diff options
author | Linus Torvalds <torvalds@ppc970.osdl.org> | 2005-04-16 15:20:36 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@ppc970.osdl.org> | 2005-04-16 15:20:36 -0700 |
commit | 1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 (patch) | |
tree | 0bba044c4ce775e45a88a51686b5d9f90697ea9d /fs/jbd |
Linux-2.6.12-rc2
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.
Let it rip!
Diffstat (limited to 'fs/jbd')
-rw-r--r-- | fs/jbd/Makefile | 7 | ||||
-rw-r--r-- | fs/jbd/checkpoint.c | 636 | ||||
-rw-r--r-- | fs/jbd/commit.c | 844 | ||||
-rw-r--r-- | fs/jbd/journal.c | 2003 | ||||
-rw-r--r-- | fs/jbd/recovery.c | 591 | ||||
-rw-r--r-- | fs/jbd/revoke.c | 702 | ||||
-rw-r--r-- | fs/jbd/transaction.c | 2062 |
7 files changed, 6845 insertions, 0 deletions
diff --git a/fs/jbd/Makefile b/fs/jbd/Makefile new file mode 100644 index 000000000000..54aca4868a36 --- /dev/null +++ b/fs/jbd/Makefile @@ -0,0 +1,7 @@ +# +# Makefile for the linux journaling routines. +# + +obj-$(CONFIG_JBD) += jbd.o + +jbd-objs := transaction.o commit.o recovery.o checkpoint.o revoke.o journal.o diff --git a/fs/jbd/checkpoint.c b/fs/jbd/checkpoint.c new file mode 100644 index 000000000000..98d830401c56 --- /dev/null +++ b/fs/jbd/checkpoint.c @@ -0,0 +1,636 @@ +/* + * linux/fs/checkpoint.c + * + * Written by Stephen C. Tweedie <sct@redhat.com>, 1999 + * + * Copyright 1999 Red Hat Software --- 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. + * + * Checkpoint routines for the generic filesystem journaling code. + * Part of the ext2fs journaling system. + * + * Checkpointing is the process of ensuring that a section of the log is + * committed fully to disk, so that that portion of the log can be + * reused. + */ + +#include <linux/time.h> +#include <linux/fs.h> +#include <linux/jbd.h> +#include <linux/errno.h> +#include <linux/slab.h> + +/* + * Unlink a buffer from a transaction. + * + * Called with j_list_lock held. + */ + +static inline void __buffer_unlink(struct journal_head *jh) +{ + transaction_t *transaction; + + transaction = jh->b_cp_transaction; + jh->b_cp_transaction = NULL; + + jh->b_cpnext->b_cpprev = jh->b_cpprev; + jh->b_cpprev->b_cpnext = jh->b_cpnext; + if (transaction->t_checkpoint_list == jh) + transaction->t_checkpoint_list = jh->b_cpnext; + if (transaction->t_checkpoint_list == jh) + transaction->t_checkpoint_list = NULL; +} + +/* + * Try to release a checkpointed buffer from its transaction. + * Returns 1 if we released it. + * Requires j_list_lock + * Called under jbd_lock_bh_state(jh2bh(jh)), and drops it + */ +static int __try_to_free_cp_buf(struct journal_head *jh) +{ + int ret = 0; + struct buffer_head *bh = jh2bh(jh); + + if (jh->b_jlist == BJ_None && !buffer_locked(bh) && !buffer_dirty(bh)) { + JBUFFER_TRACE(jh, "remove from checkpoint list"); + __journal_remove_checkpoint(jh); + jbd_unlock_bh_state(bh); + journal_remove_journal_head(bh); + BUFFER_TRACE(bh, "release"); + __brelse(bh); + ret = 1; + } else { + jbd_unlock_bh_state(bh); + } + return ret; +} + +/* + * __log_wait_for_space: wait until there is space in the journal. + * + * Called under j-state_lock *only*. It will be unlocked if we have to wait + * for a checkpoint to free up some space in the log. + */ +void __log_wait_for_space(journal_t *journal) +{ + int nblocks; + assert_spin_locked(&journal->j_state_lock); + + nblocks = jbd_space_needed(journal); + while (__log_space_left(journal) < nblocks) { + if (journal->j_flags & JFS_ABORT) + return; + spin_unlock(&journal->j_state_lock); + down(&journal->j_checkpoint_sem); + + /* + * Test