kernel_optimize_test/fs/ext4/page-io.c

433 lines
11 KiB
C
Raw Normal View History

/*
* linux/fs/ext4/page-io.c
*
* This contains the new page_io functions for ext4
*
* Written by Theodore Ts'o, 2010.
*/
#include <linux/module.h>
#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>
#include <linux/uio.h>
#include <linux/bio.h>
#include <linux/workqueue.h>
#include <linux/kernel.h>
#include <linux/slab.h>
#include "ext4_jbd2.h"
#include "xattr.h"
#include "acl.h"
#include "ext4_extents.h"
static struct kmem_cache *io_page_cachep, *io_end_cachep;
int __init ext4_init_pageio(void)
{
io_page_cachep = KMEM_CACHE(ext4_io_page, SLAB_RECLAIM_ACCOUNT);
if (io_page_cachep == NULL)
return -ENOMEM;
io_end_cachep = KMEM_CACHE(ext4_io_end, SLAB_RECLAIM_ACCOUNT);
if (io_end_cachep == NULL) {
kmem_cache_destroy(io_page_cachep);
return -ENOMEM;
}
return 0;
}
void ext4_exit_pageio(void)
{
kmem_cache_destroy(io_end_cachep);
kmem_cache_destroy(io_page_cachep);
}
ext4: handle writeback of inodes which are being freed The following BUG can occur when an inode which is getting freed when it still has dirty pages outstanding, and it gets deleted (in this because it was the target of a rename). In ordered mode, we need to make sure the data pages are written just in case we crash before the rename (or unlink) is committed. If the inode is being freed then when we try to igrab the inode, we end up tripping the BUG_ON at fs/ext4/page-io.c:146. To solve this problem, we need to keep track of the number of io callbacks which are pending, and avoid destroying the inode until they have all been completed. That way we don't have to bump the inode count to keep the inode from being destroyed; an approach which doesn't work because the count could have already been dropped down to zero before the inode writeback has started (at which point we're not allowed to bump the count back up to 1, since it's already started getting freed). Thanks to Dave Chinner for suggesting this approach, which is also used by XFS. kernel BUG at /scratch_space/linux-2.6/fs/ext4/page-io.c:146! Call Trace: [<ffffffff811075b1>] ext4_bio_write_page+0x172/0x307 [<ffffffff811033a7>] mpage_da_submit_io+0x2f9/0x37b [<ffffffff811068d7>] mpage_da_map_and_submit+0x2cc/0x2e2 [<ffffffff811069b3>] mpage_add_bh_to_extent+0xc6/0xd5 [<ffffffff81106c66>] write_cache_pages_da+0x2a4/0x3ac [<ffffffff81107044>] ext4_da_writepages+0x2d6/0x44d [<ffffffff81087910>] do_writepages+0x1c/0x25 [<ffffffff810810a4>] __filemap_fdatawrite_range+0x4b/0x4d [<ffffffff810815f5>] filemap_fdatawrite_range+0xe/0x10 [<ffffffff81122a2e>] jbd2_journal_begin_ordered_truncate+0x7b/0xa2 [<ffffffff8110615d>] ext4_evict_inode+0x57/0x24c [<ffffffff810c14a3>] evict+0x22/0x92 [<ffffffff810c1a3d>] iput+0x212/0x249 [<ffffffff810bdf16>] dentry_iput+0xa1/0xb9 [<ffffffff810bdf6b>] d_kill+0x3d/0x5d [<ffffffff810be613>] dput+0x13a/0x147 [<ffffffff810b990d>] sys_renameat+0x1b5/0x258 [<ffffffff81145f71>] ? _atomic_dec_and_lock+0x2d/0x4c [<ffffffff810b2950>] ? cp_new_stat+0xde/0xea [<ffffffff810b29c1>] ? sys_newlstat+0x2d/0x38 [<ffffffff810b99c6>] sys_rename+0x16/0x18 [<ffffffff81002a2b>] system_call_fastpath+0x16/0x1b Reported-by: Nick Bowler <nbowler@elliptictech.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Tested-by: Nick Bowler <nbowler@elliptictech.com>
2010-11-09 02:43:33 +08:00
void ext4_ioend_wait(struct inode *inode)
{
