linux/fs/jbd/commit.c

904 lines
26 KiB
C
Raw Normal View History

/*
* linux/fs/jbd/commit.c
*
* Written by Stephen C. Tweedie <sct@redhat.com>, 1998
*
* Copyright 1998 Red Hat corp --- All Rights Reserved
*
* This file is part of the Linux kernel and is made available under
* the terms of the GNU General Public License, version 2, or at your
* option, any later version, incorporated herein by reference.
*
* Journal commit routines for the generic filesystem journaling code;
* part of the ext2fs journaling system.
*/
#include <linux/time.h>
#include <linux/fs.h>
#include <linux/jbd.h>
#include <linux/errno.h>
#include <linux/slab.h>
#include <linux/mm.h>
#include <linux/pagemap.h>
/*
* Default IO end handler for temporary BJ_IO buffer_heads.
*/
static void journal_end_buffer_io_sync(struct buffer_head *bh, int uptodate)
{
BUFFER_TRACE(bh, "");
if (uptodate)
set_buffer_uptodate(bh);
else
clear_buffer_uptodate(bh);
unlock_buffer(bh);
}
/*
* When an ext3-ordered file is truncated, it is possible that many pages are
* not sucessfully freed, because they are attached to a committing transaction.
* After the transaction commits, these pages are left on the LRU, with no
* ->mapping, and with attached buffers. These pages are trivially reclaimable
* by the VM, but their apparent absence upsets the VM accounting, and it makes
* the numbers in /proc/meminfo look odd.
*
* So here, we have a buffer which has just come off the forget list. Look to
* see if we can strip all buffers from the backing page.
*
* Called under lock_journal(), and possibly under journal_datalist_lock. The
* caller provided us with a ref against the buffer, and we drop that here.
*/
static void release_buffer_page(struct buffer_head *bh)
{
struct page *page;
if (buffer_dirty(bh))
goto nope;
if (atomic_read(&bh->b_count) != 1)
goto nope;
page = bh->b_page;
if (!page)
goto nope;
if (page->mapping)
goto nope;
/* OK, it's a truncated page */
if (TestSetPageLocked(page))
goto nope;
page_cache_get(page);
__brelse(bh);
try_to_free_buffers(page);
unlock_page(page);
page_cache_release(page);
return;
nope:
__brelse(bh);
}
/*
* Try to acquire jbd_lock_bh_state() against the buffer, when j_list_lock is
* held. For ranking reasons we must trylock. If we lose, schedule away and
* return 0. j_list_lock is dropped in this case.
*/
static int inverted_lock(journal_t *journal, struct buffer_head *bh)
{
if (!jbd_trylock_bh_state(bh)) {
spin_unlock(&journal->j_list_lock);
schedule();
return 0;
}
return 1;
}
/* Done it all: now write the commit record. We should have
* cleaned up our previous buffers by now, so if we are in abort
* mode we can now just skip the rest of the journal write
* entirely.
*
* Returns 1 if the journal needs to be aborted or 0 on success
*/
static int journal_write_commit_record(journal_t *journal,
transaction_t *commit_transaction)
{
struct journal_head *descriptor;
struct buffer_head *bh;
journal_header_t *header;
int ret;
int barrier_done = 0;
if (is_journal_aborted(journal))
return 0;
descriptor = journal_get_descriptor_buffer(journal);
if (!descriptor)
return 1;
bh = jh2bh(descriptor);
header = (journal_header_t *)(bh->b_data);
header->h_magic = cpu_to_be32(JFS_MAGIC_NUMBER);
header->h_blocktype = cpu_to_be32(JFS_COMMIT_BLOCK);
header->h_sequence = cpu_to_be32(commit_transaction->t_tid);
JBUFFER_TRACE(descriptor, "write commit block");
set_buffer_dirty(bh);
if (journal->j_flags & JFS_BARRIER) {
set_buffer_ordered(bh);
barrier_done = 1;
}
ret = sync_dirty_buffer(bh);
