2010-10-28 03:30:10 +02:00
|
|
|
/*
|
|
|
|
* linux/fs/ext4/page-io.c
|
|
|
|
*
|
|
|
|
* This contains the new page_io functions for ext4
|
|
|
|
*
|
|
|
|
* Written by Theodore Ts'o, 2010.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/fs.h>
|
|
|
|
#include <linux/time.h>
|
|
|
|
#include <linux/jbd2.h>
|
|
|
|
#include <linux/highuid.h>
|
|
|
|
#include <linux/pagemap.h>
|
|
|
|
#include <linux/quotaops.h>
|
|
|
|
#include <linux/string.h>
|
|
|
|
#include <linux/buffer_head.h>
|
|
|
|
#include <linux/writeback.h>
|
|
|
|
#include <linux/pagevec.h>
|
|
|
|
#include <linux/mpage.h>
|
|
|
|
#include <linux/namei.h>
|
2013-05-08 01:19:08 +02:00
|
|
|
#include <linux/aio.h>
|
2010-10-28 03:30:10 +02:00
|
|
|
#include <linux/uio.h>
|
|
|
|
#include <linux/bio.h>
|
|
|
|
#include <linux/workqueue.h>
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/slab.h>
|
2013-01-28 15:32:54 +01:00
|
|
|
#include <linux/mm.h>
|
2013-07-12 04:42:42 +02:00
|
|
|
#include <linux/ratelimit.h>
|
2010-10-28 03:30:10 +02:00
|
|
|
|
|
|
|
#include "ext4_jbd2.h"
|
|
|
|
#include "xattr.h"
|
|
|
|
#include "acl.h"
|
|
|
|
|
2013-04-12 05:48:32 +02:00
|
|
|
static struct kmem_cache *io_end_cachep;
|
2010-10-28 03:30:10 +02:00
|
|
|
|
2010-10-28 03:30:14 +02:00
|
|
|
int __init ext4_init_pageio(void)
|
2010-10-28 03:30:10 +02:00
|
|
|
{
|
|
|
|
io_end_cachep = KMEM_CACHE(ext4_io_end, SLAB_RECLAIM_ACCOUNT);
|
2013-04-12 05:48:32 +02:00
|
|
|
if (io_end_cachep == NULL)
|
2010-10-28 03:30:10 +02:00
|
|
|
return -ENOMEM;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2010-10-28 03:30:14 +02:00
|
|
|
void ext4_exit_pageio(void)
|
2010-10-28 03:30:10 +02:00
|
|
|
{
|
|
|
|
kmem_cache_destroy(io_end_cachep);
|
|
|
|
}
|
|
|
|
|
2013-06-04 20:23:41 +02:00
|
|
|
/*
|
|
|
|
* Print an buffer I/O error compatible with the fs/buffer.c. This
|
|
|
|
* provides compatibility with dmesg scrapers that look for a specific
|
|
|
|
* buffer I/O error message. We really need a unified error reporting
|
|
|
|
* structure to userspace ala Digital Unix's uerf system, but it's
|
|
|
|
* probably not going to happen in my lifetime, due to LKML politics...
