If this device has partitions, remap block n
|
|
|
|
* of partition p to block n+start(p) of the disk.
|
|
|
|
*/
|
|
|
|
blk_partition_remap(bio);
|
|
|
|
|
2006-10-31 07:07:21 +01:00
|
|
|
if (old_sector != -1)
|
2007-07-18 13:13:10 +02:00
|
|
|
blk_add_trace_remap(q, bio, old_dev, bio->bi_sector,
|
2006-10-31 07:07:21 +01:00
|
|
|
old_sector);
|
2006-03-23 20:00:26 +01:00
|
|
|
|
|
|
|
blk_add_trace_bio(q, bio, BLK_TA_QUEUE);
|
|
|
|
|
2006-10-31 07:07:21 +01:00
|
|
|
old_sector = bio->bi_sector;
|
2006-03-23 20:00:26 +01:00
|
|
|
old_dev = bio->bi_bdev->bd_dev;
|
|
|
|
|
2007-07-18 13:27:58 +02:00
|
|
|
if (bio_check_eod(bio, nr_sectors))
|
|
|
|
goto end_io;
|
2007-11-02 08:49:08 +01:00
|
|
|
if (bio_empty_barrier(bio) && !q->prepare_flush_fn) {
|
|
|
|
err = -EOPNOTSUPP;
|
|
|
|
goto end_io;
|
|
|
|
}
|
2006-10-31 07:07:21 +01:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
ret = q->make_request_fn(q, bio);
|
|
|
|
} while (ret);
|
|
|
|
}
|
|
|
|
|
When stacked block devices are in-use (e.g. md or dm), the recursive calls
to generic_make_request can use up a lot of space, and we would rather they
didn't.
As generic_make_request is a void function, and as it is generally not
expected that it will have any effect immediately, it is safe to delay any
call to generic_make_request until there is sufficient stack space
available.
As ->bi_next is reserved for the driver to use, it can have no valid value
when generic_make_request is called, and as __make_request implicitly
assumes it will be NULL (ELEVATOR_BACK_MERGE fork of switch) we can be
certain that all callers set it to NULL. We can therefore safely use
bi_next to link pending requests together, providing we clear it before
making the real call.
So, we choose to allow each thread to only be active in one
generic_make_request at a time. If a subsequent (recursive) call is made,
the bio is linked into a per-thread list, and is handled when the active
call completes.
As the list of pending bios is per-thread, there are no locking issues to
worry about.
I say above that it is "safe to delay any call...". There are, however,
some behaviours of a make_request_fn which would make it unsafe. These
include any behaviour that assumes anything will have changed after a
recursive call to generic_make_request.
These could include:
- waiting for that call to finish and call it's bi_end_io function.
md use to sometimes do this (marking the superblock dirty before
completing a write) but doesn't any more
- inspecting the bio for fields that generic_make_request might
change, such as bi_sector or bi_bdev. It is hard to see a good
reason for this, and I don't think anyone actually does it.
- inspecing the queue to see if, e.g. it is 'full' yet. Again, I
think this is very unlikely to be useful, or to be done.
Signed-off-by: Neil Brown <neilb@suse.de>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: <dm-devel@redhat.com>
Alasdair G Kergon <agk@redhat.com> said:
I can see nothing wrong with this in principle.
For device-mapper at the moment though it's essential that, while the bio
mappings may now get delayed, they still get processed in exactly
the same order as they were passed to generic_make_request().
My main concern is whether the timing changes implicit in this patch
will make the rare data-corrupting races in the existing snapshot code
more likely. (I'm working on a fix for these races, but the unfinished
patch is already several hundred lines long.)
It would be helpful if some people on this mailing list would test
this patch in various scenarios and report back.
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-05-01 09:53:42 +02:00
|
|
|
/*
|
|
|
|
* We only want one ->make_request_fn to be active at a time,
|
|
|
|
* else stack usage with stacked devices could be a problem.
|
|
|
|
* So use current->bio_{list,tail} to keep a list of requests
|
|
|
|
* submited by a make_request_fn function.
|
|
|
|
* current->bio_tail is also used as a flag to say if
|
|
|
|
* generic_make_request is currently active in this task or not.
|
|
|
|
* If it is NULL, then no make_request is active. If it is non-NULL,
|
|
|
|
* then a make_request is active, and new requests should be added
|
|
|
|
* at the tail
|
|
|
|
*/
|
|
|
|
void generic_make_request(struct bio *bio)
|
|
|
|
{
|
|
|
|
if (current->bio_tail) {
|
|
|
|
/* make_request is active */
|
|
|
|
*(current->bio_tail) = bio;
|
|
|
|
bio->bi_next = NULL;
|
|
|
|
current->bio_tail = &bio->bi_next;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
/* following loop may be a bit non-obvious, and so deserves some
|
|
|
|
* explanation.
|
|
|
|
* Before entering the loop, bio->bi_next is NULL (as all callers
|
|
|
|
* ensure that) so we have a list with a single bio.
|
|
|
|
* We pretend that we have just taken it off a longer list, so
|
|
|
|
* we assign bio_list to the next (which is NULL) and bio_tail
|
|
|
|
* to &bio_list, thus initialising the bio_list of new bios to be
|
|
|
|
* added. __generic_make_request may indeed add some more bios
|
|
|
|
* through a recursive call to generic_make_request. If it
|
|
|
|
* did, we find a non-NULL value in bio_list and re-enter the loop
|
|
|
|
* from the top. In this case we really did just take the bio
|
|
|
|
* of the top of the list (no pretending) and so fixup bio_list and
|
|
|
|
* bio_tail or bi_next, and call into __generic_make_request again.
|
|
|
|
*
|
|
|
|
* The loop was structured like this to make only one call to
|
|
|
|
* __generic_make_request (which is important as it is large and
|
|
|
|
* inlined) and to keep the structure simple.