ext3 can fail badly when device stops accepting BIO_RW_BARRIER requests Some devices - notably dm and md - can change their behaviour in response to BIO_RW_BARRIER requests. They might start out accepting such requests but on reconfiguration, they find out that they cannot any more. ext3 (and other filesystems) deal with this by always testing if BIO_RW_BARRIER requests fail with EOPNOTSUPP, and retrying the write requests without the barrier (probably after waiting for any pending writes to complete). However there is a bug in the handling for this for ext3. When ext3 (jbd actually) decides to submit a BIO_RW_BARRIER request, it sets the buffer_ordered flag on the buffer head. If the request completes successfully, the flag STAYS SET. Other code might then write the same buffer_head after the device has been reconfigured to not accept barriers. This write will then fail, but the "other code" is not ready to handle EOPNOTSUPP errors and the error will be treated as fatal. This can be seen without having to reconfigure a device at exactly the wrong time by putting: if (buffer_ordered(bh)) printk("OH DEAR, and ordered buffer\n"); in the while loop in "commit phase 5" of journal_commit_transaction. If it ever prints the "OH DEAR ..." message (as it does sometimes for me), then that request could (in different circumstances) have failed with EOPNOTSUPP, but that isn't tested for. My proposed fix is to clear the buffer_ordered flag after it has been used, as in the following patch. Signed-off-by: Neil Brown <neilb@suse.de> Cc: <linux-ext4@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-08 12:22:13 +00:00
if (barrier_done)
clear_buffer_ordered(bh);
/* is it possible for another commit to fail at roughly
* the same time as this one? If so, we don't want to
* trust the barrier flag in the super, but instead want
* to remember if we sent a barrier request
*/
if (ret == -EOPNOTSUPP && barrier_done) {
char b[BDEVNAME_SIZE];
printk(KERN_WARNING
"JBD: barrier-based sync failed on %s - "
"disabling barriers\n",
bdevname(journal->j_dev, b));
spin_lock(&journal->j_state_lock);
journal->j_flags &= ~JFS_BARRIER;
spin_unlock(&journal->j_state_lock);
/* And try again, without the barrier */
set_buffer_uptodate(bh);
set_buffer_dirty(bh);
ret = sync_dirty_buffer(bh);
}
put_bh(bh); /* One for getblk() */
journal_put_journal_head(descriptor);
return (ret == -EIO);
}
static void journal_do_submit_data(struct buffer_head **wbuf, int bufs)
{
int i;
for (i = 0; i < bufs; i++) {
wbuf[i]->b_end_io = end_buffer_write_sync;
/* We use-up our safety reference in submit_bh() */
submit_bh(WRITE, wbuf[i]);
}
}
/*
* Submit all the data buffers to disk
*/
static void journal_submit_data_buffers(journal_t *journal,
transaction_t *commit_transaction)
{
struct journal_head *jh;
struct buffer_head *bh;
int locked;
int bufs = 0;
struct buffer_head **wbuf = journal->j_wbuf;
/*
* Whenever we unlock the journal and sleep, things can get added
* onto ->t_sync_datalist, so we have to keep looping back to
* write_out_data until we *know* that the list is empty.
*
* Cleanup any flushed data buffers from the data list. Even in
* abort mode, we want to flush this out as soon as possible.
*/
write_out_data:
cond_resched();
spin_lock(&journal->j_list_lock);
while (commit_transaction->t_sync_datalist) {
jh = commit_transaction->t_sync_datalist;
bh = jh2bh(jh);
locked = 0;
/* Get reference just to make sure buffer does not disappear
* when we are forced to drop various locks */
get_bh(bh);
/* If the buffer is dirty, we need to submit IO and hence
* we need the buffer lock. We try to lock the buffer without
* blocking. If we fail, we need to drop j_list_lock and do
* blocking lock_buffer().
