![Ryusuke Konishi](/assets/img/avatar_default.png)
This separates wait function for submitted logs from the write function nilfs_segctor_write(). A new list of segment buffers "sc_write_logs" is added to hold logs under writing, and double buffering is partially applied to hide io latency. At this point, the double buffering is disabled for blocksize < pagesize because page dirty flag is turned off during write and dirty buffers are not properly collected for pages crossing over segments. To receive full benefit of the double buffering, further refinement is needed to move the io wait outside the lock section of log writer. Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
516 lines
13 KiB
C
516 lines
13 KiB
C
/*
|
|
* segbuf.c - NILFS segment buffer
|
|
*
|
|
* Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
|
|
*
|
|
* This program is free software; you can redistribute it and/or modify
|
|
* it under the terms of the GNU General Public License as published by
|
|
* the Free Software Foundation; either version 2 of the License, or
|
|
* (at your option) any later version.
|
|
*
|
|
* This program is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
* GNU General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License
|
|
* along with this program; if not, write to the Free Software
|
|
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
|
|
*
|
|
* Written by Ryusuke Konishi <ryusuke@osrg.net>
|
|
*
|
|
*/
|
|
|
|
#include <linux/buffer_head.h>
|
|
#include <linux/writeback.h>
|
|
#include <linux/crc32.h>
|
|
#include <linux/backing-dev.h>
|
|
#include "page.h"
|
|
#include "segbuf.h"
|
|
|
|
|
|
struct nilfs_write_info {
|
|
struct the_nilfs *nilfs;
|
|
struct bio *bio;
|
|
int start, end; /* The region to be submitted */
|
|
int rest_blocks;
|
|
int max_pages;
|
|
int nr_vecs;
|
|
sector_t blocknr;
|
|
};
|
|
|
|
|
|
static struct kmem_cache *nilfs_segbuf_cachep;
|
|
|
|
static void nilfs_segbuf_init_once(void *obj)
|
|
{
|
|
memset(obj, 0, sizeof(struct nilfs_segment_buffer));
|
|
}
|
|
|
|
int __init nilfs_init_segbuf_cache(void)
|
|
{
|
|
nilfs_segbuf_cachep =
|
|
kmem_cache_create("nilfs2_segbuf_cache",
|
|
sizeof(struct nilfs_segment_buffer),
|
|
0, SLAB_RECLAIM_ACCOUNT,
|
|
nilfs_segbuf_init_once);
|
|
|
|
return (nilfs_segbuf_cachep == NULL) ? -ENOMEM : 0;
|
|
}
|
|
|
|
void nilfs_destroy_segbuf_cache(void)
|
|
{
|
|
kmem_cache_destroy(nilfs_segbuf_cachep);
|
|
}
|
|
|
|
struct nilfs_segment_buffer *nilfs_segbuf_new(struct super_block *sb)
|
|
{
|
|
struct nilfs_segment_buffer *segbuf;
|
|
|
|
segbuf = kmem_cache_alloc(nilfs_segbuf_cachep, GFP_NOFS);
|
|
if (unlikely(!segbuf))
|
|
return NULL;
|
|
|
|
segbuf->sb_super = sb;
|
|
INIT_LIST_HEAD(&segbuf->sb_list);
|
|
INIT_LIST_HEAD(&segbuf->sb_segsum_buffers);
|
|
INIT_LIST_HEAD(&segbuf->sb_payload_buffers);
|
|
|
|
init_completion(&segbuf->sb_bio_event);
|
|
atomic_set(&segbuf->sb_err, 0);
|
|