ext4: serialize unaligned asynchronous DIO ext4 has a data corruption case when doing non-block-aligned asynchronous direct IO into a sparse file, as demonstrated by xfstest 240. The root cause is that while ext4 preallocates space in the hole, mappings of that space still look "new" and dio_zero_block() will zero out the unwritten portions. When more than one AIO thread is going, they both find this "new" block and race to zero out their portion; this is uncoordinated and causes data corruption. Dave Chinner fixed this for xfs by simply serializing all unaligned asynchronous direct IO. I've done the same here. The difference is that we only wait on conversions, not all IO. This is a very big hammer, and I'm not very pleased with stuffing this into ext4_file_write(). But since ext4 is DIO_LOCKING, we need to serialize it at this high level. I tried to move this into ext4_ext_direct_IO, but by then we have the i_mutex already, and we will wait on the work queue to do conversions - which must also take the i_mutex. So that won't work. This was originally exposed by qemu-kvm installing to a raw disk image with a normal sector-63 alignment. I've tested a backport of this patch with qemu, and it does avoid the corruption. It is also quite a lot slower (14 min for package installs, vs. 8 min for well-aligned) but I'll take slow correctness over fast corruption any day. Mingming suggested that we can track outstanding conversions, and wait on those so that non-sparse files won't be affected, and I've implemented that here; unaligned AIO to nonsparse files won't take a perf hit. [tytso@mit.edu: Keep the mutex as a hashed array instead of bloating the ext4 inode] [tytso@mit.edu: Fix up namespace issues so that global variables are protected with an "ext4_" prefix.] Signed-off-by: Eric Sandeen <sandeen@redhat.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-02-12 21:17:34 +08:00
wait_queue_head_t *wq = ext4_ioend_wq(inode);
ext4: handle writeback of inodes which are being freed The following BUG can occur when an inode which is getting freed when it still has dirty pages outstanding, and it gets deleted (in this because it was the target of a rename). In ordered mode, we need to make sure the data pages are written just in case we crash before the rename (or unlink) is committed. If the inode is being freed then when we try to igrab the inode, we end up tripping the BUG_ON at fs/ext4/page-io.c:146. To solve this problem, we need to keep track of the number of io callbacks which are pending, and avoid destroying the inode until they have all been completed. That way we don't have to bump the inode count to keep the inode from being destroyed; an approach which doesn't work because the count could have already been dropped down to zero before the inode writeback has started (at which point we're not allowed to bump the count back up to 1, since it's already started getting freed). Thanks to Dave Chinner for suggesting this approach, which is also used by XFS. kernel BUG at /scratch_space/linux-2.6/fs/ext4/page-io.c:146! Call Trace: [<ffffffff811075b1>] ext4_bio_write_page+0x172/0x307 [<ffffffff811033a7>] mpage_da_submit_io+0x2f9/0x37b [<ffffffff811068d7>] mpage_da_map_and_submit+0x2cc/0x2e2 [<ffffffff811069b3>] mpage_add_bh_to_extent+0xc6/0xd5 [<ffffffff81106c66>] write_cache_pages_da+0x2a4/0x3ac [<ffffffff81107044>] ext4_da_writepages+0x2d6/0x44d [<ffffffff81087910>] do_writepages+0x1c/0x25 [<ffffffff810810a4>] __filemap_fdatawrite_range+0x4b/0x4d [<ffffffff810815f5>] filemap_fdatawrite_range+0xe/0x10 [<ffffffff81122a2e>] jbd2_journal_begin_ordered_truncate+0x7b/0xa2 [<ffffffff8110615d>] ext4_evict_inode+0x57/0x24c [<ffffffff810c14a3>] evict+0x22/0x92 [<ffffffff810c1a3d>] iput+0x212/0x249 [<ffffffff810bdf16>] dentry_iput+0xa1/0xb9 [<ffffffff810bdf6b>] d_kill+0x3d/0x5d [<ffffffff810be613>] dput+0x13a/0x147 [<ffffffff810b990d>] sys_renameat+0x1b5/0x258 [<ffffffff81145f71>] ? _atomic_dec_and_lock+0x2d/0x4c [<ffffffff810b2950>] ? cp_new_stat+0xde/0xea [<ffffffff810b29c1>] ? sys_newlstat+0x2d/0x38 [<ffffffff810b99c6>] sys_rename+0x16/0x18 [<ffffffff81002a2b>] system_call_fastpath+0x16/0x1b Reported-by: Nick Bowler <nbowler@elliptictech.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Tested-by: Nick Bowler <nbowler@elliptictech.com>
2010-11-09 02:43:33 +08:00
wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_ioend_count) == 0));