*/
if (buffer_dirty(bh)) {
if (test_set_buffer_locked(bh)) {
BUFFER_TRACE(bh, "needs blocking lock");
spin_unlock(&journal->j_list_lock);
/* Write out all data to prevent deadlocks */
journal_do_submit_data(wbuf, bufs);
bufs = 0;
lock_buffer(bh);
spin_lock(&journal->j_list_lock);
}
locked = 1;
}
/* We have to get bh_state lock. Again out of order, sigh. */
if (!inverted_lock(journal, bh)) {
jbd_lock_bh_state(bh);
spin_lock(&journal->j_list_lock);
}
/* Someone already cleaned up the buffer? */
if (!buffer_jbd(bh)
|| jh->b_transaction != commit_transaction
|| jh->b_jlist != BJ_SyncData) {
jbd_unlock_bh_state(bh);
if (locked)
unlock_buffer(bh);
BUFFER_TRACE(bh, "already cleaned up");
put_bh(bh);
continue;
}
if (locked && test_clear_buffer_dirty(bh)) {
BUFFER_TRACE(bh, "needs writeout, adding to array");
wbuf[bufs++] = bh;
__journal_file_buffer(jh, commit_transaction,
BJ_Locked);
jbd_unlock_bh_state(bh);
if (bufs == journal->j_wbufsize) {
spin_unlock(&journal->j_list_lock);
journal_do_submit_data(wbuf, bufs);
bufs = 0;
goto write_out_data;
}
} else if (!locked && buffer_locked(bh)) {
__journal_file_buffer(jh, commit_transaction,
BJ_Locked);
jbd_unlock_bh_state(bh);
put_bh(bh);
} else {
BUFFER_TRACE(bh, "writeout complete: unfile");
__journal_unfile_buffer(jh);
jbd_unlock_bh_state(bh);
if (locked)
unlock_buffer(bh);
journal_remove_journal_head(bh);
/* Once for our safety reference, once for
* journal_remove_journal_head() */
put_bh(bh);
put_bh(bh);
}
if (need_resched() || spin_needbreak(&journal->j_list_lock)) {
spin_unlock(&journal->j_list_lock);
goto write_out_data;
}
}
spin_unlock(&journal->j_list_lock);
journal_do_submit_data(wbuf, bufs);
}
/*
* journal_commit_transaction
*
* The primary function for committing a transaction to the log. This
* function is called by the journal thread to begin a complete commit.
*/
void journal_commit_transaction(journal_t *journal)
{
transaction_t *commit_transaction;
struct journal_head *jh, *new_jh, *descriptor;
struct buffer_head **wbuf = journal->j_wbuf;
int bufs;
int flags;
int err;
unsigned long blocknr;
char *tagp = NULL;
journal_header_t *header;
journal_block_tag_t *tag = NULL;
int space_left = 0;
int first_tag = 0;
int tag_flag;
int i;
/*
* First job: lock down the current transaction and wait for
* all outstanding updates to complete.
*/
#ifdef COMMIT_STATS
spin_lock(&journal->j_list_lock);
summarise_journal_usage(journal);
spin_unlock(&journal->j_list_lock);
#endif
/* Do we need to erase the effects of a prior journal_flush? */
if (journal->j_flags & JFS_FLUSHED) {
jbd_debug(3, "super block updated\n");
journal_update_superblock(journal, 1);
} else {
jbd_debug(3, "superblock not updated\n");
}
J_ASSERT(journal->j_running_transaction != NULL);
J_ASSERT(journal->j_committing_transaction == NULL);
commit_transaction = journal->j_running_transaction;
J_ASSERT(commit_transaction->t_state == T_RUNNING);
jbd_debug(1, "JBD: starting commit of transaction %d\n",
commit_transaction->t_tid);
spin_lock(&journal->j_state_lock);
commit_transaction->t_state = T_LOCKED;
spin_lock(&commit_transaction->t_handle_lock);
while (commit_transaction->t_updates) {
DEFINE_WAIT(wait);
prepare_to_wait(&journal->j_wait_updates, &wait,
TASK_UNINTERRUPTIBLE);
if (commit_transaction->t_updates) {
spin_unlock(&commit_transaction->t_handle_lock);
spin_unlock(&journal->j_state_lock);
schedule();
spin_lock(&journal->j_state_lock);
spin_lock(&commit_transaction->t_handle_lock);
}
finish_wait(&journal->j_wait_updates, &wait);
}
spin_unlock(&commit_transaction->t_handle_lock);
J_ASSERT (commit_transaction->t_outstanding_credits <=
journal->j_max_transaction_buffers);
/*
* First thing we are allowed to do is to discard any remaining
* BJ_Reserved buffers. Note, it is _not_ permissible to assume
* that there are no such buffers: if a large filesystem
* operation like a truncate needs to split itself over multiple
* transactions, then it may try to do a journal_restart() while
* there are still BJ_Reserved buffers outstanding. These must
* be released cleanly from the current transaction.