segbuf->sb_nbio = 0;
|
|
|
|
return segbuf;
|
|
}
|
|
|
|
void nilfs_segbuf_free(struct nilfs_segment_buffer *segbuf)
|
|
{
|
|
kmem_cache_free(nilfs_segbuf_cachep, segbuf);
|
|
}
|
|
|
|
void nilfs_segbuf_map(struct nilfs_segment_buffer *segbuf, __u64 segnum,
|
|
unsigned long offset, struct the_nilfs *nilfs)
|
|
{
|
|
segbuf->sb_segnum = segnum;
|
|
nilfs_get_segment_range(nilfs, segnum, &segbuf->sb_fseg_start,
|
|
&segbuf->sb_fseg_end);
|
|
|
|
segbuf->sb_pseg_start = segbuf->sb_fseg_start + offset;
|
|
segbuf->sb_rest_blocks =
|
|
segbuf->sb_fseg_end - segbuf->sb_pseg_start + 1;
|
|
}
|
|
|
|
/**
|
|
* nilfs_segbuf_map_cont - map a new log behind a given log
|
|
* @segbuf: new segment buffer
|
|
* @prev: segment buffer containing a log to be continued
|
|
*/
|
|
void nilfs_segbuf_map_cont(struct nilfs_segment_buffer *segbuf,
|
|
struct nilfs_segment_buffer *prev)
|
|
{
|
|
segbuf->sb_segnum = prev->sb_segnum;
|
|
segbuf->sb_fseg_start = prev->sb_fseg_start;
|
|
segbuf->sb_fseg_end = prev->sb_fseg_end;
|
|
segbuf->sb_pseg_start = prev->sb_pseg_start + prev->sb_sum.nblocks;
|
|
segbuf->sb_rest_blocks =
|
|
segbuf->sb_fseg_end - segbuf->sb_pseg_start + 1;
|
|
}
|
|
|
|
void nilfs_segbuf_set_next_segnum(struct nilfs_segment_buffer *segbuf,
|
|
__u64 nextnum, struct the_nilfs *nilfs)
|
|
{
|
|
segbuf->sb_nextnum = nextnum;
|
|
segbuf->sb_sum.next = nilfs_get_segment_start_blocknr(nilfs, nextnum);
|
|
}
|
|
|
|
int nilfs_segbuf_extend_segsum(struct nilfs_segment_buffer *segbuf)
|
|
{
|
|
struct buffer_head *bh;
|
|
|
|
bh = sb_getblk(segbuf->sb_super,
|
|
segbuf->sb_pseg_start + segbuf->sb_sum.nsumblk);
|
|
if (unlikely(!bh))
|
|
return -ENOMEM;
|
|
|
|
nilfs_segbuf_add_segsum_buffer(segbuf, bh);
|
|
return 0;
|
|
}
|
|
|
|
int nilfs_segbuf_extend_payload(struct nilfs_segment_buffer *segbuf,
|
|
struct buffer_head **bhp)
|
|
{
|
|
struct buffer_head *bh;
|
|
|
|
bh = sb_getblk(segbuf->sb_super,
|
|
segbuf->sb_pseg_start + segbuf->sb_sum.nblocks);
|
|
if (unlikely(!bh))
|
|
return -ENOMEM;
|
|
|
|
nilfs_segbuf_add_payload_buffer(segbuf, bh);
|
|
*bhp = bh;
|
|
return 0;
|
|
}
|
|
|
|
int nilfs_segbuf_reset(struct nilfs_segment_buffer *segbuf, unsigned flags,
|
|
time_t ctime)
|
|
{
|
|
int err;
|
|
|
|
segbuf->sb_sum.nblocks = segbuf->sb_sum.nsumblk = 0;
|
|
err = nilfs_segbuf_extend_segsum(segbuf);
|
|
if (unlikely(err))
|
|
return err;
|
|
|
|
segbuf->sb_sum.flags = flags;
|
|
segbuf->sb_sum.sumbytes = sizeof(struct nilfs_segment_summary);
|
|
segbuf->sb_sum.nfinfo = segbuf->sb_sum.nfileblk = 0;
|
|
segbuf->sb_sum.ctime = ctime;
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* Setup segument summary
|
|
*/
|
|
void nilfs_segbuf_fill_in_segsum(struct nilfs_segment_buffer *segbuf)
|
|
{
|
|
struct nilfs_segment_summary *raw_sum;
|
|
struct buffer_head *bh_sum;
|
|
|
|
bh_sum = list_entry(segbuf->sb_segsum_buffers.next,
|
|
struct buffer_head, b_assoc_buffers);
|
|