}
static void put_io_page(struct ext4_io_page *io_page)
{
if (atomic_dec_and_test(&io_page->p_count)) {
end_page_writeback(io_page->p_page);
put_page(io_page->p_page);
kmem_cache_free(io_page_cachep, io_page);
}
}
void ext4_free_io_end(ext4_io_end_t *io)
{
int i;
BUG_ON(!io);
if (io->page)
put_page(io->page);
for (i = 0; i < io->num_io_pages; i++)
put_io_page(io->pages[i]);
io->num_io_pages = 0;
if (atomic_dec_and_test(&EXT4_I(io->inode)->i_ioend_count))
wake_up_all(ext4_ioend_wq(io->inode));
kmem_cache_free(io_end_cachep, io);
}
/*
* check a range of space and convert unwritten extents to written.
*
* Called with inode->i_mutex; we depend on this when we manipulate
* io->flag, since we could otherwise race with ext4_flush_completed_IO()
*/
int ext4_end_io_nolock(ext4_io_end_t *io)
{
struct inode *inode = io->inode;
loff_t offset = io->offset;
ssize_t size = io->size;
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);
if (!(io->flag & EXT4_IO_END_UNWRITTEN))
return ret;
ret = ext4_convert_unwritten_extents(inode, offset, size);
if (ret < 0) {
printk(KERN_EMERG "%s: failed to convert unwritten "
"extents to written extents, error is %d "
"io is still on inode %lu aio dio list\n",
__func__, ret, inode->i_ino);
return ret;
}
if (io->iocb)
aio_complete(io->iocb, io->result, 0);
/* clear the DIO AIO unwritten flag */
ext4: serialize unaligned asynchronous DIO ext4 has a data corruption case when doing non-block-aligned asynchronous direct IO into a sparse file, as demonstrated by xfstest 240. The root cause is that while ext4 preallocates space in the hole, mappings of that space still look "new" and dio_zero_block() will zero out the unwritten portions. When more than one AIO thread is going, they both find this "new" block and race to zero out their portion; this is uncoordinated and causes data corruption. Dave Chinner fixed this for xfs by simply serializing all unaligned asynchronous direct IO. I've done the same here. The difference is that we only wait on conversions, not all IO. This is a very big hammer, and I'm not very pleased with stuffing this into ext4_file_write(). But since ext4 is DIO_LOCKING, we need to serialize it at this high level. I tried to move this into ext4_ext_direct_IO, but by then we have the i_mutex already, and we will wait on the work queue to do conversions - which must also take the i_mutex. So that won't work. This was originally exposed by qemu-kvm installing to a raw disk image with a normal sector-63 alignment. I've tested a backport of this patch with qemu, and it does avoid the corruption. It is also quite a lot slower (14 min for package installs, vs. 8 min for well-aligned) but I'll take slow correctness over fast corruption any day. Mingming suggested that we can track outstanding conversions, and wait on those so that non-sparse files won't be affected, and I've implemented that here; unaligned AIO to nonsparse files won't take a perf hit. [tytso@mit.edu: Keep the mutex as a hashed array instead of bloating the ext4 inode] [tytso@mit.edu: Fix up namespace issues so that global variables are protected with an "ext4_" prefix.] Signed-off-by: Eric Sandeen <sandeen@redhat.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-02-12 21:17:34 +08:00
if (io->flag & EXT4_IO_END_UNWRITTEN) {
io->flag &= ~EXT4_IO_END_UNWRITTEN;
/* Wake up anyone waiting on unwritten extent conversion */
if (atomic_dec_and_test(&EXT4_I(inode)->i_aiodio_unwritten))
wake_up_all(ext4_ioend_wq(io->inode));