*
* In this case, the filesystem must still reserve write access
* again before modifying the buffer in the new transaction, but
* we do not require it to remember exactly which old buffers it
* has reserved. This is consistent with the existing behaviour
* that multiple journal_get_write_access() calls to the same
* buffer are perfectly permissable.
*/
while (commit_transaction->t_reserved_list) {
jh = commit_transaction->t_reserved_list;
JBUFFER_TRACE(jh, "reserved, unused: refile");
/*
* A journal_get_undo_access()+journal_release_buffer() may
* leave undo-committed data.
*/
if (jh->b_committed_data) {
struct buffer_head *bh = jh2bh(jh);
jbd_lock_bh_state(bh);
jbd_free(jh->b_committed_data, bh->b_size);
jh->b_committed_data = NULL;
jbd_unlock_bh_state(bh);
}
journal_refile_buffer(journal, jh);
}
/*
* Now try to drop any written-back buffers from the journal's
* checkpoint lists. We do this *before* commit because it potentially
* frees some memory
*/
spin_lock(&journal->j_list_lock);
__journal_clean_checkpoint_list(journal);
spin_unlock(&journal->j_list_lock);
jbd_debug (3, "JBD: commit phase 1\n");
/*
* Switch to a new revoke table.
*/
journal_switch_revoke_table(journal);
commit_transaction->t_state = T_FLUSH;
journal->j_committing_transaction = commit_transaction;
journal->j_running_transaction = NULL;
commit_transaction->t_log_start = journal->j_head;
wake_up(&journal->j_wait_transaction_locked);
spin_unlock(&journal->j_state_lock);
jbd_debug (3, "JBD: commit phase 2\n");
/*
* Now start flushing things to disk, in the order they appear
* on the transaction lists. Data blocks go first.
*/
err = 0;
journal_submit_data_buffers(journal, commit_transaction);
/*
* Wait for all previously submitted IO to complete.
*/
spin_lock(&journal->j_list_lock);
while (commit_transaction->t_locked_list) {
struct buffer_head *bh;
jh = commit_transaction->t_locked_list->b_tprev;
bh = jh2bh(jh);
get_bh(bh);
if (buffer_locked(bh)) {
spin_unlock(&journal->j_list_lock);
wait_on_buffer(bh);
if (unlikely(!buffer_uptodate(bh)))
err = -EIO;
spin_lock(&journal->j_list_lock);
}
if (!inverted_lock(journal, bh)) {
put_bh(bh);
spin_lock(&journal->j_list_lock);
continue;
}
if (buffer_jbd(bh) && jh->b_jlist == BJ_Locked) {
__journal_unfile_buffer(jh);
jbd_unlock_bh_state(bh);
journal_remove_journal_head(bh);
put_bh(bh);
} else {
jbd_unlock_bh_state(bh);
}
put_bh(bh);
cond_resched_lock(&journal->j_list_lock);
}
spin_unlock(&journal->j_list_lock);
if (err)
journal_abort(journal, err);
journal_write_revoke_records(journal, commit_transaction);
jbd_debug(3, "JBD: commit phase 2\n");
/*
* If we found any dirty or locked buffers, then we should have
* looped back up to the write_out_data label. If there weren't
* any then journal_clean_data_list should have wiped the list
* clean by now, so check that it is in fact empty.