|
|
|
|
*/
|
|
|
|
BUG_ON(bio->bi_next);
|
|
|
|
do {
|
|
|
|
current->bio_list = bio->bi_next;
|
|
|
|
if (bio->bi_next == NULL)
|
|
|
|
current->bio_tail = ¤t->bio_list;
|
|
|
|
else
|
|
|
|
bio->bi_next = NULL;
|
|
|
|
__generic_make_request(bio);
|
|
|
|
bio = current->bio_list;
|
|
|
|
} while (bio);
|
|
|
|
current->bio_tail = NULL; /* deactivate */
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
EXPORT_SYMBOL(generic_make_request);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* submit_bio: submit a bio to the block device layer for I/O
|
|
|
|
* @rw: whether to %READ or %WRITE, or maybe to %READA (read ahead)
|
|
|
|
* @bio: The &struct bio which describes the I/O
|
|
|
|
*
|
|
|
|
* submit_bio() is very similar in purpose to generic_make_request(), and
|
|
|
|
* uses that function to do most of the work. Both are fairly rough
|
|
|
|
* interfaces, @bio must be presetup and ready for I/O.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
void submit_bio(int rw, struct bio *bio)
|
|
|
|
{
|
|
|
|
int count = bio_sectors(bio);
|
|
|
|
|
2005-06-27 10:55:12 +02:00
|
|
|
bio->bi_rw |= rw;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-09-27 13:01:25 +02:00
|
|
|
/*
|
|
|
|
* If it's a regular read/write or a barrier with data attached,
|
|
|
|
* go through the normal accounting stuff before submission.
|
|
|
|
*/
|
|
|
|
if (!bio_empty_barrier(bio)) {
|
|
|
|
|
|
|
|
BIO_BUG_ON(!bio->bi_size);
|
|
|
|
BIO_BUG_ON(!bio->bi_io_vec);
|
|
|
|
|
|
|
|
if (rw & WRITE) {
|
|
|
|
count_vm_events(PGPGOUT, count);
|
|
|
|
} else {
|
|
|
|
task_io_account_read(bio->bi_size);
|
|
|
|
count_vm_events(PGPGIN, count);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (unlikely(block_dump)) {
|
|
|
|
char b[BDEVNAME_SIZE];
|
|
|
|
printk(KERN_DEBUG "%s(%d): %s block %Lu on %s\n",
|
2007-10-19 08:40:40 +02:00
|
|
|
current->comm, task_pid_nr(current),
|
2007-09-27 13:01:25 +02:00
|
|
|
(rw & WRITE) ? "WRITE" : "READ",
|
|
|
|
(unsigned long long)bio->bi_sector,
|
|
|
|
bdevname(bio->bi_bdev,b));
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
generic_make_request(bio);
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(submit_bio);
|
|
|
|
|
2005-06-25 23:59:10 +02:00
|
|
|
static void blk_recalc_rq_sectors(struct request *rq, int nsect)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
if (blk_fs_request(rq)) {
|
|
|
|
rq->hard_sector += nsect;
|
|
|
|
rq->hard_nr_sectors -= nsect;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Move the I/O submission pointers ahead if required.
|
|
|
|
*/
|
|
|
|
if ((rq->nr_sectors >= rq->hard_nr_sectors) &&
|
|
|
|
(rq->sector <= rq->hard_sector)) {
|
|
|
|
rq->sector = rq->hard_sector;
|
|
|
|
rq->nr_sectors = rq->hard_nr_sectors;
|
|
|
|
rq->hard_cur_sectors = bio_cur_sectors(rq->bio);
|
|
|
|
rq->current_nr_sectors = rq->hard_cur_sectors;
|
|
|
|
rq->buffer = bio_data(rq->bio);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* if total number of sectors is less than the first segment
|
|
|
|
* size, something has gone terribly wrong
|
|
|
|
*/
|
|
|
|
if (rq->nr_sectors < rq->current_nr_sectors) {
|
|
|
|
printk("blk: request botched\n");
|
|
|
|
rq->nr_sectors = rq->current_nr_sectors;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int __end_that_request_first(struct request *req, int uptodate,
|
|
|
|
int nr_bytes)
|
|
|
|
{
|
|
|
|
int total_bytes, bio_nbytes, error, next_idx = 0;
|
|
|
|
struct bio *bio;
|
|
|
|
|
2006-03-23 20:00:26 +01:00
|
|
|
blk_add_trace_rq(req->q, req, BLK_TA_COMPLETE);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* extend uptodate bool to allow < 0 value to be direct io error
|
|
|
|
*/
|
|
|
|
error = 0;
|
|
|
|
if (end_io_error(uptodate))
|
|
|
|
error = !uptodate ? -EIO : uptodate;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* for a REQ_BLOCK_PC request, we want to carry any eventual
|
|
|
|
* sense key with us all the way through
|
|
|
|
*/
|
|
|
|
if (!blk_pc_request(req))
|
|
|
|
req->errors = 0;
|
|
|
|
|
|
|
|
if (!uptodate) {
|
2006-08-10 08:44:47 +02:00
|
|
|
if (blk_fs_request(req) && !(req->cmd_flags & REQ_QUIET))
|
2005-04-17 00:20:36 +02:00
|
|
|
printk("end_request: I/O error, dev %s, sector %llu\n",
|
|
|
|
req->rq_disk ? req->rq_disk->disk_name : "?",
|
|
|
|
(unsigned long long)req->sector);
|
|
|
|
}
|
|
|
|
|
2005-11-01 08:35:42 +01:00
|
|
|
if (blk_fs_request(req) && req->rq_disk) {
|
2005-11-01 09:26:16 +01:00
|
|
|
const int rw = rq_data_dir(req);
|
|
|
|
|
2006-01-17 11:09:27 +01:00
|
|
|
disk_stat_add(req->rq_disk, sectors[rw], nr_bytes >> 9);
|
2005-11-01 08:35:42 +01:00
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
total_bytes = bio_nbytes = 0;
|
|
|
|
while ((bio = req->bio) != NULL) {
|
|
|
|
int nbytes;
|
|
|
|
|
2007-09-27 13:01:25 +02:00
|
|
|
/*
|
|
|
|
* For an empty barrier request, the low level driver must
|
|
|
|
* store a potential error location in ->sector. We pass
|
|
|
|
* that back up in ->bi_sector.