|
|
|
|
*/
|
|
|
|
static void buffer_io_error(struct buffer_head *bh)
|
|
|
|
{
|
|
|
|
char b[BDEVNAME_SIZE];
|
2013-07-12 04:42:42 +02:00
|
|
|
printk_ratelimited(KERN_ERR "Buffer I/O error on device %s, logical block %llu\n",
|
2013-06-04 20:23:41 +02:00
|
|
|
bdevname(bh->b_bdev, b),
|
|
|
|
(unsigned long long)bh->b_blocknr);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ext4_finish_bio(struct bio *bio)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
int error = !test_bit(BIO_UPTODATE, &bio->bi_flags);
|
2013-11-07 21:20:26 +01:00
|
|
|
struct bio_vec *bvec;
|
2013-06-04 20:23:41 +02:00
|
|
|
|
2013-11-07 21:20:26 +01:00
|
|
|
bio_for_each_segment_all(bvec, bio, i) {
|
2013-06-04 20:23:41 +02:00
|
|
|
struct page *page = bvec->bv_page;
|
|
|
|
struct buffer_head *bh, *head;
|
|
|
|
unsigned bio_start = bvec->bv_offset;
|
|
|
|
unsigned bio_end = bio_start + bvec->bv_len;
|
|
|
|
unsigned under_io = 0;
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
if (!page)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
if (error) {
|
|
|
|
SetPageError(page);
|
|
|
|
set_bit(AS_EIO, &page->mapping->flags);
|
|
|
|
}
|
|
|
|
bh = head = page_buffers(page);
|
|
|
|
/*
|
|
|
|
* We check all buffers in the page under BH_Uptodate_Lock
|
|
|
|
* to avoid races with other end io clearing async_write flags
|
|
|
|
*/
|
|
|
|
local_irq_save(flags);
|
|
|
|
bit_spin_lock(BH_Uptodate_Lock, &head->b_state);
|
|
|
|
do {
|
|
|
|
if (bh_offset(bh) < bio_start ||
|
|
|
|
bh_offset(bh) + bh->b_size > bio_end) {
|
|
|
|
if (buffer_async_write(bh))
|
|
|
|
under_io++;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
clear_buffer_async_write(bh);
|
|
|
|
if (error)
|
|
|
|
buffer_io_error(bh);
|
|
|
|
} while ((bh = bh->b_this_page) != head);
|
|
|
|
bit_spin_unlock(BH_Uptodate_Lock, &head->b_state);
|
|
|
|
local_irq_restore(flags);
|
|
|
|
if (!under_io)
|
|
|
|
end_page_writeback(page);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-06-04 17:58:58 +02:00
|
|
|
static void ext4_release_io_end(ext4_io_end_t *io_end)
|
2010-10-28 03:30:10 +02:00
|
|
|
{
|
2013-06-04 20:23:41 +02:00
|
|
|
struct bio *bio, *next_bio;
|
|
|
|
|
2013-06-04 17:58:58 +02:00
|
|
|
BUG_ON(!list_empty(&io_end->list));
|
|
|
|
BUG_ON(io_end->flag & EXT4_IO_END_UNWRITTEN);
|
2013-06-04 19:21:11 +02:00
|
|
|
WARN_ON(io_end->handle);
|
2013-06-04 17:58:58 +02:00
|
|
|
|
|
|
|
if (atomic_dec_and_test(&EXT4_I(io_end->inode)->i_ioend_count))
|
|
|
|
wake_up_all(ext4_ioend_wq(io_end->inode));
|
2013-06-04 20:23:41 +02:00
|
|
|
|
|
|
|
for (bio = io_end->bio; bio; bio = next_bio) {
|
|
|
|
next_bio = bio->bi_private;
|
|
|
|
ext4_finish_bio(bio);
|
|
|
|
bio_put(bio);
|
|
|
|
}
|
2013-06-04 17:58:58 +02:00
|
|
|
kmem_cache_free(io_end_cachep, io_end);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
|
|
|
|
{
|
|
|
|
struct inode *inode = io_end->inode;
|
2012-09-29 05:36:25 +02:00
|
|
|
|
2013-06-04 17:58:58 +02:00
|
|
|
io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
|
|
|
|
/* Wake up anyone waiting on unwritten extent conversion */
|
|
|
|
if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
|
|
|
|
wake_up_all(ext4_ioend_wq(inode));
|
2010-10-28 03:30:10 +02:00
|
|
|
}
|
|
|
|
|
2013-06-04 20:30:00 +02:00
|
|
|
/*
|
|
|
|
* Check a range of space and convert unwritten extents to written. Note that
|
|
|
|
* we are protected from truncate touching same part of extent tree by the
|
|
|
|
* fact that truncate code waits for all DIO to finish (thus exclusion from
|
|
|
|
* direct IO is achieved) and also waits for PageWriteback bits. Thus we
|
|
|
|
* cannot get to ext4_ext_truncate() before all IOs overlapping that range are
|
|
|
|
* completed (happens from ext4_free_ioend()).