*/
J_ASSERT (commit_transaction->t_sync_datalist == NULL);
jbd_debug (3, "JBD: commit phase 3\n");
/*
* Way to go: we have now written out all of the data for a
* transaction! Now comes the tricky part: we need to write out
* metadata. Loop over the transaction's entire buffer list:
*/
spin_lock(&journal->j_state_lock);
commit_transaction->t_state = T_COMMIT;
spin_unlock(&journal->j_state_lock);
jbd: fix possible journal overflow issues There are several cases where the running transaction can get buffers added to its BJ_Metadata list which it never dirtied, which makes its t_nr_buffers counter end up larger than its t_outstanding_credits counter. This will cause issues when starting new transactions as while we are logging buffers we decrement t_outstanding_buffers, so when t_outstanding_buffers goes negative, we will report that we need less space in the journal than we actually need, so transactions will be started even though there may not be enough room for them. In the worst case scenario (which admittedly is almost impossible to reproduce) this will result in the journal running out of space. The fix is to only refile buffers from the committing transaction to the running transactions BJ_Modified list when b_modified is set on that journal, which is the only way to be sure if the running transaction has modified that buffer. This patch also fixes an accounting error in journal_forget, it is possible that we can call journal_forget on a buffer without having modified it, only gotten write access to it, so instead of freeing a credit, we only do so if the buffer was modified. The assert will help catch if this problem occurs. Without these two patches I could hit this assert within minutes of running postmark, with them this issue no longer arises. Thank you, Signed-off-by: Josef Bacik <jbacik@redhat.com> Cc: <linux-ext4@vger.kernel.org> Acked-by: Jan Kara <jack@ucw.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 09:16:12 +00:00
J_ASSERT(commit_transaction->t_nr_buffers <=
commit_transaction->t_outstanding_credits);
descriptor = NULL;
bufs = 0;
while (commit_transaction->t_buffers) {
/* Find the next buffer to be journaled... */
jh = commit_transaction->t_buffers;
/* If we're in abort mode, we just un-journal the buffer and
release it for background writing. */
if (is_journal_aborted(journal)) {
JBUFFER_TRACE(jh, "journal is aborting: refile");
journal_refile_buffer(journal, jh);
/* If that was the last one, we need to clean up
* any descriptor buffers which may have been
* already allocated, even if we are now
* aborting. */
if (!commit_transaction->t_buffers)
goto start_journal_io;
continue;
}
/* Make sure we have a descriptor block in which to
record the metadata buffer. */
if (!descriptor) {
struct buffer_head *bh;
J_ASSERT (bufs == 0);
jbd_debug(4, "JBD: get descriptor\n");
descriptor = journal_get_descriptor_buffer(journal);
if (!descriptor) {
journal_abort(journal, -EIO);
continue;
}
bh = jh2bh(descriptor);
jbd_debug(4, "JBD: got buffer %llu (%p)\n",
(unsigned long long)bh->b_blocknr, bh->b_data);
header = (journal_header_t *)&bh->b_data[0];
header->h_magic = cpu_to_be32(JFS_MAGIC_NUMBER);
header->h_blocktype = cpu_to_be32(JFS_DESCRIPTOR_BLOCK);
header->h_sequence = cpu_to_be32(commit_transaction->t_tid);
tagp = &bh->b_data[sizeof(journal_header_t)];
space_left = bh->b_size - sizeof(journal_header_t);
first_tag = 1;
set_buffer_jwrite(bh);
set_buffer_dirty(bh);
wbuf[bufs++] = bh;
/* Record it so that we can wait for IO
completion later */
BUFFER_TRACE(bh, "ph3: file as descriptor");
journal_file_buffer(descriptor, commit_transaction,
BJ_LogCtl);
}
/* Where is the buffer to be written? */
err = journal_next_log_block(journal, &blocknr);
/* If the block mapping failed, just abandon the buffer
and repeat this loop: we'll fall into the
refile-on-abort condition above. */
if (err) {
journal_abort(journal, err);
continue;
}
/*
* start_this_handle() uses t_outstanding_credits to determine
* the free space in the log, but this counter is changed
* by journal_next_log_block() also.
*/
commit_transaction->t_outstanding_credits--;
/* Bump b_count to prevent truncate from stumbling over
the shadowed buffer! @@@ This can go if we ever get
rid of the BJ_IO/BJ_Shadow pairing of buffers. */
atomic_inc(&jh2bh(jh)->b_count);
/* Make a temporary IO buffer with which to write it out
(this will requeue both the metadata buffer and the
temporary IO buffer). new_bh goes on BJ_IO*/
set_bit(BH_JWrite, &jh2bh(jh)->b_state);
/*
* akpm: journal_write_metadata_buffer() sets
* new_bh->b_transaction to commit_transaction.