ext4: serialize unaligned asynchronous DIO ext4 has a data corruption case when doing non-block-aligned asynchronous direct IO into a sparse file, as demonstrated by xfstest 240. The root cause is that while ext4 preallocates space in the hole, mappings of that space still look "new" and dio_zero_block() will zero out the unwritten portions. When more than one AIO thread is going, they both find this "new" block and race to zero out their portion; this is uncoordinated and causes data corruption. Dave Chinner fixed this for xfs by simply serializing all unaligned asynchronous direct IO. I've done the same here. The difference is that we only wait on conversions, not all IO. This is a very big hammer, and I'm not very pleased with stuffing this into ext4_file_write(). But since ext4 is DIO_LOCKING, we need to serialize it at this high level. I tried to move this into ext4_ext_direct_IO, but by then we have the i_mutex already, and we will wait on the work queue to do conversions - which must also take the i_mutex. So that won't work. This was originally exposed by qemu-kvm installing to a raw disk image with a normal sector-63 alignment. I've tested a backport of this patch with qemu, and it does avoid the corruption. It is also quite a lot slower (14 min for package installs, vs. 8 min for well-aligned) but I'll take slow correctness over fast corruption any day. Mingming suggested that we can track outstanding conversions, and wait on those so that non-sparse files won't be affected, and I've implemented that here; unaligned AIO to nonsparse files won't take a perf hit. [tytso@mit.edu: Keep the mutex as a hashed array instead of bloating the ext4 inode] [tytso@mit.edu: Fix up namespace issues so that global variables are protected with an "ext4_" prefix.] Signed-off-by: Eric Sandeen <sandeen@redhat.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-02-12 21:17:34 +08:00
}
return ret;
}
/*
* work on completed aio dio IO, to convert unwritten extents to extents
*/
static void ext4_end_io_work(struct work_struct *work)
{
ext4_io_end_t *io = container_of(work, ext4_io_end_t, work);
struct inode *inode = io->inode;
struct ext4_inode_info *ei = EXT4_I(inode);
unsigned long flags;
int ret;
spin_lock_irqsave(&ei->i_completed_io_lock, flags);
if (list_empty(&io->list)) {
spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
goto free;
}
spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
ext4: remove i_mutex lock in ext4_evict_inode to fix lockdep complaining The i_mutex lock and flush_completed_IO() added by commit 2581fdc810 in ext4_evict_inode() causes lockdep complaining about potential deadlock in several places. In most/all of these LOCKDEP complaints it looks like it's a false positive, since many of the potential circular locking cases can't take place by the time the ext4_evict_inode() is called; but since at the very least it may mask real problems, we need to address this. This change removes the flush_completed_IO() and i_mutex lock in ext4_evict_inode(). Instead, we take a different approach to resolve the software lockup that commit 2581fdc810 intends to fix. Rather than having ext4-dio-unwritten thread wait for grabing the i_mutex lock of an inode, we use mutex_trylock() instead, and simply requeue the work item if we fail to grab the inode's i_mutex lock. This should speed up work queue processing in general and also prevents the following deadlock scenario: During page fault, shrink_icache_memory is called that in turn evicts another inode B. Inode B has some pending io_end work so it calls ext4_ioend_wait() that waits for inode B's i_ioend_count to become zero. However, inode B's ioend work was queued behind some of inode A's ioend work on the same cpu's ext4-dio-unwritten workqueue. As the ext4-dio-unwritten thread on that cpu is processing inode A's ioend work, it tries to grab inode A's i_mutex lock. Since the i_mutex lock of inode A is still hold before the page fault happened, we enter a deadlock. Signed-off-by: Jiaying Zhang <jiayingz@google.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-08-31 23:50:51 +08:00
if (!mutex_trylock(&inode->i_mutex)) {
/*
* Requeue the work instead of waiting so that the work
* items queued after this can be processed.