|
|
|
|
*/
|
|
|
|
if (blk_empty_barrier(req))
|
|
|
|
bio->bi_sector = req->sector;
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
if (nr_bytes >= bio->bi_size) {
|
|
|
|
req->bio = bio->bi_next;
|
|
|
|
nbytes = bio->bi_size;
|
2007-09-27 12:46:13 +02:00
|
|
|
req_bio_endio(req, bio, nbytes, error);
|
2005-04-17 00:20:36 +02:00
|
|
|
next_idx = 0;
|
|
|
|
bio_nbytes = 0;
|
|
|
|
} else {
|
|
|
|
int idx = bio->bi_idx + next_idx;
|
|
|
|
|
|
|
|
if (unlikely(bio->bi_idx >= bio->bi_vcnt)) {
|
|
|
|
blk_dump_rq_flags(req, "__end_that");
|
|
|
|
printk("%s: bio idx %d >= vcnt %d\n",
|
|
|
|
__FUNCTION__,
|
|
|
|
bio->bi_idx, bio->bi_vcnt);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
nbytes = bio_iovec_idx(bio, idx)->bv_len;
|
|
|
|
BIO_BUG_ON(nbytes > bio->bi_size);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* not a complete bvec done
|
|
|
|
*/
|
|
|
|
if (unlikely(nbytes > nr_bytes)) {
|
|
|
|
bio_nbytes += nr_bytes;
|
|
|
|
total_bytes += nr_bytes;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* advance to the next vector
|
|
|
|
*/
|
|
|
|
next_idx++;
|
|
|
|
bio_nbytes += nbytes;
|
|
|
|
}
|
|
|
|
|
|
|
|
total_bytes += nbytes;
|
|
|
|
nr_bytes -= nbytes;
|
|
|
|
|
|
|
|
if ((bio = req->bio)) {
|
|
|
|
/*
|
|
|
|
* end more in this run, or just return 'not-done'
|
|
|
|
*/
|
|
|
|
if (unlikely(nr_bytes <= 0))
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* completely done
|
|
|
|
*/
|
|
|
|
if (!req->bio)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* if the request wasn't completed, update state
|
|
|
|
*/
|
|
|
|
if (bio_nbytes) {
|
2007-09-27 12:46:13 +02:00
|
|
|
req_bio_endio(req, bio, bio_nbytes, error);
|
2005-04-17 00:20:36 +02:00
|
|
|
bio->bi_idx += next_idx;
|
|
|
|
bio_iovec(bio)->bv_offset += nr_bytes;
|
|
|
|
bio_iovec(bio)->bv_len -= nr_bytes;
|
|
|
|
}
|
|
|
|
|
|
|
|
blk_recalc_rq_sectors(req, total_bytes >> 9);
|
|
|
|
blk_recalc_rq_segments(req);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* end_that_request_first - end I/O on a request
|
|
|
|
* @req: the request being processed
|
|
|
|
* @uptodate: 1 for success, 0 for I/O error, < 0 for specific error
|
|
|
|
* @nr_sectors: number of sectors to end I/O on
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Ends I/O on a number of sectors attached to @req, and sets it up
|
|
|
|
* for the next range of segments (if any) in the cluster.
|
|
|
|
*
|
|
|
|
* Return:
|
|
|
|
* 0 - we are done with this request, call end_that_request_last()
|
|
|
|
* 1 - still buffers pending for this request
|
|
|
|
**/
|
|
|
|
int end_that_request_first(struct request *req, int uptodate, int nr_sectors)
|
|
|
|
{
|
|
|
|
return __end_that_request_first(req, uptodate, nr_sectors << 9);
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(end_that_request_first);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* end_that_request_chunk - end I/O on a request
|
|
|
|
* @req: the request being processed
|
|
|
|
* @uptodate: 1 for success, 0 for I/O error, < 0 for specific error
|
|
|
|
* @nr_bytes: number of bytes to complete
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Ends I/O on a number of bytes attached to @req, and sets it up
|
|
|
|
* for the next range of segments (if any). Like end_that_request_first(),
|
|
|
|
* but deals with bytes instead of sectors.
|
|
|
|
*
|
|
|
|
* Return:
|
|
|
|
* 0 - we are done with this request, call end_that_request_last()
|
|
|
|
* 1 - still buffers pending for this request
|
|
|
|
**/
|
|
|
|
int end_that_request_chunk(struct request *req, int uptodate, int nr_bytes)
|
|
|
|
{
|
|
|
|
return __end_that_request_first(req, uptodate, nr_bytes);
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(end_that_request_chunk);
|
|
|
|
|
2006-01-09 16:02:34 +01:00
|
|
|
/*
|
|
|
|
* splice the completion data to a local structure and hand off to
|
|
|
|
* process_completion_queue() to complete the requests
|
|
|
|
*/
|
|
|
|
static void blk_done_softirq(struct softirq_action *h)
|
|
|
|
{
|
2006-06-23 11:05:55 +02:00
|
|
|
struct list_head *cpu_list, local_list;
|
2006-01-09 16:02:34 +01:00
|
|
|
|
|
|
|
local_irq_disable();
|
|
|
|
cpu_list = &__get_cpu_var(blk_cpu_done);
|
2006-06-23 11:05:55 +02:00
|
|
|
list_replace_init(cpu_list, &local_list);
|
2006-01-09 16:02:34 +01:00
|
|
|
local_irq_enable();
|
|
|
|
|
|
|
|
while (!list_empty(&local_list)) {
|
|
|
|
struct request *rq = list_entry(local_list.next, struct request, donelist);
|
|
|
|
|
|
|
|
list_del_init(&rq->donelist);
|
|
|
|
rq->q->softirq_done_fn(rq);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-08-23 09:29:40 +02:00
|
|
|
static int __cpuinit blk_cpu_notify(struct notifier_block *self, unsigned long action,
|
2006-01-09 16:02:34 +01:00
|
|
|
void *hcpu)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* If a CPU goes away, splice its entries to the current CPU
|
|
|
|
* and trigger a run of the softirq
|
|
|
|
*/
|
2007-05-09 11:35:10 +02:00
|
|
|
if (action == CPU_DEAD || action == CPU_DEAD_FROZEN) {
|
2006-01-09 16:02:34 +01:00
|
|
|
int cpu = (unsigned long) hcpu;
|
|
|
|
|
|
|
|
local_irq_disable();
|
|
|
|
list_splice_init(&per_cpu(blk_cpu_done, cpu),
|
|
|
|
&__get_cpu_var(blk_cpu_done));
|
|
|
|
raise_softirq_irqoff(BLOCK_SOFTIRQ);
|
|
|
|
local_irq_enable();
|
|
|
|
}
|
|
|
|
|
|
|
|
return NOTIFY_OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2007-08-23 09:29:40 +02:00
|
|
|
static struct notifier_block blk_cpu_notifier __cpuinitdata = {
|
2006-01-09 16:02:34 +01:00
|
|
|
.notifier_call = blk_cpu_notify,
|
|
|
|
};
|
|
|
|
|
|
|
|
/**
|
|
|
|
* blk_complete_request - end I/O on a request
|
|
|
|
* @req: the request being processed
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Ends all I/O on a request. It does not handle partial completions,
|
2006-06-26 18:35:02 +02:00
|
|
|
* unless the driver actually implements this in its completion callback
|
2007-07-18 13:13:10 +02:00
|
|
|
* through requeueing. The actual completion happens out-of-order,
|
2006-01-09 16:02:34 +01:00
|
|
|
* through a softirq handler. The user must have registered a completion
|
|
|
|
* callback through blk_queue_softirq_done().