* We need to clean this up before we release new_bh
* (which is of type BJ_IO)
*/
JBUFFER_TRACE(jh, "ph3: write metadata");
flags = journal_write_metadata_buffer(commit_transaction,
jh, &new_jh, blocknr);
set_bit(BH_JWrite, &jh2bh(new_jh)->b_state);
wbuf[bufs++] = jh2bh(new_jh);
/* Record the new block's tag in the current descriptor
buffer */
tag_flag = 0;
if (flags & 1)
tag_flag |= JFS_FLAG_ESCAPE;
if (!first_tag)
tag_flag |= JFS_FLAG_SAME_UUID;
tag = (journal_block_tag_t *) tagp;
tag->t_blocknr = cpu_to_be32(jh2bh(jh)->b_blocknr);
tag->t_flags = cpu_to_be32(tag_flag);
tagp += sizeof(journal_block_tag_t);
space_left -= sizeof(journal_block_tag_t);
if (first_tag) {
memcpy (tagp, journal->j_uuid, 16);
tagp += 16;
space_left -= 16;
first_tag = 0;
}
/* If there's no more to do, or if the descriptor is full,
let the IO rip! */
if (bufs == journal->j_wbufsize ||
commit_transaction->t_buffers == NULL ||
space_left < sizeof(journal_block_tag_t) + 16) {
jbd_debug(4, "JBD: Submit %d IOs\n", bufs);
/* Write an end-of-descriptor marker before
submitting the IOs. "tag" still points to
the last tag we set up. */
tag->t_flags |= cpu_to_be32(JFS_FLAG_LAST_TAG);
start_journal_io:
for (i = 0; i < bufs; i++) {
struct buffer_head *bh = wbuf[i];
lock_buffer(bh);
clear_buffer_dirty(bh);
set_buffer_uptodate(bh);
bh->b_end_io = journal_end_buffer_io_sync;
submit_bh(WRITE, bh);
}
cond_resched();
/* Force a new descriptor to be generated next
time round the loop. */
descriptor = NULL;
bufs = 0;
}
}
/* Lo and behold: we have just managed to send a transaction to
the log. Before we can commit it, wait for the IO so far to
complete. Control buffers being written are on the
transaction's t_log_list queue, and metadata buffers are on
the t_iobuf_list queue.
Wait for the buffers in reverse order. That way we are
less likely to be woken up until all IOs have completed, and
so we incur less scheduling load.
*/
jbd_debug(3, "JBD: commit phase 4\n");
/*
* akpm: these are BJ_IO, and j_list_lock is not needed.
* See __journal_try_to_free_buffer.
*/
wait_for_iobuf:
while (commit_transaction->t_iobuf_list != NULL) {
struct buffer_head *bh;
jh = commit_transaction->t_iobuf_list->b_tprev;
bh = jh2bh(jh);
if (buffer_locked(bh)) {
wait_on_buffer(bh);
goto wait_for_iobuf;
}
if (cond_resched())
goto wait_for_iobuf;
if (unlikely(!buffer_uptodate(bh)))
err = -EIO;
clear_buffer_jwrite(bh);
JBUFFER_TRACE(jh, "ph4: unfile after journal write");
journal_unfile_buffer(journal, jh);
/*
* ->t_iobuf_list should contain only dummy buffer_heads
* which were created by journal_write_metadata_buffer().