raw_sum = (struct nilfs_segment_summary *)bh_sum->b_data;
|
|
|
|
raw_sum->ss_magic = cpu_to_le32(NILFS_SEGSUM_MAGIC);
|
|
raw_sum->ss_bytes = cpu_to_le16(sizeof(*raw_sum));
|
|
raw_sum->ss_flags = cpu_to_le16(segbuf->sb_sum.flags);
|
|
raw_sum->ss_seq = cpu_to_le64(segbuf->sb_sum.seg_seq);
|
|
raw_sum->ss_create = cpu_to_le64(segbuf->sb_sum.ctime);
|
|
raw_sum->ss_next = cpu_to_le64(segbuf->sb_sum.next);
|
|
raw_sum->ss_nblocks = cpu_to_le32(segbuf->sb_sum.nblocks);
|
|
raw_sum->ss_nfinfo = cpu_to_le32(segbuf->sb_sum.nfinfo);
|
|
raw_sum->ss_sumbytes = cpu_to_le32(segbuf->sb_sum.sumbytes);
|
|
raw_sum->ss_pad = 0;
|
|
}
|
|
|
|
/*
|
|
* CRC calculation routines
|
|
*/
|
|
void nilfs_segbuf_fill_in_segsum_crc(struct nilfs_segment_buffer *segbuf,
|
|
u32 seed)
|
|
{
|
|
struct buffer_head *bh;
|
|
struct nilfs_segment_summary *raw_sum;
|
|
unsigned long size, bytes = segbuf->sb_sum.sumbytes;
|
|
u32 crc;
|
|
|
|
bh = list_entry(segbuf->sb_segsum_buffers.next, struct buffer_head,
|
|
b_assoc_buffers);
|
|
|
|
raw_sum = (struct nilfs_segment_summary *)bh->b_data;
|
|
size = min_t(unsigned long, bytes, bh->b_size);
|
|
crc = crc32_le(seed,
|
|
(unsigned char *)raw_sum +
|
|
sizeof(raw_sum->ss_datasum) + sizeof(raw_sum->ss_sumsum),
|
|
size - (sizeof(raw_sum->ss_datasum) +
|
|
sizeof(raw_sum->ss_sumsum)));
|
|
|
|
list_for_each_entry_continue(bh, &segbuf->sb_segsum_buffers,
|
|
b_assoc_buffers) {
|
|
bytes -= size;
|
|
size = min_t(unsigned long, bytes, bh->b_size);
|
|
crc = crc32_le(crc, bh->b_data, size);
|
|
}
|
|
raw_sum->ss_sumsum = cpu_to_le32(crc);
|
|
}
|
|
|
|
void nilfs_segbuf_fill_in_data_crc(struct nilfs_segment_buffer *segbuf,
|
|
u32 seed)
|
|
{
|
|
struct buffer_head *bh;
|
|
struct nilfs_segment_summary *raw_sum;
|
|
void *kaddr;
|
|
u32 crc;
|
|
|
|
bh = list_entry(segbuf->sb_segsum_buffers.next, struct buffer_head,
|
|
b_assoc_buffers);
|
|
raw_sum = (struct nilfs_segment_summary *)bh->b_data;
|
|
crc = crc32_le(seed,
|
|
(unsigned char *)raw_sum + sizeof(raw_sum->ss_datasum),
|
|
bh->b_size - sizeof(raw_sum->ss_datasum));
|
|
|
|
list_for_each_entry_continue(bh, &segbuf->sb_segsum_buffers,
|
|
b_assoc_buffers) {
|
|
crc = crc32_le(crc, bh->b_data, bh->b_size);
|
|
}
|
|
list_for_each_entry(bh, &segbuf->sb_payload_buffers, b_assoc_buffers) {
|
|
kaddr = kmap_atomic(bh->b_page, KM_USER0);
|
|
crc = crc32_le(crc, kaddr + bh_offset(bh), bh->b_size);
|
|
kunmap_atomic(kaddr, KM_USER0);
|
|
}
|
|
raw_sum->ss_datasum = cpu_to_le32(crc);
|
|
}
|
|
|
|
static void nilfs_release_buffers(struct list_head *list)
|
|
{
|
|
struct buffer_head *bh, *n;
|
|
|
|
list_for_each_entry_safe(bh, n, list, b_assoc_buffers) {
|
|
list_del_init(&bh->b_assoc_buffers);
|
|
if (buffer_nilfs_allocated(bh)) {
|
|
struct page *clone_page = bh->b_page;
|
|
|
|
/* remove clone page */
|
|
brelse(bh);
|
|
page_cache_release(clone_page); /* for each bh */
|
|
if (page_count(clone_page) <= 2) {
|
|
lock_page(clone_page);