|
|
|
|
*/
|
2012-09-29 06:14:55 +02:00
|
|
|
static int ext4_end_io(ext4_io_end_t *io)
|
2010-10-28 03:30:10 +02:00
|
|
|
{
|
|
|
|
struct inode *inode = io->inode;
|
|
|
|
loff_t offset = io->offset;
|
|
|
|
ssize_t size = io->size;
|
2013-06-04 19:21:11 +02:00
|
|
|
handle_t *handle = io->handle;
|
2010-10-28 03:30:10 +02:00
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
ext4_debug("ext4_end_io_nolock: io 0x%p from inode %lu,list->next 0x%p,"
|
|
|
|
"list->prev 0x%p\n",
|
|
|
|
io, inode->i_ino, io->list.next, io->list.prev);
|
|
|
|
|
2013-06-04 19:21:11 +02:00
|
|
|
io->handle = NULL; /* Following call will use up the handle */
|
|
|
|
ret = ext4_convert_unwritten_extents(handle, inode, offset, size);
|
2010-10-28 03:30:10 +02:00
|
|
|
if (ret < 0) {
|
2011-10-31 15:56:32 +01:00
|
|
|
ext4_msg(inode->i_sb, KERN_EMERG,
|
|
|
|
"failed to convert unwritten extents to written "
|
|
|
|
"extents -- potential data loss! "
|
|
|
|
"(inode %lu, offset %llu, size %zd, error %d)",
|
|
|
|
inode->i_ino, offset, size, ret);
|
2010-10-28 03:30:10 +02:00
|
|
|
}
|
2013-06-04 17:58:58 +02:00
|
|
|
ext4_clear_io_unwritten_flag(io);
|
|
|
|
ext4_release_io_end(io);
|
2010-10-28 03:30:10 +02:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2013-06-04 20:21:02 +02:00
|
|
|
static void dump_completed_IO(struct inode *inode, struct list_head *head)
|
2012-09-29 06:14:55 +02:00
|
|
|
{
|
|
|
|
#ifdef EXT4FS_DEBUG
|
|
|
|
struct list_head *cur, *before, *after;
|
|
|
|
ext4_io_end_t *io, *io0, *io1;
|
|
|
|
|
2013-06-04 20:21:02 +02:00
|
|
|
if (list_empty(head))
|
2012-09-29 06:14:55 +02:00
|
|
|
return;
|
|
|
|
|
2013-06-04 20:21:02 +02:00
|
|
|
ext4_debug("Dump inode %lu completed io list\n", inode->i_ino);
|
|
|
|
list_for_each_entry(io, head, list) {
|
2012-09-29 06:14:55 +02:00
|
|
|
cur = &io->list;
|
|
|
|
before = cur->prev;
|
|
|
|
io0 = container_of(before, ext4_io_end_t, list);
|
|
|
|
after = cur->next;
|
|
|
|
io1 = container_of(after, ext4_io_end_t, list);
|
|
|
|
|
|
|
|
ext4_debug("io 0x%p from inode %lu,prev 0x%p,next 0x%p\n",
|
|
|
|
io, inode->i_ino, io0, io1);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Add the io_end to per-inode completed end_io list. */
|
2013-06-04 17:58:58 +02:00
|
|
|
static void ext4_add_complete_io(ext4_io_end_t *io_end)
|
2010-10-28 03:30:10 +02:00
|
|
|
{
|
2012-09-29 06:14:55 +02:00
|
|
|
struct ext4_inode_info *ei = EXT4_I(io_end->inode);
|
2013-10-16 14:25:11 +02:00
|
|
|
struct ext4_sb_info *sbi = EXT4_SB(io_end->inode->i_sb);
|
2012-09-29 06:14:55 +02:00
|
|
|
struct workqueue_struct *wq;
|
|
|
|
unsigned long flags;
|
|
|
|
|
2013-09-04 15:04:39 +02:00
|
|
|
/* Only reserved conversions from writeback should enter here */
|
|
|
|
WARN_ON(!(io_end->flag & EXT4_IO_END_UNWRITTEN));
|
2013-10-16 14:25:11 +02:00
|
|
|
WARN_ON(!io_end->handle && sbi->s_journal);
|
2011-10-30 23:26:08 +01:00
|
|
|