*/
BUFFER_TRACE(bh, "dumping temporary bh");
journal_put_journal_head(jh);
__brelse(bh);
J_ASSERT_BH(bh, atomic_read(&bh->b_count) == 0);
free_buffer_head(bh);
/* We also have to unlock and free the corresponding
shadowed buffer */
jh = commit_transaction->t_shadow_list->b_tprev;
bh = jh2bh(jh);
clear_bit(BH_JWrite, &bh->b_state);
J_ASSERT_BH(bh, buffer_jbddirty(bh));
/* The metadata is now released for reuse, but we need
to remember it against this transaction so that when
we finally commit, we can do any checkpointing
required. */
JBUFFER_TRACE(jh, "file as BJ_Forget");
journal_file_buffer(jh, commit_transaction, BJ_Forget);
/* Wake up any transactions which were waiting for this
IO to complete */
wake_up_bit(&bh->b_state, BH_Unshadow);
JBUFFER_TRACE(jh, "brelse shadowed buffer");
__brelse(bh);
}
J_ASSERT (commit_transaction->t_shadow_list == NULL);
jbd_debug(3, "JBD: commit phase 5\n");
/* Here we wait for the revoke record and descriptor record buffers */
wait_for_ctlbuf:
while (commit_transaction->t_log_list != NULL) {
struct buffer_head *bh;
jh = commit_transaction->t_log_list->b_tprev;
bh = jh2bh(jh);
if (buffer_locked(bh)) {
wait_on_buffer(bh);
goto wait_for_ctlbuf;
}
if (cond_resched())
goto wait_for_ctlbuf;
if (unlikely(!buffer_uptodate(bh)))
err = -EIO;
BUFFER_TRACE(bh, "ph5: control buffer writeout done: unfile");
clear_buffer_jwrite(bh);
journal_unfile_buffer(journal, jh);
journal_put_journal_head(jh);
__brelse(bh); /* One for getblk */
/* AKPM: bforget here */
}
jbd_debug(3, "JBD: commit phase 6\n");
if (journal_write_commit_record(journal, commit_transaction))
err = -EIO;
if (err)
journal_abort(journal, err);
/* End of a transaction! Finally, we can do checkpoint
processing: any buffers committed as a result of this
transaction can be removed from any checkpoint list it was on
before. */
jbd_debug(3, "JBD: commit phase 7\n");
J_ASSERT(commit_transaction->t_sync_datalist == NULL);
J_ASSERT(commit_transaction->t_buffers == NULL);
J_ASSERT(commit_transaction->t_checkpoint_list == NULL);
J_ASSERT(commit_transaction->t_iobuf_list == NULL);
J_ASSERT(commit_transaction->t_shadow_list == NULL);
J_ASSERT(commit_transaction->t_log_list == NULL);
restart_loop:
/*
* As there are other places (journal_unmap_buffer()) adding buffers
* to this list we have to be careful and hold the j_list_lock.
*/
spin_lock(&journal->j_list_lock);
while (commit_transaction->t_forget) {
transaction_t *cp_transaction;
struct buffer_head *bh;
jh = commit_transaction->t_forget;
spin_unlock(&journal->j_list_lock);
bh = jh2bh(jh);
jbd_lock_bh_state(bh);
J_ASSERT_JH(jh, jh->b_transaction == commit_transaction ||
jh->b_transaction == journal->j_running_transaction);
/*
* If there is undo-protected committed data against
* this buffer, then we can remove it now. If it is a
* buffer needing such protection, the old frozen_data
* field now points to a committed version of the
* buffer, so rotate that field to the new committed
* data.
*
* Otherwise, we can just throw away the frozen data now.
*/
if (jh->b_committed_data) {
jbd_free(jh->b_committed_data, bh->b_size);
jh->b_committed_data = NULL;
if (jh->b_frozen_data) {
jh->b_committed_data = jh->b_frozen_data;
jh->b_frozen_data = NULL;
}
} else if (jh->b_frozen_data) {
jbd_free(jh->b_frozen_data, bh->b_size);
jh->b_frozen_data = NULL;
}
spin_lock(&journal->j_list_lock);
cp_transaction = jh->b_cp_transaction;
if (cp_transaction) {
JBUFFER_TRACE(jh, "remove from old cp transaction");
__journal_remove_checkpoint(jh);
}
/* Only re-checkpoint the buffer_head if it is marked
* dirty. If the buffer was added to the BJ_Forget list
* by journal_forget, it may no longer be dirty and
* there's no point in keeping a checkpoint record for
* it. */
/* A buffer which has been freed while still being
* journaled by a previous transaction may end up still
* being dirty here, but we want to avoid writing back
* that buffer in the future now that the last use has
* been committed. That's not only a performance gain,
* it also stops aliasing problems if the buffer is left
* behind for writeback and gets reallocated for another
* use in a different page. */
if (buffer_freed(bh)) {
clear_buffer_freed(bh);
clear_buffer_jbddirty(bh);
}
if (buffer_jbddirty(bh)) {