|
|
nilfs_free_private_page(clone_page);
|
|
}
|
|
continue;
|
|
}
|
|
brelse(bh);
|
|
}
|
|
}
|
|
|
|
static void nilfs_segbuf_clear(struct nilfs_segment_buffer *segbuf)
|
|
{
|
|
nilfs_release_buffers(&segbuf->sb_segsum_buffers);
|
|
nilfs_release_buffers(&segbuf->sb_payload_buffers);
|
|
}
|
|
|
|
/*
|
|
* Iterators for segment buffers
|
|
*/
|
|
void nilfs_clear_logs(struct list_head *logs)
|
|
{
|
|
struct nilfs_segment_buffer *segbuf;
|
|
|
|
list_for_each_entry(segbuf, logs, sb_list)
|
|
nilfs_segbuf_clear(segbuf);
|
|
}
|
|
|
|
void nilfs_truncate_logs(struct list_head *logs,
|
|
struct nilfs_segment_buffer *last)
|
|
{
|
|
struct nilfs_segment_buffer *n, *segbuf;
|
|
|
|
segbuf = list_prepare_entry(last, logs, sb_list);
|
|
list_for_each_entry_safe_continue(segbuf, n, logs, sb_list) {
|
|
list_del_init(&segbuf->sb_list);
|
|
nilfs_segbuf_clear(segbuf);
|
|
nilfs_segbuf_free(segbuf);
|
|
}
|
|
}
|
|
|
|
int nilfs_wait_on_logs(struct list_head *logs)
|
|
{
|
|
struct nilfs_segment_buffer *segbuf;
|
|
int err;
|
|
|
|
list_for_each_entry(segbuf, logs, sb_list) {
|
|
err = nilfs_segbuf_wait(segbuf);
|
|
if (err)
|
|
return err;
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* BIO operations
|
|
*/
|
|
static void nilfs_end_bio_write(struct bio *bio, int err)
|
|
{
|
|
const int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
|
|
struct nilfs_segment_buffer *segbuf = bio->bi_private;
|
|
|
|
if (err == -EOPNOTSUPP) {
|
|
set_bit(BIO_EOPNOTSUPP, &bio->bi_flags);
|
|
bio_put(bio);
|
|
/* to be detected by submit_seg_bio() */
|
|
}
|
|
|
|
if (!uptodate)
|
|
atomic_inc(&segbuf->sb_err);
|
|
|
|
bio_put(bio);
|
|
complete(&segbuf->sb_bio_event);
|
|
}
|
|
|
|
static int nilfs_segbuf_submit_bio(struct nilfs_segment_buffer *segbuf,
|
|
struct nilfs_write_info *wi, int mode)
|
|
{
|
|
struct bio *bio = wi->bio;
|
|
int err;
|
|
|
|
if (segbuf->sb_nbio > 0 && bdi_write_congested(wi->nilfs->ns_bdi)) {
|
|
wait_for_completion(&segbuf->sb_bio_event);
|
|
segbuf->sb_nbio--;
|
|
if (unlikely(atomic_read(&segbuf->sb_err))) {
|
|
bio_put(bio);
|
|
err = -EIO;
|
|
goto failed;
|
|
}
|
|
}
|
|
|
|
bio->bi_end_io = nilfs_end_bio_write;
|
|
bio->bi_private = segbuf;
|
|
bio_get(bio);
|
|
submit_bio(mode, bio);
|
|
if (bio_flagged(bio, BIO_EOPNOTSUPP)) {
|
|
bio_put(bio);
|
|
err = -EOPNOTSUPP;
|
|
goto failed;
|
|
}
|
|
segbuf->sb_nbio++;
|
|
bio_put(bio);
|
|
|
|
wi->bio = NULL;
|
|
wi->rest_blocks -= wi->end - wi->start;
|
|
wi->nr_vecs = min(wi->max_pages, wi->rest_blocks);
|
|
wi->start = wi->end;
|
|
return 0;
|
|
|
|
failed:
|
|
wi->bio = NULL;
|
|
return err;
|
|
}
|
|
|
|
/**
|
|
* nilfs_alloc_seg_bio - allocate a new bio for writing log
|
|
* @nilfs: nilfs object
|
|
* @start: start block number of the bio
|
|
* @nr_vecs: request size of page vector.
|
|
*
|
|
* Return Value: On success, pointer to the struct bio is returned.
|
|
* On error, NULL is returned.