|
|
|
|
**/
|
|
|
|
|
|
|
|
void blk_complete_request(struct request *req)
|
|
|
|
{
|
|
|
|
struct list_head *cpu_list;
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
BUG_ON(!req->q->softirq_done_fn);
|
|
|
|
|
|
|
|
local_irq_save(flags);
|
|
|
|
|
|
|
|
cpu_list = &__get_cpu_var(blk_cpu_done);
|
|
|
|
list_add_tail(&req->donelist, cpu_list);
|
|
|
|
raise_softirq_irqoff(BLOCK_SOFTIRQ);
|
|
|
|
|
|
|
|
local_irq_restore(flags);
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(blk_complete_request);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* queue lock must be held
|
|
|
|
*/
|
2006-01-06 09:49:03 +01:00
|
|
|
void end_that_request_last(struct request *req, int uptodate)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct gendisk *disk = req->rq_disk;
|
2006-01-06 09:49:03 +01:00
|
|
|
int error;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* extend uptodate bool to allow < 0 value to be direct io error
|
|
|
|
*/
|
|
|
|
error = 0;
|
|
|
|
if (end_io_error(uptodate))
|
|
|
|
error = !uptodate ? -EIO : uptodate;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (unlikely(laptop_mode) && blk_fs_request(req))
|
|
|
|
laptop_io_completion();
|
|
|
|
|
2006-05-23 11:23:49 +02:00
|
|
|
/*
|
|
|
|
* Account IO completion. bar_rq isn't accounted as a normal
|
|
|
|
* IO on queueing nor completion. Accounting the containing
|
|
|
|
* request is enough.
|
|
|
|
*/
|
|
|
|
if (disk && blk_fs_request(req) && req != &req->q->bar_rq) {
|
2005-04-17 00:20:36 +02:00
|
|
|
unsigned long duration = jiffies - req->start_time;
|
2005-11-01 09:26:16 +01:00
|
|
|
const int rw = rq_data_dir(req);
|
|
|
|
|
|
|
|
__disk_stat_inc(disk, ios[rw]);
|
|
|
|
__disk_stat_add(disk, ticks[rw], duration);
|
2005-04-17 00:20:36 +02:00
|
|
|
disk_round_stats(disk);
|
|
|
|
disk->in_flight--;
|
|
|
|
}
|
|
|
|
if (req->end_io)
|
2006-01-06 09:49:03 +01:00
|
|
|
req->end_io(req, error);
|
2005-04-17 00:20:36 +02:00
|
|
|
else
|
|
|
|
__blk_put_request(req->q, req);
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(end_that_request_last);
|
|
|
|
|
2007-09-21 10:41:07 +02:00
|
|
|
static inline void __end_request(struct request *rq, int uptodate,
|
|
|
|
unsigned int nr_bytes, int dequeue)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2007-09-21 10:41:07 +02:00
|
|
|
if (!end_that_request_chunk(rq, uptodate, nr_bytes)) {
|
|
|
|
if (dequeue)
|
|
|
|
blkdev_dequeue_request(rq);
|
|
|
|
add_disk_randomness(rq->rq_disk);
|
|
|
|
end_that_request_last(rq, uptodate);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-09-21 10:41:07 +02:00
|
|
|
static unsigned int rq_byte_size(struct request *rq)
|
|
|
|
{
|
|
|
|
if (blk_fs_request(rq))
|
|
|
|
return rq->hard_nr_sectors << 9;
|
|
|
|
|
|
|
|
return rq->data_len;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* end_queued_request - end all I/O on a queued request
|
|
|
|
* @rq: the request being processed
|
|
|
|
* @uptodate: error value or 0/1 uptodate flag
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Ends all I/O on a request, and removes it from the block layer queues.
|
|
|
|
* Not suitable for normal IO completion, unless the driver still has
|
|
|
|
* the request attached to the block layer.
|
|
|
|
*
|
|
|
|
**/
|
|
|
|
void end_queued_request(struct request *rq, int uptodate)
|
|
|
|
{
|
|
|
|
__end_request(rq, uptodate, rq_byte_size(rq), 1);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(end_queued_request);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* end_dequeued_request - end all I/O on a dequeued request
|
|
|
|
* @rq: the request being processed
|
|
|
|
* @uptodate: error value or 0/1 uptodate flag
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Ends all I/O on a request. The request must already have been
|
|
|
|
* dequeued using blkdev_dequeue_request(), as is normally the case
|
|
|
|
* for most drivers.
|
|
|
|
*
|
|
|
|
**/
|
|
|
|
void end_dequeued_request(struct request *rq, int uptodate)
|
|
|
|
{
|
|
|
|
__end_request(rq, uptodate, rq_byte_size(rq), 0);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(end_dequeued_request);
|
|
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
* end_request - end I/O on the current segment of the request
|
2007-10-19 08:39:28 +02:00
|
|
|
* @req: the request being processed
|
2007-09-21 10:41:07 +02:00
|
|
|
* @uptodate: error value or 0/1 uptodate flag
|
|
|
|
*
|
|
|
|
* Description:
|
|
|
|
* Ends I/O on the current segment of a request. If that is the only
|
|
|
|
* remaining segment, the request is also completed and freed.