JBUFFER_TRACE(jh, "add to new checkpointing trans");
__journal_insert_checkpoint(jh, commit_transaction);
JBUFFER_TRACE(jh, "refile for checkpoint writeback");
__journal_refile_buffer(jh);
jbd_unlock_bh_state(bh);
} else {
J_ASSERT_BH(bh, !buffer_dirty(bh));
[PATCH] jbd: fix BUG in journal_commit_transaction() Fix possible assertion failure in journal_commit_transaction() on jh->b_next_transaction == NULL (when we are processing BJ_Forget list and buffer is not jbddirty). !jbddirty buffers can be placed on BJ_Forget list for example by journal_forget() or by __dispose_buffer() - generally such buffer means that it has been freed by this transaction. Freed buffers should not be reallocated until the transaction has committed (that's why we have the assertion there) but they *can* be reallocated when the transaction has already been committed to disk and we are just processing the BJ_Forget list (as soon as we remove b_committed_data from the bitmap bh, ext3 will be able to reallocate buffers freed by the committing transaction). So we have to also count with the case that the buffer has been reallocated and b_next_transaction has been already set. And one more subtle point: it can happen that we manage to reallocate the buffer and also mark it jbddirty. Then we also add the freed buffer to the checkpoint list of the committing trasaction. But that should do no harm. Non-jbddirty buffers should be filed to BJ_Reserved and not BJ_Metadata list. It can actually happen that we refile such buffers during the commit phase when we reallocate in the running transaction blocks deleted in committing transaction (and that can happen if the committing transaction already wrote all the data and is just cleaning up BJ_Forget list). Signed-off-by: Jan Kara <jack@suse.cz> Acked-by: "Stephen C. Tweedie" <sct@redhat.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 09:05:25 +00:00
/* The buffer on BJ_Forget list and not jbddirty means
* it has been freed by this transaction and hence it
* could not have been reallocated until this
* transaction has committed. *BUT* it could be
* reallocated once we have written all the data to
* disk and before we process the buffer on BJ_Forget
* list. */
JBUFFER_TRACE(jh, "refile or unfile freed buffer");
__journal_refile_buffer(jh);
if (!jh->b_transaction) {
jbd_unlock_bh_state(bh);
/* needs a brelse */
journal_remove_journal_head(bh);
release_buffer_page(bh);
} else
jbd_unlock_bh_state(bh);
}
cond_resched_lock(&journal->j_list_lock);
}
spin_unlock(&journal->j_list_lock);
/*
* This is a bit sleazy. We use j_list_lock to protect transition
* of a transaction into T_FINISHED state and calling
* __journal_drop_transaction(). Otherwise we could race with
* other checkpointing code processing the transaction...
*/
spin_lock(&journal->j_state_lock);
spin_lock(&journal->j_list_lock);
/*
* Now recheck if some buffers did not get attached to the transaction
* while the lock was dropped...
*/
if (commit_transaction->t_forget) {
spin_unlock(&journal->j_list_lock);
spin_unlock(&journal->j_state_lock);
goto restart_loop;
}
/* Done with this transaction! */
jbd_debug(3, "JBD: commit phase 8\n");
J_ASSERT(commit_transaction->t_state == T_COMMIT);
commit_transaction->t_state = T_FINISHED;
J_ASSERT(commit_transaction == journal->j_committing_transaction);
journal->j_commit_sequence = commit_transaction->t_tid;
journal->j_committing_transaction = NULL;
spin_unlock(&journal->j_state_lock);
if (commit_transaction->t_checkpoint_list == NULL &&
commit_transaction->t_checkpoint_io_list == NULL) {
__journal_drop_transaction(journal, commit_transaction);
} else {
if (journal->j_checkpoint_transactions == NULL) {
journal->j_checkpoint_transactions = commit_transaction;
commit_transaction->t_cpnext = commit_transaction;
commit_transaction->t_cpprev = commit_transaction;
} else {
commit_transaction->t_cpnext =
journal->j_checkpoint_transactions;
commit_transaction->t_cpprev =
commit_transaction->t_cpnext->t_cpprev;
commit_transaction->t_cpnext->t_cpprev =
commit_transaction;
commit_transaction->t_cpprev->t_cpnext =
commit_transaction;
}
}
spin_unlock(&journal->j_list_lock);
jbd_debug(1, "JBD: commit %d complete, head %d\n",
journal->j_commit_sequence, journal->j_tail_sequence);
wake_up(&journal->j_wait_done_commit);
}