spin_lock_irqsave(&ei->i_completed_io_lock, flags);
|
2013-10-16 14:25:11 +02:00
|
|
|
wq = sbi->rsv_conversion_wq;
|
2013-09-04 15:04:39 +02:00
|
|
|
if (list_empty(&ei->i_rsv_conversion_list))
|
|
|
|
queue_work(wq, &ei->i_rsv_conversion_work);
|
|
|
|
list_add_tail(&io_end->list, &ei->i_rsv_conversion_list);
|
2012-09-29 06:14:55 +02:00
|
|
|
spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
|
|
|
|
}
|
2011-10-30 23:26:08 +01:00
|
|
|
|
2013-06-04 20:21:02 +02:00
|
|
|
static int ext4_do_flush_completed_IO(struct inode *inode,
|
|
|
|
struct list_head *head)
|
2012-09-29 06:14:55 +02:00
|
|
|
{
|
|
|
|
ext4_io_end_t *io;
|
2013-01-28 15:49:15 +01:00
|
|
|
struct list_head unwritten;
|
2012-09-29 06:14:55 +02:00
|
|
|
unsigned long flags;
|
|
|
|
struct ext4_inode_info *ei = EXT4_I(inode);
|
|
|
|
int err, ret = 0;
|
|
|
|
|
|
|
|
spin_lock_irqsave(&ei->i_completed_io_lock, flags);
|
2013-06-04 20:21:02 +02:00
|
|
|
dump_completed_IO(inode, head);
|
|
|
|
list_replace_init(head, &unwritten);
|
2012-09-29 06:14:55 +02:00
|
|
|
spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
|
|
|
|
|
|
|
|
while (!list_empty(&unwritten)) {
|
|
|
|
io = list_entry(unwritten.next, ext4_io_end_t, list);
|
|
|
|
BUG_ON(!(io->flag & EXT4_IO_END_UNWRITTEN));
|
|
|
|
list_del_init(&io->list);
|
|
|
|
|
|
|
|
err = ext4_end_io(io);
|
|
|
|
if (unlikely(!ret && err))
|
|
|
|
ret = err;
|
|
|
|
}
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2013-06-04 20:21:02 +02:00
|
|
|
* work on completed IO, to convert unwritten extents to extents
|
2012-09-29 06:14:55 +02:00
|
|
|
*/
|
2013-06-04 20:21:02 +02:00
|
|
|
void ext4_end_io_rsv_work(struct work_struct *work)
|
|
|
|
{
|
|
|
|
struct ext4_inode_info *ei = container_of(work, struct ext4_inode_info,
|
|
|
|
i_rsv_conversion_work);
|
|
|
|
ext4_do_flush_completed_IO(&ei->vfs_inode, &ei->i_rsv_conversion_list);
|
|
|
|
}
|
|
|
|
|
2010-10-28 03:30:10 +02:00
|
|
|
ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags)
|
|
|
|
{
|
2010-12-20 03:41:55 +01:00
|
|
|
ext4_io_end_t *io = kmem_cache_zalloc(io_end_cachep, flags);
|
2010-10-28 03:30:10 +02:00
|
|
|
if (io) {
|
2010-11-08 19:43:33 +01:00
|
|
|
atomic_inc(&EXT4_I(inode)->i_ioend_count);
|
|
|
|
io->inode = inode;
|
2010-10-28 03:30:10 +02:00
|
|
|
INIT_LIST_HEAD(&io->list);
|
2013-06-04 17:58:58 +02:00
|
|
|
atomic_set(&io->count, 1);
|
2010-10-28 03:30:10 +02:00
|
|
|
}
|
|
|
|
return io;
|
|
|
|
}
|
|
|
|
|
2013-06-04 17:58:58 +02:00
|
|
|
void ext4_put_io_end_defer(ext4_io_end_t *io_end)
|
|
|
|
{
|
|
|
|
if (atomic_dec_and_test(&io_end->count)) {
|
|
|
|
if (!(io_end->flag & EXT4_IO_END_UNWRITTEN) || !io_end->size) {
|
|
|
|
ext4_release_io_end(io_end);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
ext4_add_complete_io(io_end);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
int ext4_put_io_end(ext4_io_end_t *io_end)
|
|
|
|
{
|
|
|
|
int err = 0;
|
|
|
|
|
|