|
|
*/
|
|
static struct bio *nilfs_alloc_seg_bio(struct the_nilfs *nilfs, sector_t start,
|
|
int nr_vecs)
|
|
{
|
|
struct bio *bio;
|
|
|
|
bio = bio_alloc(GFP_NOIO, nr_vecs);
|
|
if (bio == NULL) {
|
|
while (!bio && (nr_vecs >>= 1))
|
|
bio = bio_alloc(GFP_NOIO, nr_vecs);
|
|
}
|
|
if (likely(bio)) {
|
|
bio->bi_bdev = nilfs->ns_bdev;
|
|
bio->bi_sector = start << (nilfs->ns_blocksize_bits - 9);
|
|
}
|
|
return bio;
|
|
}
|
|
|
|
static void nilfs_segbuf_prepare_write(struct nilfs_segment_buffer *segbuf,
|
|
struct nilfs_write_info *wi)
|
|
{
|
|
wi->bio = NULL;
|
|
wi->rest_blocks = segbuf->sb_sum.nblocks;
|
|
wi->max_pages = bio_get_nr_vecs(wi->nilfs->ns_bdev);
|
|
wi->nr_vecs = min(wi->max_pages, wi->rest_blocks);
|
|
wi->start = wi->end = 0;
|
|
wi->blocknr = segbuf->sb_pseg_start;
|
|
}
|
|
|
|
static int nilfs_segbuf_submit_bh(struct nilfs_segment_buffer *segbuf,
|
|
struct nilfs_write_info *wi,
|
|
struct buffer_head *bh, int mode)
|
|
{
|
|
int len, err;
|
|
|
|
BUG_ON(wi->nr_vecs <= 0);
|
|
repeat:
|
|
if (!wi->bio) {
|
|
wi->bio = nilfs_alloc_seg_bio(wi->nilfs, wi->blocknr + wi->end,
|
|
wi->nr_vecs);
|
|
if (unlikely(!wi->bio))
|
|
return -ENOMEM;
|
|
}
|
|
|
|
len = bio_add_page(wi->bio, bh->b_page, bh->b_size, bh_offset(bh));
|
|
if (len == bh->b_size) {
|
|
wi->end++;
|
|
return 0;
|
|
}
|
|
/* bio is FULL */
|
|
err = nilfs_segbuf_submit_bio(segbuf, wi, mode);
|
|
/* never submit current bh */
|
|
if (likely(!err))
|
|
goto repeat;
|
|
return err;
|
|
}
|
|
|
|
/**
|
|
* nilfs_segbuf_write - submit write requests of a log
|
|
* @segbuf: buffer storing a log to be written
|
|
* @nilfs: nilfs object
|
|
*
|
|
* Return Value: On Success, 0 is returned. On Error, one of the following
|
|
* negative error code is returned.
|
|
*
|
|
* %-EIO - I/O error
|
|
*
|
|
* %-ENOMEM - Insufficient memory available.
|
|
*/
|
|
int nilfs_segbuf_write(struct nilfs_segment_buffer *segbuf,
|
|
struct the_nilfs *nilfs)
|
|
{
|
|
struct nilfs_write_info wi;
|
|
struct buffer_head *bh;
|
|
int res = 0, rw = WRITE;
|
|
|
|
wi.nilfs = nilfs;
|
|
nilfs_segbuf_prepare_write(segbuf, &wi);
|
|
|
|
list_for_each_entry(bh, &segbuf->sb_segsum_buffers, b_assoc_buffers) {
|
|
res = nilfs_segbuf_submit_bh(segbuf, &wi, bh, rw);
|
|
if (unlikely(res))
|
|
goto failed_bio;
|
|
}
|
|
|
|
list_for_each_entry(bh, &segbuf->sb_payload_buffers, b_assoc_buffers) {
|
|
res = nilfs_segbuf_submit_bh(segbuf, &wi, bh, rw);
|
|
if (unlikely(res))
|
|
goto failed_bio;
|
|
}
|
|
|
|
if (wi.bio) {
|
|
/*
|
|
* Last BIO is always sent through the following
|
|
* submission.
|
|
*/
|
|
rw |= (1 << BIO_RW_SYNCIO) | (1 << BIO_RW_UNPLUG);
|
|
res = nilfs_segbuf_submit_bio(segbuf, &wi, rw);
|
|
}
|
|
|
|
failed_bio:
|
|
return res;
|
|
}
|
|
|
|
/**
|
|
* nilfs_segbuf_wait - wait for completion of requested BIOs
|
|
* @segbuf: segment buffer
|
|
*
|
|
* Return Value: On Success, 0 is returned. On Error, one of the following
|
|
* negative error code is returned.
|
|
*
|
|
* %-EIO - I/O error
|
|
*/
|
|
int nilfs_segbuf_wait(struct nilfs_segment_buffer *segbuf)
|
|
{
|
|
int err = 0;
|
|
|
|
if (!segbuf->sb_nbio)
|
|
return 0;
|
|
|
|
do {
|
|
wait_for_completion(&segbuf->sb_bio_event);
|
|
} while (--segbuf->sb_nbio > 0);
|
|
|
|
if (unlikely(atomic_read(&segbuf->sb_err) > 0)) {
|
|
printk(KERN_ERR "NILFS: IO error writing segment\n");
|
|
err = -EIO;
|
|
}
|
|
return err;
|
|
}
|