|
|
|
|
*
|
|
|
|
* This is a remnant of how older block drivers handled IO completions.
|
|
|
|
* Modern drivers typically end IO on the full request in one go, unless
|
|
|
|
* they have a residual value to account for. For that case this function
|
|
|
|
* isn't really useful, unless the residual just happens to be the
|
|
|
|
* full current segment. In other words, don't use this function in new
|
|
|
|
* code. Either use end_request_completely(), or the
|
|
|
|
* end_that_request_chunk() (along with end_that_request_last()) for
|
|
|
|
* partial completions.
|
|
|
|
*
|
|
|
|
**/
|
|
|
|
void end_request(struct request *req, int uptodate)
|
|
|
|
{
|
|
|
|
__end_request(req, uptodate, req->hard_cur_sectors << 9, 1);
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
EXPORT_SYMBOL(end_request);
|
|
|
|
|
2007-08-16 13:31:28 +02:00
|
|
|
static void blk_rq_bio_prep(struct request_queue *q, struct request *rq,
|
|
|
|
struct bio *bio)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-08-10 08:44:47 +02:00
|
|
|
/* first two bits are identical in rq->cmd_flags and bio->bi_rw */
|
|
|
|
rq->cmd_flags |= (bio->bi_rw & 3);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
rq->nr_phys_segments = bio_phys_segments(q, bio);
|
|
|
|
rq->nr_hw_segments = bio_hw_segments(q, bio);
|
|
|
|
rq->current_nr_sectors = bio_cur_sectors(bio);
|
|
|
|
rq->hard_cur_sectors = rq->current_nr_sectors;
|
|
|
|
rq->hard_nr_sectors = rq->nr_sectors = bio_sectors(bio);
|
|
|
|
rq->buffer = bio_data(bio);
|
2006-12-01 10:40:55 +01:00
|
|
|
rq->data_len = bio->bi_size;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
rq->bio = rq->biotail = bio;
|
|
|
|
|
2007-08-16 13:31:28 +02:00
|
|
|
if (bio->bi_bdev)
|
|
|
|
rq->rq_disk = bio->bi_bdev->bd_disk;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int kblockd_schedule_work(struct work_struct *work)
|
|
|
|
{
|
|
|
|
return queue_work(kblockd_workqueue, work);
|
|
|
|
}
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(kblockd_schedule_work);
|
|
|
|
|
2007-05-09 11:33:56 +02:00
|
|
|
void kblockd_flush_work(struct work_struct *work)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2007-05-09 11:34:22 +02:00
|
|
|
cancel_work_sync(work);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2007-05-09 11:33:56 +02:00
|
|
|
EXPORT_SYMBOL(kblockd_flush_work);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int __init blk_dev_init(void)
|
|
|
|
{
|
2006-01-09 16:02:34 +01:00
|
|
|
int i;
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
kblockd_workqueue = create_workqueue("kblockd");
|
|
|
|
if (!kblockd_workqueue)
|
|
|
|
panic("Failed to create kblockd\n");
|
|
|
|
|
|
|
|
request_cachep = kmem_cache_create("blkdev_requests",
|
2007-07-20 03:11:58 +02:00
|
|
|
sizeof(struct request), 0, SLAB_PANIC, NULL);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
requestq_cachep = kmem_cache_create("blkdev_queue",
|
2007-07-24 09:28:11 +02:00
|
|
|
sizeof(struct request_queue), 0, SLAB_PANIC, NULL);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
iocontext_cachep = kmem_cache_create("blkdev_ioc",
|
2007-07-20 03:11:58 +02:00
|
|
|
sizeof(struct io_context), 0, SLAB_PANIC, NULL);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-03-28 11:56:37 +02:00
|
|
|
for_each_possible_cpu(i)
|
2006-01-09 16:02:34 +01:00
|
|
|
INIT_LIST_HEAD(&per_cpu(blk_cpu_done, i));
|
|
|
|
|
|
|
|
open_softirq(BLOCK_SOFTIRQ, blk_done_softirq, NULL);
|
2006-06-27 11:54:11 +02:00
|
|
|
register_hotcpu_notifier(&blk_cpu_notifier);
|
2006-01-09 16:02:34 +01:00
|
|
|
|
block: blk_max_pfn is somtimes wrong
There is a small problem in handling page bounce.
At the moment blk_max_pfn equals max_pfn, which is in fact not maximum
possible _number_ of a page frame, but the _amount_ of page frames. For
example for the 32bit x86 node with 4Gb RAM, max_pfn = 0x100000, but not
0xFFFF.
request_queue structure has a member q->bounce_pfn and queue needs bounce
pages for the pages _above_ this limit. This routine is handled by
blk_queue_bounce(), where the following check is produced:
if (q->bounce_pfn >= blk_max_pfn)
return;
Assume, that a driver has set q->bounce_pfn to 0xFFFF, but blk_max_pfn
equals 0x10000. In such situation the check above fails and for each bio
we always fall down for iterating over pages tied to the bio.
I want to notice, that for quite a big range of device drivers (ide, md,
...) such problem doesn't happen because they use BLK_BOUNCE_ANY for
bounce_pfn. BLK_BOUNCE_ANY is defined as blk_max_pfn << PAGE_SHIFT, and
then the check above doesn't fail. But for other drivers, which obtain
reuired value from drivers, it fails. For example sata_nv uses
ATA_DMA_MASK or dev->dma_mask.
I propose to use (max_pfn - 1) for blk_max_pfn. And the same for
blk_max_low_pfn. The patch also cleanses some checks related with
bounce_pfn.