|
|
if (atomic_dec_and_test(&io_end->count)) {
|
|
|
|
if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
|
2013-06-04 19:21:11 +02:00
|
|
|
err = ext4_convert_unwritten_extents(io_end->handle,
|
|
|
|
io_end->inode, io_end->offset,
|
|
|
|
io_end->size);
|
|
|
|
io_end->handle = NULL;
|
2013-06-04 17:58:58 +02:00
|
|
|
ext4_clear_io_unwritten_flag(io_end);
|
|
|
|
}
|
|
|
|
ext4_release_io_end(io_end);
|
|
|
|
}
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
|
|
|
ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end)
|
|
|
|
{
|
|
|
|
atomic_inc(&io_end->count);
|
|
|
|
return io_end;
|
|
|
|
}
|
|
|
|
|
2013-07-11 03:31:04 +02:00
|
|
|
/* BIO completion function for page writeback */
|
2010-10-28 03:30:10 +02:00
|
|
|
static void ext4_end_bio(struct bio *bio, int error)
|
|
|
|
{
|
|
|
|
ext4_io_end_t *io_end = bio->bi_private;
|
2013-10-12 00:44:27 +02:00
|
|
|
sector_t bi_sector = bio->bi_iter.bi_sector;
|
2010-10-28 03:30:10 +02:00
|
|
|
|
|
|
|
BUG_ON(!io_end);
|
|
|
|
bio->bi_end_io = NULL;
|
|
|
|
if (test_bit(BIO_UPTODATE, &bio->bi_flags))
|
|
|
|
error = 0;
|
2013-04-12 05:48:32 +02:00
|
|
|
|
2010-11-08 19:43:33 +01:00
|
|
|
if (error) {
|
2013-06-04 20:23:41 +02:00
|
|
|
struct inode *inode = io_end->inode;
|
|
|
|
|
2014-04-07 16:54:20 +02:00
|
|
|
ext4_warning(inode->i_sb, "I/O error %d writing to inode %lu "
|
2010-11-08 19:43:33 +01:00
|
|
|
"(offset %llu size %ld starting block %llu)",
|
2014-04-07 16:54:20 +02:00
|
|
|
error, inode->i_ino,
|
2010-11-08 19:43:33 +01:00
|
|
|
(unsigned long long) io_end->offset,
|
|
|
|
(long) io_end->size,
|
|
|
|
(unsigned long long)
|
ext4: Fix data corruption with multi-block writepages support
This fixes a corruption problem with the multi-block
writepages submittal change for ext4, from commit
bd2d0210cf22f2bd0cef72eb97cf94fc7d31d8cc ("ext4: use bio
layer instead of buffer layer in mpage_da_submit_io").
(Note that this corruption is not present in 2.6.37 on
ext4, because the corruption was detected after the
feature was merged in 2.6.37-rc1, and so it was turned
off by adding a non-default mount option,
mblk_io_submit. With this commit, which hopefully
fixes the last of the bugs with this feature, we'll be
able to turn on this performance feature by default in
2.6.38, and remove the mblk_io_submit option.)
The ext4 code path to bundle multiple pages for
writeback in ext4_bio_write_page() had a bug: we should
be clearing buffer head dirty flags *before* we submit
the bio, not in the completion routine.
The patch below was tested on 2.6.37 under KVM with the
postgresql script which was submitted by Jon Nelson as
documented in commit 1449032be1.
Without the patch, I'd hit the corruption problem about
50-70% of the time. With the patch, I executed the
script > 100 times with no corruption seen.
I also fixed a bug to make sure ext4_end_bio() doesn't
dereference the bio after the bio_put() call.