again, another process may have checkpointed while we + * were waiting for the checkpoint lock + */ + spin_lock(&journal->j_state_lock); + nblocks = jbd_space_needed(journal); + if (__log_space_left(journal) < nblocks) { + spin_unlock(&journal->j_state_lock); + log_do_checkpoint(journal); + spin_lock(&journal->j_state_lock); + } + up(&journal->j_checkpoint_sem); + } +} + +/* + * We were unable to perform jbd_trylock_bh_state() inside j_list_lock. + * The caller must restart a list walk. Wait for someone else to run + * jbd_unlock_bh_state(). + */ +static void jbd_sync_bh(journal_t *journal, struct buffer_head *bh) +{ + get_bh(bh); + spin_unlock(&journal->j_list_lock); + jbd_lock_bh_state(bh); + jbd_unlock_bh_state(bh); + put_bh(bh); +} + +/* + * Clean up a transaction's checkpoint list. + * + * We wait for any pending IO to complete and make sure any clean + * buffers are removed from the transaction. + * + * Return 1 if we performed any actions which might have destroyed the + * checkpoint. (journal_remove_checkpoint() deletes the transaction when + * the last checkpoint buffer is cleansed) + * + * Called with j_list_lock held. + */ +static int __cleanup_transaction(journal_t *journal, transaction_t *transaction) +{ + struct journal_head *jh, *next_jh, *last_jh; + struct buffer_head *bh; + int ret = 0; + + assert_spin_locked(&journal->j_list_lock); + jh = transaction->t_checkpoint_list; + if (!jh) + return 0; + + last_jh = jh->b_cpprev; + next_jh = jh; + do { + jh = next_jh; + bh = jh2bh(jh); + if (buffer_locked(bh)) { + atomic_inc(&bh->b_count); + spin_unlock(&journal->j_list_lock); + wait_on_buffer(bh); + /* the journal_head may have gone by now */ + BUFFER_TRACE(bh, "brelse"); + __brelse(bh); + goto out_return_1; + } + + /* + * This is foul + */ + if (!jbd_trylock_bh_state(bh)) { + jbd_sync_bh(journal, bh); + goto out_return_1; + } + + if (jh->b_transaction != NULL) { + transaction_t *t = jh->b_transaction; + tid_t tid = t->t_tid; + + spin_unlock(&journal->j_list_lock); + jbd_unlock_bh_state(bh); + log_start_commit(journal, tid); + log_wait_commit(journal, tid); + goto out_return_1; + } + + /* + * AKPM: I think the buffer_jbddirty test is redundant - it + * shouldn't have NULL b_transaction? + */ + next_jh = jh->b_cpnext; + if (!buffer_dirty(bh) && !buffer_jbddirty(bh)) { + BUFFER_TRACE(bh, "remove from checkpoint"); + __journal_remove_checkpoint(jh); + jbd_unlock_bh_state(bh); + journal_remove_journal_head(bh); + __brelse(bh); + ret = 1; + } else { + jbd_unlock_bh_state(bh); + } + jh = next_jh; + } while (jh != last_jh); + + return ret; +out_return_1: + spin_lock(&journal->j_list_lock); + return 1; +} + +#define NR_BATCH 64 + +static void +__flush_batch(journal_t *journal, struct buffer_head **bhs, int *batch_count) +{ + int i; + + spin_unlock(&journal->j_list_lock); + ll_rw_block(WRITE, *batch_count, bhs); + spin_lock(&journal->j_list_lock); + for (i = 0; i < *batch_count; i++) { + struct buffer_head *bh = bhs[i]; + clear_buffer_jwrite(bh); + BUFFER_TRACE(bh, "brelse"); + __brelse(bh); + } + *batch_count = 0; +} + +/* + * Try to flush one buffer from the checkpoint list to disk. + * + * Return 1 if something happened which requires us to abort the current + * scan of the checkpoint list. + * + * Called with j_list_lock held. + * Called under jbd_lock_bh_state(jh2bh(jh)), and drops it + */ +static int __flush_buffer(journal_t *journal, struct journal_head *jh, + struct buffer_head **bhs, int *batch_count, + int *drop_count) +{ + struct