Signed-off-by: Vasily Tarasov <vtaras@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2007-03-27 08:52:47 +02:00
|
|
|
blk_max_low_pfn = max_low_pfn - 1;
|
|
|
|
blk_max_pfn = max_pfn - 1;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* IO Context helper functions
|
|
|
|
*/
|
|
|
|
void put_io_context(struct io_context *ioc)
|
|
|
|
{
|
|
|
|
if (ioc == NULL)
|
|
|
|
return;
|
|
|
|
|
|
|
|
BUG_ON(atomic_read(&ioc->refcount) == 0);
|
|
|
|
|
|
|
|
if (atomic_dec_and_test(&ioc->refcount)) {
|
2006-03-28 08:59:01 +02:00
|
|
|
struct cfq_io_context *cic;
|
|
|
|
|
2006-03-18 21:05:53 +01:00
|
|
|
rcu_read_lock();
|
2005-04-17 00:20:36 +02:00
|
|
|
if (ioc->aic && ioc->aic->dtor)
|
|
|
|
ioc->aic->dtor(ioc->aic);
|
2006-03-28 08:59:01 +02:00
|
|
|
if (ioc->cic_root.rb_node != NULL) {
|
2006-03-28 09:00:28 +02:00
|
|
|
struct rb_node *n = rb_first(&ioc->cic_root);
|
|
|
|
|
|
|
|
cic = rb_entry(n, struct cfq_io_context, rb_node);
|
2006-03-28 08:59:01 +02:00
|
|
|
cic->dtor(ioc);
|
|
|
|
}
|
2006-03-18 21:05:53 +01:00
|
|
|
rcu_read_unlock();
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
kmem_cache_free(iocontext_cachep, ioc);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(put_io_context);
|
|
|
|
|
|
|
|
/* Called by the exitting task */
|
|
|
|
void exit_io_context(void)
|
|
|
|
{
|
|
|
|
struct io_context *ioc;
|
2006-03-28 08:59:01 +02:00
|
|
|
struct cfq_io_context *cic;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2005-06-27 10:55:12 +02:00
|
|
|
task_lock(current);
|
2005-04-17 00:20:36 +02:00
|
|
|
ioc = current->io_context;
|
|
|
|
current->io_context = NULL;
|
2005-06-27 10:55:12 +02:00
|
|
|
task_unlock(current);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-08-29 09:15:14 +02:00
|
|
|
ioc->task = NULL;
|
2005-04-17 00:20:36 +02:00
|
|
|
if (ioc->aic && ioc->aic->exit)
|
|
|
|
ioc->aic->exit(ioc->aic);
|
2006-03-28 08:59:01 +02:00
|
|
|
if (ioc->cic_root.rb_node != NULL) {
|
|
|
|
cic = rb_entry(rb_first(&ioc->cic_root), struct cfq_io_context, rb_node);
|
|
|
|
cic->exit(ioc);
|
|
|
|
}
|
2006-08-29 09:15:14 +02:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
put_io_context(ioc);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If the current task has no IO context then create one and initialise it.
|
2005-06-29 05:45:15 +02:00
|
|
|
* Otherwise, return its existing IO context.
|
2005-04-17 00:20:36 +02:00
|
|
|
*
|
2005-06-29 05:45:15 +02:00
|
|
|
* This returned IO context doesn't have a specifically elevated refcount,
|
|
|
|
* but since the current task itself holds a reference, the context can be
|
|
|
|
* used in general code, so long as it stays within `current` context.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
2006-07-19 23:39:40 +02:00
|
|
|
static struct io_context *current_io_context(gfp_t gfp_flags, int node)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct task_struct *tsk = current;
|
|
|
|
struct io_context *ret;
|
|
|
|
|
|
|
|
ret = tsk->io_context;
|
2005-06-29 05:45:15 +02:00
|
|
|
if (likely(ret))
|
|
|
|
return ret;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-07-19 23:39:40 +02:00
|
|
|
ret = kmem_cache_alloc_node(iocontext_cachep, gfp_flags, node);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (ret) {
|
|
|
|
atomic_set(&ret->refcount, 1);
|
2005-06-27 10:55:12 +02:00
|
|
|
ret->task = current;
|
2006-08-29 09:05:44 +02:00
|
|
|
ret->ioprio_changed = 0;
|
2005-04-17 00:20:36 +02:00
|
|
|
ret->last_waited = jiffies; /* doesn't matter... */
|
|
|
|
ret->nr_batch_requests = 0; /* because this is 0 */
|
|
|
|
ret->aic = NULL;
|
2006-03-28 08:59:01 +02:00
|
|
|
ret->cic_root.rb_node = NULL;
|
2007-04-24 21:17:33 +02:00
|
|
|
ret->ioc_data = NULL;
|
2006-08-21 08:34:15 +02:00
|
|
|
/* make sure set_task_ioprio() sees the settings above */
|
|
|
|
smp_wmb();
|
2005-06-29 05:45:15 +02:00
|
|
|
tsk->io_context = ret;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2005-06-29 05:45:15 +02:00
|
|
|
return ret;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2005-06-29 05:45:15 +02:00
|
|
|
/*
|
|
|
|
* If the current task has no IO context then create one and initialise it.
|
|
|
|
* If it does have a context, take a ref on it.
|
|
|
|
*
|
|
|
|
* This is always called in the context of the task which submitted the I/O.