Reported-by: Jon Nelson <jnelson@jamponi.net>
Reported-by: Matthias Bayer <jackdachef@gmail.com>
Signed-off-by: Curt Wohlgemuth <curtw@google.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Cc: stable@kernel.org
2011-02-07 18:46:14 +01:00
|
|
|
bi_sector >> (inode->i_blkbits - 9));
|
2014-04-07 16:54:20 +02:00
|
|
|
mapping_set_error(inode->i_mapping, error);
|
2010-11-08 19:43:33 +01:00
|
|
|
}
|
2013-07-11 03:31:04 +02:00
|
|
|
|
|
|
|
if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
|
|
|
|
/*
|
|
|
|
* Link bio into list hanging from io_end. We have to do it
|
|
|
|
* atomically as bio completions can be racing against each
|
|
|
|
* other.
|
|
|
|
*/
|
|
|
|
bio->bi_private = xchg(&io_end->bio, bio);
|
|
|
|
ext4_put_io_end_defer(io_end);
|
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* Drop io_end reference early. Inode can get freed once
|
|
|
|
* we finish the bio.
|
|
|
|
*/
|
|
|
|
ext4_put_io_end_defer(io_end);
|
|
|
|
ext4_finish_bio(bio);
|
|
|
|
bio_put(bio);
|
|
|
|
}
|
2010-10-28 03:30:10 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
void ext4_io_submit(struct ext4_io_submit *io)
|
|
|
|
{
|
|
|
|
struct bio *bio = io->io_bio;
|
|
|
|
|
|
|
|
if (bio) {
|
|
|
|
bio_get(io->io_bio);
|
|
|
|
submit_bio(io->io_op, io->io_bio);
|
|
|
|
BUG_ON(bio_flagged(io->io_bio, BIO_EOPNOTSUPP));
|
|
|
|
bio_put(io->io_bio);
|
|
|
|
}
|
2011-02-22 03:01:42 +01:00
|
|
|
io->io_bio = NULL;
|
2013-06-04 17:58:58 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
void ext4_io_submit_init(struct ext4_io_submit *io,
|
|
|
|
struct writeback_control *wbc)
|
|
|
|
{
|
|
|
|
io->io_op = (wbc->sync_mode == WB_SYNC_ALL ? WRITE_SYNC : WRITE);
|
|
|
|
io->io_bio = NULL;
|
2011-02-22 03:01:42 +01:00
|
|
|
io->io_end = NULL;
|
2010-10-28 03:30:10 +02:00
|
|
|
}
|
|
|
|
|
2013-06-04 17:58:58 +02:00
|
|
|
static int io_submit_init_bio(struct ext4_io_submit *io,
|
|
|
|
struct buffer_head *bh)
|
2010-10-28 03:30:10 +02:00
|
|
|
{
|
|
|
|
int nvecs = bio_get_nr_vecs(bh->b_bdev);
|
|
|
|
struct bio *bio;
|
|
|
|
|
2011-06-30 03:44:45 +02:00
|
|
|
bio = bio_alloc(GFP_NOIO, min(nvecs, BIO_MAX_PAGES));
|
2013-06-06 16:18:22 +02:00
|
|
|
if (!bio)
|
|
|
|
return -ENOMEM;
|
2013-10-12 00:44:27 +02:00
|
|
|
bio->bi_iter.bi_sector = bh->b_blocknr * (bh->b_size >> 9);
|
2010-10-28 03:30:10 +02:00
|
|
|
bio->bi_bdev = bh->b_bdev;
|
|
|
|
bio->bi_end_io = ext4_end_bio;
|
2013-06-04 17:58:58 +02:00
|
|
|
bio->bi_private = ext4_get_io_end(io->io_end);
|
2010-10-28 03:30:10 +02:00
|
|
|
io->io_bio = bio;
|
|
|
|
io->io_next_block = bh->b_blocknr;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int io_submit_add_bh(struct ext4_io_submit *io,
|
|
|
|
struct inode *inode,
|
|
|
|
struct buffer_head *bh)
|
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
if (io->io_bio && bh->b_blocknr != io->io_next_block) {
|
|
|
|
submit_and_retry:
|
|
|
|
ext4_io_submit(io);
|
|
|
|
}
|
|
|
|
if (io->io_bio == NULL) {
|
2013-06-04 17:58:58 +02:00
|
|
|
ret = io_submit_init_bio(io, bh);
|
2010-10-28 03:30:10 +02:00
|
|
|
if (ret)
|
|
|
|
return ret;
|
|
|
|
}
|
2013-06-04 17:58:58 +02:00
|
|
|