*/
queue_work(EXT4_SB(inode->i_sb)->dio_unwritten_wq, &io->work);
/*
* To prevent the ext4-dio-unwritten thread from keeping
* requeueing end_io requests and occupying cpu for too long,
* yield the cpu if it sees an end_io request that has already
* been requeued.
*/
if (io->flag & EXT4_IO_END_QUEUED)
yield();
io->flag |= EXT4_IO_END_QUEUED;
return;
}
ret = ext4_end_io_nolock(io);
if (ret < 0) {
mutex_unlock(&inode->i_mutex);
return;
}
spin_lock_irqsave(&ei->i_completed_io_lock, flags);
if (!list_empty(&io->list))
list_del_init(&io->list);
spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
mutex_unlock(&inode->i_mutex);
free:
ext4_free_io_end(io);
}
ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags)
{
ext4_io_end_t *io = kmem_cache_zalloc(io_end_cachep, flags);
if (io) {
ext4: handle writeback of inodes which are being freed The following BUG can occur when an inode which is getting freed when it still has dirty pages outstanding, and it gets deleted (in this because it was the target of a rename). In ordered mode, we need to make sure the data pages are written just in case we crash before the rename (or unlink) is committed. If the inode is being freed then when we try to igrab the inode, we end up tripping the BUG_ON at fs/ext4/page-io.c:146. To solve this problem, we need to keep track of the number of io callbacks which are pending, and avoid destroying the inode until they have all been completed. That way we don't have to bump the inode count to keep the inode from being destroyed; an approach which doesn't work because the count could have already been dropped down to zero before the inode writeback has started (at which point we're not allowed to bump the count back up to 1, since it's already started getting freed). Thanks to Dave Chinner for suggesting this approach, which is also used by XFS. kernel BUG at /scratch_space/linux-2.6/fs/ext4/page-io.c:146! Call Trace: [<ffffffff811075b1>] ext4_bio_write_page+0x172/0x307 [<ffffffff811033a7>] mpage_da_submit_io+0x2f9/0x37b [<ffffffff811068d7>] mpage_da_map_and_submit+0x2cc/0x2e2 [<ffffffff811069b3>] mpage_add_bh_to_extent+0xc6/0xd5 [<ffffffff81106c66>] write_cache_pages_da+0x2a4/0x3ac [<ffffffff81107044>] ext4_da_writepages+0x2d6/0x44d [<ffffffff81087910>] do_writepages+0x1c/0x25 [<ffffffff810810a4>] __filemap_fdatawrite_range+0x4b/0x4d [<ffffffff810815f5>] filemap_fdatawrite_range+0xe/0x10 [<ffffffff81122a2e>] jbd2_journal_begin_ordered_truncate+0x7b/0xa2 [<ffffffff8110615d>] ext4_evict_inode+0x57/0x24c [<ffffffff810c14a3>] evict+0x22/0x92 [<ffffffff810c1a3d>] iput+0x212/0x249 [<ffffffff810bdf16>] dentry_iput+0xa1/0xb9 [<ffffffff810bdf6b>] d_kill+0x3d/0x5d [<ffffffff810be613>] dput+0x13a/0x147 [<ffffffff810b990d>] sys_renameat+0x1b5/0x258 [<ffffffff81145f71>] ? _atomic_dec_and_lock+0x2d/0x4c [<ffffffff810b2950>] ? cp_new_stat+0xde/0xea [<ffffffff810b29c1>] ? sys_newlstat+0x2d/0x38 [<ffffffff810b99c6>] sys_rename+0x16/0x18 [<ffffffff81002a2b>] system_call_fastpath+0x16/0x1b Reported-by: Nick Bowler <nbowler@elliptictech.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Tested-by: Nick Bowler <nbowler@elliptictech.com>
2010-11-09 02:43:33 +08:00
atomic_inc(&EXT4_I(inode)->i_ioend_count);
io->inode = inode;
INIT_WORK(&io->work, ext4_end_io_work);
INIT_LIST_HEAD(&io->list);
}
return io;
}
/*
* 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];