buffer_head *bh = jh2bh(jh); + int ret = 0; + + if (buffer_dirty(bh) && !buffer_locked(bh) && jh->b_jlist == BJ_None) { + J_ASSERT_JH(jh, jh->b_transaction == NULL); + + /* + * Important: we are about to write the buffer, and + * possibly block, while still holding the journal lock. + * We cannot afford to let the transaction logic start + * messing around with this buffer before we write it to + * disk, as that would break recoverability. + */ + BUFFER_TRACE(bh, "queue"); + get_bh(bh); + J_ASSERT_BH(bh, !buffer_jwrite(bh)); + set_buffer_jwrite(bh); + bhs[*batch_count] = bh; + jbd_unlock_bh_state(bh); + (*batch_count)++; + if (*batch_count == NR_BATCH) { + __flush_batch(journal, bhs, batch_count); + ret = 1; + } + } else { + int last_buffer = 0; + if (jh->b_cpnext == jh) { + /* We may be about to drop the transaction. Tell the + * caller that the lists have changed. + */ + last_buffer = 1; + } + if (__try_to_free_cp_buf(jh)) { + (*drop_count)++; + ret = last_buffer; + } + } + return ret; +} + +/* + * Perform an actual checkpoint. We don't write out only enough to + * satisfy the current blocked requests: rather we submit a reasonably + * sized chunk of the outstanding data to disk at once for + * efficiency. __log_wait_for_space() will retry if we didn't free enough. + * + * However, we _do_ take into account the amount requested so that once + * the IO has been queued, we can return as soon as enough of it has + * completed to disk. + * + * The journal should be locked before calling this function. + */ +int log_do_checkpoint(journal_t *journal) +{ + int result; + int batch_count = 0; + struct buffer_head *bhs[NR_BATCH]; + + jbd_debug(1, "Start checkpoint\n"); + + /* + * First thing: if there are any transactions in the log which + * don't need checkpointing, just eliminate them from the + * journal straight away. + */ + result = cleanup_journal_tail(journal); + jbd_debug(1, "cleanup_journal_tail returned %d\n", result); + if (result <= 0) + return result; + + /* + * OK, we need to start writing disk blocks. Try to free up a + * quarter of the log in a single checkpoint if we can. + */ + /* + * AKPM: check this code. I had a feeling a while back that it + * degenerates into a busy loop at unmount time. + */ + spin_lock(&journal->j_list_lock); + while (journal->j_checkpoint_transactions) { + transaction_t *transaction; + struct journal_head *jh, *last_jh, *next_jh; + int drop_count = 0; + int cleanup_ret, retry = 0; + tid_t this_tid; + + transaction = journal->j_checkpoint_transactions; + this_tid = transaction->t_tid; + jh = transaction->t_checkpoint_list; + last_jh = jh->b_cpprev; + next_jh = jh; + do { + struct buffer_head *bh; + + jh = next_jh; + next_jh = jh->b_cpnext; + bh = jh2bh(jh); + if (!jbd_trylock_bh_state(bh)) { + jbd_sync_bh(journal, bh); + spin_lock(&journal->j_list_lock); + retry = 1; + break; + } + retry = __flush_buffer(journal, jh, bhs, &batch_count, &drop_count); + if (cond_resched_lock(&journal->j_list_lock)) { + retry = 1; + break; + } + } while (jh != last_jh && !retry); + + if (batch_count) + __flush_batch(journal, bhs, &batch_count); + + /* + * If someone cleaned up this transaction while we slept, we're + * done + */ + if (journal->j_checkpoint_transactions != transaction) + break; + if (retry) + continue; + /* + * Maybe it's a new transaction, but it fell at the same + * address + */ + if (transaction->t_tid != this_tid) + continue; + /* + * We have walked the whole transaction list without + * finding anything to write to disk. We had better be + * able to