|
|
|
|
*/
|
2006-07-19 23:39:40 +02:00
|
|
|
struct io_context *get_io_context(gfp_t gfp_flags, int node)
|
2005-06-29 05:45:15 +02:00
|
|
|
{
|
|
|
|
struct io_context *ret;
|
2006-07-19 23:39:40 +02:00
|
|
|
ret = current_io_context(gfp_flags, node);
|
2005-06-29 05:45:15 +02:00
|
|
|
if (likely(ret))
|
2005-04-17 00:20:36 +02:00
|
|
|
atomic_inc(&ret->refcount);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(get_io_context);
|
|
|
|
|
|
|
|
void copy_io_context(struct io_context **pdst, struct io_context **psrc)
|
|
|
|
{
|
|
|
|
struct io_context *src = *psrc;
|
|
|
|
struct io_context *dst = *pdst;
|
|
|
|
|
|
|
|
if (src) {
|
|
|
|
BUG_ON(atomic_read(&src->refcount) == 0);
|
|
|
|
atomic_inc(&src->refcount);
|
|
|
|
put_io_context(dst);
|
|
|
|
*pdst = src;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(copy_io_context);
|
|
|
|
|
|
|
|
void swap_io_context(struct io_context **ioc1, struct io_context **ioc2)
|
|
|
|
{
|
|
|
|
struct io_context *temp;
|
|
|
|
temp = *ioc1;
|
|
|
|
*ioc1 = *ioc2;
|
|
|
|
*ioc2 = temp;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(swap_io_context);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* sysfs parts below
|
|
|
|
*/
|
|
|
|
struct queue_sysfs_entry {
|
|
|
|
struct attribute attr;
|
|
|
|
ssize_t (*show)(struct request_queue *, char *);
|
|
|
|
ssize_t (*store)(struct request_queue *, const char *, size_t);
|
|
|
|
};
|
|
|
|
|
|
|
|
static ssize_t
|
|
|
|
queue_var_show(unsigned int var, char *page)
|
|
|
|
{
|
|
|
|
return sprintf(page, "%d\n", var);
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t
|
|
|
|
queue_var_store(unsigned long *var, const char *page, size_t count)
|
|
|
|
{
|
|
|
|
char *p = (char *) page;
|
|
|
|
|
|
|
|
*var = simple_strtoul(p, &p, 10);
|
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t queue_requests_show(struct request_queue *q, char *page)
|
|
|
|
{
|
|
|
|
return queue_var_show(q->nr_requests, (page));
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t
|
|
|
|
queue_requests_store(struct request_queue *q, const char *page, size_t count)
|
|
|
|
{
|
|
|
|
struct request_list *rl = &q->rq;
|
2006-03-18 19:51:29 +01:00
|
|
|
unsigned long nr;
|
|
|
|
int ret = queue_var_store(&nr, page, count);
|
|
|
|
if (nr < BLKDEV_MIN_RQ)
|
|
|
|
nr = BLKDEV_MIN_RQ;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-03-18 19:51:29 +01:00
|
|
|
spin_lock_irq(q->queue_lock);
|
|
|
|
q->nr_requests = nr;
|
2005-04-17 00:20:36 +02:00
|
|
|
blk_queue_congestion_threshold(q);
|
|
|
|
|
|
|
|
if (rl->count[READ] >= queue_congestion_on_threshold(q))
|
2006-10-20 08:28:15 +02:00
|
|
|
blk_set_queue_congested(q, READ);
|
2005-04-17 00:20:36 +02:00
|
|
|
else if (rl->count[READ] < queue_congestion_off_threshold(q))
|
2006-10-20 08:28:15 +02:00
|
|
|
blk_clear_queue_congested(q, READ);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (rl->count[WRITE] >= queue_congestion_on_threshold(q))
|
2006-10-20 08:28:15 +02:00
|
|
|
blk_set_queue_congested(q, WRITE);
|
2005-04-17 00:20:36 +02:00
|
|
|
else if (rl->count[WRITE] < queue_congestion_off_threshold(q))
|
2006-10-20 08:28:15 +02:00
|
|
|
blk_clear_queue_congested(q, WRITE);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (rl->count[READ] >= q->nr_requests) {
|
|
|
|
blk_set_queue_full(q, READ);
|
|
|
|
} else if (rl->count[READ]+1 <= q->nr_requests) {
|
|
|
|
blk_clear_queue_full(q, READ);
|
|
|
|
wake_up(&rl->wait[READ]);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (rl->count[WRITE] >= q->nr_requests) {
|
|
|
|
blk_set_queue_full(q, WRITE);
|
|
|
|
} else if (rl->count[WRITE]+1 <= q->nr_requests) {
|
|
|
|
blk_clear_queue_full(q, WRITE);
|
|
|
|
wake_up(&rl->wait[WRITE]);
|
|
|
|
}
|
2006-03-18 19:51:29 +01:00
|
|
|
spin_unlock_irq(q->queue_lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t queue_ra_show(struct request_queue *q, char *page)
|
|
|
|
{
|
|
|
|
int ra_kb = q->backing_dev_info.ra_pages << (PAGE_CACHE_SHIFT - 10);
|
|
|
|
|
|
|
|
return queue_var_show(ra_kb, (page));
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t
|
|
|
|
queue_ra_store(struct request_queue *q, const char *page, size_t count)
|
|
|
|
{
|
|
|
|
unsigned long ra_kb;
|
|
|
|
ssize_t ret = queue_var_store(&ra_kb, page, count);
|
|
|
|
|
|
|
|
spin_lock_irq(q->queue_lock);
|
|
|
|
q->backing_dev_info.ra_pages = ra_kb >> (PAGE_CACHE_SHIFT - 10);
|
|
|
|
spin_unlock_irq(q->queue_lock);
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t queue_max_sectors_show(struct request_queue *q, char *page)
|
|
|
|
{
|
|
|
|
int max_sectors_kb = q->max_sectors >> 1;
|
|
|
|
|
|
|
|
return queue_var_show(max_sectors_kb, (page));
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t
|
|
|
|
queue_max_sectors_store(struct request_queue *q, const char *page, size_t count)
|
|
|
|
{
|
|
|
|
unsigned long max_sectors_kb,
|
|
|
|
max_hw_sectors_kb = q->max_hw_sectors >> 1,
|
|
|
|
page_kb = 1 << (PAGE_CACHE_SHIFT - 10);
|
|
|
|
ssize_t ret = queue_var_store(&max_sectors_kb, page, count);
|
|
|
|
|
|
|
|
if (max_sectors_kb > max_hw_sectors_kb || max_sectors_kb < page_kb)
|
|
|
|
return -EINVAL;
|
|
|
|
/*
|
|
|
|
* Take the queue lock to update the readahead and max_sectors
|
|