printk(KERN_ERR "Buffer I/O error on device %s, logical block %llu\n",
bdevname(bh->b_bdev, b),
(unsigned long long)bh->b_blocknr);
}
static void ext4_end_bio(struct bio *bio, int error)
{
ext4_io_end_t *io_end = bio->bi_private;
struct workqueue_struct *wq;
struct inode *inode;
unsigned long flags;
int i;
2011-02-08 01:46:14 +08:00
sector_t bi_sector = bio->bi_sector;
BUG_ON(!io_end);
bio->bi_private = NULL;
bio->bi_end_io = NULL;
if (test_bit(BIO_UPTODATE, &bio->bi_flags))
error = 0;
bio_put(bio);
for (i = 0; i < io_end->num_io_pages; i++) {
struct page *page = io_end->pages[i]->p_page;
struct buffer_head *bh, *head;
loff_t offset;
loff_t io_end_offset;
if (error) {
SetPageError(page);
set_bit(AS_EIO, &page->mapping->flags);
head = page_buffers(page);
BUG_ON(!head);
io_end_offset = io_end->offset + io_end->size;
offset = (sector_t) page->index << PAGE_CACHE_SHIFT;
bh = head;
do {
if ((offset >= io_end->offset) &&
(offset+bh->b_size <= io_end_offset))
buffer_io_error(bh);
offset += bh->b_size;
bh = bh->b_this_page;
} while (bh != head);
}
put_io_page(io_end->pages[i]);
}
io_end->num_io_pages = 0;
ext4: handle writeback of inodes which are being freed The following BUG can occur when an inode which is getting freed when it still has dirty pages outstanding, and it gets deleted (in this because it was the target of a rename). In ordered mode, we need to make sure the data pages are written just in case we crash before the rename (or unlink) is committed. If the inode is being freed then when we try to igrab the inode, we end up tripping the BUG_ON at fs/ext4/page-io.c:146. To solve this problem, we need to keep track of the number of io callbacks which are pending, and avoid destroying the inode until they have all been completed. That way we don't have to bump the inode count to keep the inode from being destroyed; an approach which doesn't work because the count could have already been dropped down to zero before the inode writeback has started (at which point we're not allowed to bump the count back up to 1, since it's already started getting freed). Thanks to Dave Chinner for suggesting this approach, which is also used by XFS. kernel BUG at /scratch_space/linux-2.6/fs/ext4/page-io.c:146! Call Trace: [<ffffffff811075b1>] ext4_bio_write_page+0x172/0x307 [<ffffffff811033a7>] mpage_da_submit_io+0x2f9/0x37b [<ffffffff811068d7>] mpage_da_map_and_submit+0x2cc/0x2e2 [<ffffffff811069b3>] mpage_add_bh_to_extent+0xc6/0xd5 [<ffffffff81106c66>] write_cache_pages_da+0x2a4/0x3ac [<ffffffff81107044>] ext4_da_writepages+0x2d6/0x44d [<ffffffff81087910>] do_writepages+0x1c/0x25 [<ffffffff810810a4>] __filemap_fdatawrite_range+0x4b/0x4d [<ffffffff810815f5>] filemap_fdatawrite_range+0xe/0x10 [<ffffffff81122a2e>] jbd2_journal_begin_ordered_truncate+0x7b/0xa2 [<ffffffff8110615d>] ext4_evict_inode+0x57/0x24c [<ffffffff810c14a3>] evict+0x22/0x92 [<ffffffff810c1a3d>] iput+0x212/0x249 [<ffffffff810bdf16>] dentry_iput+0xa1/0xb9 [<ffffffff810bdf6b>] d_kill+0x3d/0x5d [<ffffffff810be613>] dput+0x13a/0x147 [<ffffffff810b990d>] sys_renameat+0x1b5/0x258 [<ffffffff81145f71>] ? _atomic_dec_and_lock+0x2d/0x4c [<ffffffff810b2950>] ? cp_new_stat+0xde/0xea [<ffffffff810b29c1>] ? sys_newlstat+0x2d/0x38 [<ffffffff810b99c6>] sys_rename+0x16/0x18 [<ffffffff81002a2b>] system_call_fastpath+0x16/0x1b Reported-by: Nick Bowler <nbowler@elliptictech.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Tested-by: Nick Bowler <nbowler@elliptictech.com>