ret = bio_add_page(io->io_bio, bh->b_page, bh->b_size, bh_offset(bh));
|
|
|
|
if (ret != bh->b_size)
|
|
|
|
goto submit_and_retry;
|
2010-10-28 03:30:10 +02:00
|
|
|
io->io_next_block++;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
int ext4_bio_write_page(struct ext4_io_submit *io,
|
|
|
|
struct page *page,
|
|
|
|
int len,
|
2014-05-12 14:12:25 +02:00
|
|
|
struct writeback_control *wbc,
|
|
|
|
bool keep_towrite)
|
2010-10-28 03:30:10 +02:00
|
|
|
{
|
|
|
|
struct inode *inode = page->mapping->host;
|
2013-04-12 05:48:32 +02:00
|
|
|
unsigned block_start, blocksize;
|
2010-10-28 03:30:10 +02:00
|
|
|
struct buffer_head *bh, *head;
|
|
|
|
int ret = 0;
|
2013-04-12 05:48:32 +02:00
|
|
|
int nr_submitted = 0;
|
2010-10-28 03:30:10 +02:00
|
|
|
|
|
|
|
blocksize = 1 << inode->i_blkbits;
|
|
|
|
|
ext4: Fix data corruption with multi-block writepages support
This fixes a corruption problem with the multi-block
writepages submittal change for ext4, from commit
bd2d0210cf22f2bd0cef72eb97cf94fc7d31d8cc ("ext4: use bio
layer instead of buffer layer in mpage_da_submit_io").
(Note that this corruption is not present in 2.6.37 on
ext4, because the corruption was detected after the
feature was merged in 2.6.37-rc1, and so it was turned
off by adding a non-default mount option,
mblk_io_submit. With this commit, which hopefully
fixes the last of the bugs with this feature, we'll be
able to turn on this performance feature by default in
2.6.38, and remove the mblk_io_submit option.)
The ext4 code path to bundle multiple pages for
writeback in ext4_bio_write_page() had a bug: we should
be clearing buffer head dirty flags *before* we submit
the bio, not in the completion routine.
The patch below was tested on 2.6.37 under KVM with the
postgresql script which was submitted by Jon Nelson as
documented in commit 1449032be1.
Without the patch, I'd hit the corruption problem about
50-70% of the time. With the patch, I executed the
script > 100 times with no corruption seen.
I also fixed a bug to make sure ext4_end_bio() doesn't
dereference the bio after the bio_put() call.
Reported-by: Jon Nelson <jnelson@jamponi.net>
Reported-by: Matthias Bayer <jackdachef@gmail.com>
Signed-off-by: Curt Wohlgemuth <curtw@google.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Cc: stable@kernel.org
2011-02-07 18:46:14 +01:00
|
|
|
BUG_ON(!PageLocked(page));
|
2010-10-28 03:30:10 +02:00
|
|
|
BUG_ON(PageWriteback(page));
|
|
|
|
|
2014-05-12 14:12:25 +02:00
|
|
|
if (keep_towrite)
|
|
|
|
set_page_writeback_keepwrite(page);
|
|
|
|
else
|
|
|
|
set_page_writeback(page);
|
2011-02-27 22:43:24 +01:00
|
|
|
ClearPageError(page);
|
2010-10-28 03:30:10 +02:00
|
|
|
|
2014-05-27 18:48:55 +02:00
|
|
|
/*
|
|
|
|
* Comments copied from block_write_full_page_endio:
|
|
|
|
*
|
|
|
|
* The page straddles i_size. It must be zeroed out on each and every
|
|
|
|
* writepage invocation because it may be mmapped. "A file is mapped
|
|
|
|
* in multiples of the page size. For a file that is not a multiple of
|
|
|
|
* the page size, the remaining memory is zeroed when mapped, and
|
|
|
|
* writes to that region are not written out to the file."