|
|
* values synchronously:
|
|
|
|
*/
|
|
|
|
spin_lock_irq(q->queue_lock);
|
|
|
|
q->max_sectors = max_sectors_kb << 1;
|
|
|
|
spin_unlock_irq(q->queue_lock);
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t queue_max_hw_sectors_show(struct request_queue *q, char *page)
|
|
|
|
{
|
|
|
|
int max_hw_sectors_kb = q->max_hw_sectors >> 1;
|
|
|
|
|
|
|
|
return queue_var_show(max_hw_sectors_kb, (page));
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static struct queue_sysfs_entry queue_requests_entry = {
|
|
|
|
.attr = {.name = "nr_requests", .mode = S_IRUGO | S_IWUSR },
|
|
|
|
.show = queue_requests_show,
|
|
|
|
.store = queue_requests_store,
|
|
|
|
};
|
|
|
|
|
|
|
|
static struct queue_sysfs_entry queue_ra_entry = {
|
|
|
|
.attr = {.name = "read_ahead_kb", .mode = S_IRUGO | S_IWUSR },
|
|
|
|
.show = queue_ra_show,
|
|
|
|
.store = queue_ra_store,
|
|
|
|
};
|
|
|
|
|
|
|
|
static struct queue_sysfs_entry queue_max_sectors_entry = {
|
|
|
|
.attr = {.name = "max_sectors_kb", .mode = S_IRUGO | S_IWUSR },
|
|
|
|
.show = queue_max_sectors_show,
|
|
|
|
.store = queue_max_sectors_store,
|
|
|
|
};
|
|
|
|
|
|
|
|
static struct queue_sysfs_entry queue_max_hw_sectors_entry = {
|
|
|
|
.attr = {.name = "max_hw_sectors_kb", .mode = S_IRUGO },
|
|
|
|
.show = queue_max_hw_sectors_show,
|
|
|
|
};
|
|
|
|
|
|
|
|
static struct queue_sysfs_entry queue_iosched_entry = {
|
|
|
|
.attr = {.name = "scheduler", .mode = S_IRUGO | S_IWUSR },
|
|
|
|
.show = elv_iosched_show,
|
|
|
|
.store = elv_iosched_store,
|
|
|
|
};
|
|
|
|
|
|
|
|
static struct attribute *default_attrs[] = {
|
|
|
|
&queue_requests_entry.attr,
|
|
|
|
&queue_ra_entry.attr,
|
|
|
|
&queue_max_hw_sectors_entry.attr,
|
|
|
|
&queue_max_sectors_entry.attr,
|
|
|
|
&queue_iosched_entry.attr,
|
|
|
|
NULL,
|
|
|
|
};
|
|
|
|
|
|
|
|
#define to_queue(atr) container_of((atr), struct queue_sysfs_entry, attr)
|
|
|
|
|
|
|
|
static ssize_t
|
|
|
|
queue_attr_show(struct kobject *kobj, struct attribute *attr, char *page)
|
|
|
|
{
|
|
|
|
struct queue_sysfs_entry *entry = to_queue(attr);
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q =
|
|
|
|
container_of(kobj, struct request_queue, kobj);
|
2006-03-19 00:34:37 +01:00
|
|
|
ssize_t res;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (!entry->show)
|
2005-04-29 08:26:06 +02:00
|
|
|
return -EIO;
|
2006-03-19 00:34:37 +01:00
|
|
|
mutex_lock(&q->sysfs_lock);
|
|
|
|
if (test_bit(QUEUE_FLAG_DEAD, &q->queue_flags)) {
|
|
|
|
mutex_unlock(&q->sysfs_lock);
|
|
|
|
return -ENOENT;
|
|
|
|
}
|
|
|
|
res = entry->show(q, page);
|
|
|
|
mutex_unlock(&q->sysfs_lock);
|
|
|
|
return res;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t
|
|
|
|
queue_attr_store(struct kobject *kobj, struct attribute *attr,
|
|
|
|
const char *page, size_t length)
|
|
|
|
{
|
|
|
|
struct queue_sysfs_entry *entry = to_queue(attr);
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q = container_of(kobj, struct request_queue, kobj);
|
2006-03-19 00:34:37 +01:00
|
|
|
|
|
|
|
ssize_t res;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (!entry->store)
|
2005-04-29 08:26:06 +02:00
|
|
|
return -EIO;
|
2006-03-19 00:34:37 +01:00
|
|
|
mutex_lock(&q->sysfs_lock);
|
|
|
|
if (test_bit(QUEUE_FLAG_DEAD, &q->queue_flags)) {
|
|
|
|
mutex_unlock(&q->sysfs_lock);
|
|
|
|
return -ENOENT;
|
|
|
|
}
|
|
|
|
res = entry->store(q, page, length);
|
|
|
|
mutex_unlock(&q->sysfs_lock);
|
|
|
|
return res;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
static struct sysfs_ops queue_sysfs_ops = {
|
|
|
|
.show = queue_attr_show,
|
|
|
|
.store = queue_attr_store,
|
|
|
|
};
|
|
|
|
|
2005-06-25 23:59:10 +02:00
|
|
|
static struct kobj_type queue_ktype = {
|
2005-04-17 00:20:36 +02:00
|
|
|
.sysfs_ops = &queue_sysfs_ops,
|
|
|
|
.default_attrs = default_attrs,
|
2006-03-19 00:34:37 +01:00
|
|
|
.release = blk_release_queue,
|
2005-04-17 00:20:36 +02:00
|
|
|
};
|
|
|
|
|
|
|
|
int blk_register_queue(struct gendisk *disk)
|
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q = disk->queue;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (!q || !q->request_fn)
|
|
|
|
return -ENXIO;
|
|
|
|
|
2007-12-18 07:05:35 +01:00
|
|
|
ret = kobject_add_ng(&q->kobj, kobject_get(&disk->dev.kobj),
|
|
|
|
"%s", "queue");
|
2005-04-17 00:20:36 +02:00
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
|
2006-03-19 00:34:37 +01:00
|
|
|
kobject_uevent(&q->kobj, KOBJ_ADD);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
ret = elv_register_queue(q);
|
|
|
|
if (ret) {
|
2006-03-19 00:34:37 +01:00
|
|
|
kobject_uevent(&q->kobj, KOBJ_REMOVE);
|
|
|
|
kobject_del(&q->kobj);
|
2005-04-17 00:20:36 +02:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
void blk_unregister_queue(struct gendisk *disk)
|
|
|
|
{
|
2007-07-24 09:28:11 +02:00
|
|
|
struct request_queue *q = disk->queue;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (q && q->request_fn) {
|
|
|
|
elv_unregister_queue(q);
|
|
|
|
|
2006-03-19 00:34:37 +01:00
|
|
|
kobject_uevent(&q->kobj, KOBJ_REMOVE);
|
|
|
|
kobject_del(&q->kobj);
|
2007-05-21 22:08:01 +02:00
|
|
|
kobject_put(&disk->dev.kobj);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
}
|