2010-11-09 02:43:33 +08:00
inode = io_end->inode;
if (error) {
io_end->flag |= EXT4_IO_END_ERROR;
ext4_warning(inode->i_sb, "I/O error writing to inode %lu "
"(offset %llu size %ld starting block %llu)",
inode->i_ino,
(unsigned long long) io_end->offset,
(long) io_end->size,
(unsigned long long)
2011-02-08 01:46:14 +08:00
bi_sector >> (inode->i_blkbits - 9));
ext4: handle writeback of inodes which are being freed The following BUG can occur when an inode which is getting freed when it still has dirty pages outstanding, and it gets deleted (in this because it was the target of a rename). In ordered mode, we need to make sure the data pages are written just in case we crash before the rename (or unlink) is committed. If the inode is being freed then when we try to igrab the inode, we end up tripping the BUG_ON at fs/ext4/page-io.c:146. To solve this problem, we need to keep track of the number of io callbacks which are pending, and avoid destroying the inode until they have all been completed. That way we don't have to bump the inode count to keep the inode from being destroyed; an approach which doesn't work because the count could have already been dropped down to zero before the inode writeback has started (at which point we're not allowed to bump the count back up to 1, since it's already started getting freed). Thanks to Dave Chinner for suggesting this approach, which is also used by XFS. kernel BUG at /scratch_space/linux-2.6/fs/ext4/page-io.c:146! Call Trace: [<ffffffff811075b1>] ext4_bio_write_page+0x172/0x307 [<ffffffff811033a7>] mpage_da_submit_io+0x2f9/0x37b [<ffffffff811068d7>] mpage_da_map_and_submit+0x2cc/0x2e2 [<ffffffff811069b3>] mpage_add_bh_to_extent+0xc6/0xd5 [<ffffffff81106c66>] write_cache_pages_da+0x2a4/0x3ac [<ffffffff81107044>] ext4_da_writepages+0x2d6/0x44d [<ffffffff81087910>] do_writepages+0x1c/0x25 [<ffffffff810810a4>] __filemap_fdatawrite_range+0x4b/0x4d [<ffffffff810815f5>] filemap_fdatawrite_range+0xe/0x10 [<ffffffff81122a2e>] jbd2_journal_begin_ordered_truncate+0x7b/0xa2 [<ffffffff8110615d>] ext4_evict_inode+0x57/0x24c [<ffffffff810c14a3>] evict+0x22/0x92 [<ffffffff810c1a3d>] iput+0x212/0x249 [<ffffffff810bdf16>] dentry_iput+0xa1/0xb9 [<ffffffff810bdf6b>] d_kill+0x3d/0x5d [<ffffffff810be613>] dput+0x13a/0x147 [<ffffffff810b990d>] sys_renameat+0x1b5/0x258 [<ffffffff81145f71>] ? _atomic_dec_and_lock+0x2d/0x4c [<ffffffff810b2950>] ? cp_new_stat+0xde/0xea [<ffffffff810b29c1>] ? sys_newlstat+0x2d/0x38 [<ffffffff810b99c6>] sys_rename+0x16/0x18 [<ffffffff81002a2b>] system_call_fastpath+0x16/0x1b Reported-by: Nick Bowler <nbowler@elliptictech.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Tested-by: Nick Bowler <nbowler@elliptictech.com>
2010-11-09 02:43:33 +08:00
}
if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
ext4_free_io_end(io_end);
return;
}
/* Add the io_end to per-inode completed io list*/
spin_lock_irqsave(&EXT4_I(inode)->i_completed_io_lock, flags);
list_add_tail(&io_end->list, &EXT4_I(inode)->i_completed_io_list);
spin_unlock_irqrestore(&EXT4_I(inode)->i_completed_io_lock, flags);
wq = EXT4_SB(inode->i_sb)->dio_unwritten_wq;
/* queue the work to convert unwritten extents to written */
queue_work(wq, &io_end->work);
}
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);
}
io->io_bio = NULL;
io->io_op = 0;
io->io_end = NULL;
}
static int io_submit_init(struct ext4_io_submit *io,