|
|
|
|
*/
|
|
|
|
if (len < PAGE_CACHE_SIZE)
|
|
|
|
zero_user_segment(page, len, PAGE_CACHE_SIZE);
|
2013-04-12 05:48:32 +02:00
|
|
|
/*
|
|
|
|
* In the first loop we prepare and mark buffers to submit. We have to
|
|
|
|
* mark all buffers in the page before submitting so that
|
|
|
|
* end_page_writeback() cannot be called from ext4_bio_end_io() when IO
|
|
|
|
* on the first buffer finishes and we are still working on submitting
|
|
|
|
* the second buffer.
|
|
|
|
*/
|
|
|
|
bh = head = page_buffers(page);
|
|
|
|
do {
|
|
|
|
block_start = bh_offset(bh);
|
2010-10-28 03:30:10 +02:00
|
|
|
if (block_start >= len) {
|
|
|
|
clear_buffer_dirty(bh);
|
|
|
|
set_buffer_uptodate(bh);
|
|
|
|
continue;
|
|
|
|
}
|
2013-01-29 02:53:28 +01:00
|
|
|
if (!buffer_dirty(bh) || buffer_delay(bh) ||
|
|
|
|
!buffer_mapped(bh) || buffer_unwritten(bh)) {
|
|
|
|
/* A hole? We can safely clear the dirty bit */
|
|
|
|
if (!buffer_mapped(bh))
|
|
|
|
clear_buffer_dirty(bh);
|
|
|
|
if (io->io_bio)
|
|
|
|
ext4_io_submit(io);
|
|
|
|
continue;
|
|
|
|
}
|
2013-04-12 05:48:32 +02:00
|
|
|
if (buffer_new(bh)) {
|
|
|
|
clear_buffer_new(bh);
|
|
|
|
unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr);
|
|
|
|
}
|
|
|
|
set_buffer_async_write(bh);
|
|
|
|
} while ((bh = bh->b_this_page) != head);
|
|
|
|
|
|
|
|
/* Now submit buffers to write */
|
|
|
|
bh = head = page_buffers(page);
|
|
|
|
do {
|
|
|
|
if (!buffer_async_write(bh))
|
|
|
|
continue;
|
2013-06-04 17:58:58 +02:00
|
|
|
ret = io_submit_add_bh(io, inode, bh);
|
2010-10-28 03:30:10 +02:00
|
|
|
if (ret) {
|
|
|
|
/*
|
|
|
|
* We only get here on ENOMEM. Not much else
|
|
|
|
* we can do but mark the page as dirty, and
|
|
|
|
* better luck next time.
|
|
|
|
*/
|
2013-01-28 15:32:54 +01:00
|
|
|
redirty_page_for_writepage(wbc, page);
|
2010-10-28 03:30:10 +02:00
|
|
|
break;
|
|
|
|
}
|
2013-04-12 05:48:32 +02:00
|
|
|
nr_submitted++;
|
2013-01-28 15:32:54 +01:00
|
|
|
clear_buffer_dirty(bh);
|
2013-04-12 05:48:32 +02:00
|
|
|
} while ((bh = bh->b_this_page) != head);
|
|
|
|
|
|
|
|
/* Error stopped previous loop? Clean up buffers... */
|
|
|
|
if (ret) {
|
|
|
|
do {
|
|
|
|
clear_buffer_async_write(bh);
|
|
|
|
bh = bh->b_this_page;
|
|
|
|
} while (bh != head);
|
2010-10-28 03:30:10 +02:00
|
|
|
}
|
|
|
|
unlock_page(page);
|
2013-04-12 05:48:32 +02:00
|
|
|
/* Nothing submitted - we have to end page writeback */
|
|
|
|
if (!nr_submitted)
|
|
|
|
end_page_writeback(page);
|
2010-10-28 03:30:10 +02:00
|
|
|
return ret;
|
|
|
|
}
|