struct inode *inode,
struct writeback_control *wbc,
struct buffer_head *bh)
{
ext4_io_end_t *io_end;
struct page *page = bh->b_page;
int nvecs = bio_get_nr_vecs(bh->b_bdev);
struct bio *bio;
io_end = ext4_init_io_end(inode, GFP_NOFS);
if (!io_end)
return -ENOMEM;
bio = bio_alloc(GFP_NOIO, min(nvecs, BIO_MAX_PAGES));
bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
bio->bi_bdev = bh->b_bdev;
bio->bi_private = io->io_end = io_end;
bio->bi_end_io = ext4_end_bio;
io_end->offset = (page->index << PAGE_CACHE_SHIFT) + bh_offset(bh);
io->io_bio = bio;
io->io_op = (wbc->sync_mode == WB_SYNC_ALL ? WRITE_SYNC : WRITE);
io->io_next_block = bh->b_blocknr;
return 0;
}
static int io_submit_add_bh(struct ext4_io_submit *io,
struct ext4_io_page *io_page,
struct inode *inode,
struct writeback_control *wbc,
struct buffer_head *bh)
{
ext4_io_end_t *io_end;
int ret;
if (buffer_new(bh)) {
clear_buffer_new(bh);
unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr);
}
if (!buffer_mapped(bh) || buffer_delay(bh)) {
if (!buffer_mapped(bh))
clear_buffer_dirty(bh);
if (io->io_bio)
ext4_io_submit(io);
return 0;
}
if (io->io_bio && bh->b_blocknr != io->io_next_block) {
submit_and_retry:
ext4_io_submit(io);
}
if (io->io_bio == NULL) {
ret = io_submit_init(io, inode, wbc, bh);
if (ret)
return ret;
}
io_end = io->io_end;
if ((io_end->num_io_pages >= MAX_IO_PAGES) &&
(io_end->pages[io_end->num_io_pages-1] != io_page))
goto submit_and_retry;
if (buffer_uninit(bh) && !(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
io_end->flag |= EXT4_IO_END_UNWRITTEN;
atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
}
io->io_end->size += bh->b_size;
io->io_next_block++;
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;
if ((io_end->num_io_pages == 0) ||
(io_end->pages[io_end->num_io_pages-1] != io_page)) {
io_end->pages[io_end->num_io_pages++] = io_page;
atomic_inc(&io_page->p_count);
}
return 0;
}
int ext4_bio_write_page(struct ext4_io_submit *io,
struct page *page,
int len,
struct writeback_control *wbc)
{
struct inode *inode = page->mapping->host;
unsigned block_start, block_end, blocksize;
struct ext4_io_page *io_page;
struct buffer_head *bh, *head;
int ret = 0;
blocksize = 1 << inode->i_blkbits;
2011-02-08 01:46:14 +08:00
BUG_ON(!PageLocked(page));
BUG_ON(PageWriteback(page));
io_page = kmem_cache_alloc(io_page_cachep, GFP_NOFS);
if (!io_page) {
set_page_dirty(page);
unlock_page(page);
return -ENOMEM;
}
io_page->p_page = page;
atomic_set(&io_page->p_count, 1);
get_page(page);
set_page_writeback(page);
ClearPageError(page);
for (bh = head = page_buffers(page), block_start = 0;
bh != head || !block_start;
block_start = block_end, bh = bh->b_this_page) {
2011-02-08 01:46:14 +08:00
block_end = block_start + blocksize;
if (block_start >= len) {
clear_buffer_dirty(bh);
set_buffer_uptodate(bh);
continue;
}
2011-02-08 01:46:14 +08:00
clear_buffer_dirty(bh);
ret = io_submit_add_bh(io, io_page, inode, wbc, bh);
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.
*/
set_page_dirty(page);
break;
}
}
unlock_page(page);
/*
* If the page was truncated before we could do the writeback,
* or we had a memory allocation error while trying to write
* the first buffer head, we won't have submitted any pages for
* I/O. In that case we need to make sure we've cleared the
* PageWriteback bit from the page to prevent the system from
* wedging later on.
*/
put_io_page(io_page);
return ret;
}