make some progress or we are in trouble. + */ + cleanup_ret = __cleanup_transaction(journal, transaction); + J_ASSERT(drop_count != 0 || cleanup_ret != 0); + if (journal->j_checkpoint_transactions != transaction) + break; + } + spin_unlock(&journal->j_list_lock); + result = cleanup_journal_tail(journal); + if (result < 0) + return result; + + return 0; +} + +/* + * Check the list of checkpoint transactions for the journal to see if + * we have already got rid of any since the last update of the log tail + * in the journal superblock. If so, we can instantly roll the + * superblock forward to remove those transactions from the log. + * + * Return <0 on error, 0 on success, 1 if there was nothing to clean up. + * + * Called with the journal lock held. + * + * This is the only part of the journaling code which really needs to be + * aware of transaction aborts. Checkpointing involves writing to the + * main filesystem area rather than to the journal, so it can proceed + * even in abort state, but we must not update the journal superblock if + * we have an abort error outstanding. + */ + +int cleanup_journal_tail(journal_t *journal) +{ + transaction_t * transaction; + tid_t first_tid; + unsigned long blocknr, freed; + + /* OK, work out the oldest transaction remaining in the log, and + * the log block it starts at. + * + * If the log is now empty, we need to work out which is the + * next transaction ID we will write, and where it will + * start. */ + + spin_lock(&journal->j_state_lock); + spin_lock(&journal->j_list_lock); + transaction = journal->j_checkpoint_transactions; + if (transaction) { + first_tid = transaction->t_tid; + blocknr = transaction->t_log_start; + } else if ((transaction = journal->j_committing_transaction) != NULL) { + first_tid = transaction->t_tid; + blocknr = transaction->t_log_start; + } else if ((transaction = journal->j_running_transaction) != NULL) { + first_tid = transaction->t_tid; + blocknr = journal->j_head; + } else { + first_tid = journal->j_transaction_sequence; + blocknr = journal->j_head; + } + spin_unlock(&journal->j_list_lock); + J_ASSERT(blocknr != 0); + + /* If the oldest pinned transaction is at the tail of the log + already then there's not much we can do right now. */ + if (journal->j_tail_sequence == first_tid) { + spin_unlock(&journal->j_state_lock); + return 1; + } + + /* OK, update the superblock to recover the freed space. + * Physical blocks come first: have we wrapped beyond the end of + * the log? */ + freed = blocknr - journal->j_tail; + if (blocknr < journal->j_tail) + freed = freed + journal->j_last - journal->j_first; + + jbd_debug(1, + "Cleaning journal tail from %d to %d (offset %lu), " + "freeing %lu\n", + journal->j_tail_sequence, first_tid, blocknr, freed); + + journal->j_free += freed; + journal->j_tail_sequence = first_tid; + journal->j_tail = blocknr; + spin_unlock(&journal->j_state_lock); + if (!(journal->j_flags & JFS_ABORT)) + journal_update_superblock(journal, 1); + return 0; +} + + +/* Checkpoint list management */ + +/* + * journal_clean_checkpoint_list + * + * Find all the written-back checkpoint buffers in the journal and release them. + * + * Called with the journal locked. + * Called with j_list_lock held. + * Returns number of bufers reaped (for debug) + */ + +int __journal_clean_checkpoint_list(journal_t *journal) +{ + transaction_t *transaction, *last_transaction, *next_transaction; + int ret = 0; + + transaction = journal->j_checkpoint_transactions; + if (transaction == 0) + goto out; + + last_transaction = transaction->t_cpprev; + next_transaction = transaction; + do { + struct journal_head *jh; + + transaction = next_transaction; + next_transaction = transaction->t_cpnext; + jh = transaction->t_checkpoint_list; + if (jh) { + struct journal_head *last_jh = jh->b_cpprev; + struct journal_head *next_jh = jh; + + do { + jh = next_jh; + next_jh = jh->b_cpnext; + /* Use trylock because of the ranknig */ + if (jbd_trylock_bh_state(jh2bh(jh))) + ret += __try_to_free_cp_buf(jh); + /* + * This function only frees up some memory + * if possible so we dont have an obligation + * to finish processing. Bail out if preemption + * requested: + */ + if (need_resched()) + goto out; + } while (jh != last_jh); + } + } while (transaction != last_transaction); +out: + return ret; +} + +/* + * journal_remove_checkpoint: called after a buffer has been committed + * to disk (either by being write-back flushed to disk, or being + * committed to the log). + * + * We cannot safely clean a transaction out of the log until all of the + * buffer updates committed in that transaction have safely been stored + * elsewhere on disk. To achieve this, all of the buffers in a + * transaction need to be maintained on the transaction's checkpoint + * list until they have been rewritten, at which point this function is + * called to remove the buffer from the existing transaction's + * checkpoint list. + * + * This function is called with the journal locked. + * This function is called with j_list_lock held. + */ + +void __journal_remove_checkpoint(struct journal_head *jh) +{ + transaction_t *transaction; + journal_t *journal; + + JBUFFER_TRACE(jh, "entry"); + + if ((transaction = jh->b_cp_transaction) == NULL) { + JBUFFER_TRACE(jh, "not on transaction"); + goto out; + } + journal = transaction->t_journal; + + __buffer_unlink(jh); + + if (transaction->t_checkpoint_list != NULL) + goto out; + JBUFFER_TRACE(jh, "transaction has no more buffers"); + + /* + * There is one special case to worry about: if we have just pulled the + * buffer off a committing transaction's forget list, then even if the + * checkpoint list is empty, the transaction obviously cannot be + * dropped! + * + * The locking here around j_committing_transaction is a bit sleazy. + * See the comment at the end of journal_commit_transaction(). + */ + if (transaction == journal->j_committing_transaction) { + JBUFFER_TRACE(jh, "belongs to committing transaction"); + goto out; + } + + /* OK, that was the last buffer for the transaction: we can now + safely remove this transaction from the log */ + + __journal_drop_transaction(journal, transaction); + + /* Just in case anybody was waiting for more transactions to be + checkpointed... */ + wake_up(&journal->j_wait_logspace); +out: + JBUFFER_TRACE(jh, "exit"); +} + +/* + * journal_insert_checkpoint: put a committed buffer onto a checkpoint + * list so that we know when it is safe to clean the transaction out of + * the log. + * + * Called with the journal locked. + * Called with j_list_lock held. + */ +void __journal_insert_checkpoint(struct journal_head *jh, + transaction_t *transaction) +{ + JBUFFER_TRACE(jh, "entry"); + J_ASSERT_JH(jh, buffer_dirty(jh2bh(jh)) || buffer_jbddirty(jh2bh(jh))); + J_ASSERT_JH(jh, jh->b_cp_transaction == NULL); + + jh->b_cp_transaction = transaction; + + if (!transaction->t_checkpoint_list) { + jh->b_cpnext = jh->b_cpprev = jh; + } else { + jh->b_cpnext = transaction->t_checkpoint_list; + jh->b_cpprev = transaction->t_checkpoint_list->b_cpprev; + jh->b_cpprev->b_cpnext = jh; + jh->b_cpnext->b_cpprev = jh; + } + transaction->t_checkpoint_list = jh; +} + +/* + * We've finished with this transaction structure: adios... + * + * The transaction must have no links except for the checkpoint by this + * point. + * + * Called with the journal locked. + * Called with j_list_lock held. + */ + +void __journal_drop_transaction(journal_t *journal, transaction_t *transaction) +{ + assert_spin_locked(&journal->j_list_lock); + if (transaction->t_cpnext) { + transaction->t_cpnext->t_cpprev = transaction->t_cpprev; + transaction->t_cpprev->t_cpnext = transaction->t_cpnext; + if (journal->j_checkpoint_transactions == transaction) + journal->j_checkpoint_transactions = + transaction->t_cpnext; + if (journal->j_checkpoint_transactions == transaction) + journal->j_checkpoint_transactions = NULL; + } + + J_ASSERT(transaction->t_state == T_FINISHED); + J_ASSERT(transaction->t_buffers == NULL); + J_ASSERT(transaction->t_sync_datalist == NULL); + J_ASSERT(transaction->t_forget == NULL); + J_ASSERT(transaction->t_iobuf_list == NULL); + J_ASSERT(transaction->t_shadow_list == NULL); + J_ASSERT(transaction->t_log_list == NULL); + J_ASSERT(transaction->t_checkpoint_list == NULL); + J_ASSERT(transaction->t_updates == 0); + J_ASSERT(journal->j_committing_transaction != transaction); + J_ASSERT(journal->j_running_transaction != transaction); + + jbd_debug(1, "Dropping transaction %d, all done\n", transaction->t_tid); + kfree(transaction); +} diff --git a/fs/jbd/commit.c b/fs/jbd/commit.c new file mode 100644 index 000000000000..dac720c837ab --- /dev/null +++ b/fs/jbd/commit.c @@ -0,0 +1,844 @@ +/* + * linux/fs/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> +#include <linux/smp_lock.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; + int i, 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); + + /* AKPM: buglet - add `i' to tmp! */ + for (i = 0; i < bh->b_size; i += 512) { + journal_header_t *tmp = (journal_header_t*)bh->b_data; + tmp->h_magic = cpu_to_be32(JFS_MAGIC_NUMBER); + tmp->h_blocktype = cpu_to_be32(JFS_COMMIT_BLOCK); + tmp->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); + /* 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 */ + clear_buffer_ordered(bh); + 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); +} + +/* + * 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); + if (jh->b_committed_data) { + kfree(jh->b_committed_data); + 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"); + + /* + * First, drop modified flag: all accesses to the buffers + * will be tracked for a new trasaction only -bzzz + */ + spin_lock(&journal->j_list_lock); + if (commit_transaction->t_buffers) { + new_jh = jh = commit_transaction->t_buffers->b_tnext; + do { + J_ASSERT_JH(new_jh, new_jh->b_modified == 1 || + new_jh->b_modified == 0); + new_jh->b_modified = 0; + new_jh = new_jh->b_tnext; + } while (new_jh != jh); + } + spin_unlock(&journal->j_list_lock); + + /* + * Now start flushing things to disk, in the order they appear + * on the transaction lists. Data blocks go first. + */ + + err = 0; + /* + * 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. + */ + bufs = 0; + /* + * 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) { + struct buffer_head *bh; + + jh = commit_transaction->t_sync_datalist; + commit_transaction->t_sync_datalist = jh->b_tnext; + bh = jh2bh(jh); + if (buffer_locked(bh)) { + BUFFER_TRACE(bh, "locked"); + if (!inverted_lock(journal, bh)) + goto write_out_data; + __journal_temp_unlink_buffer(jh); + __journal_file_buffer(jh, commit_transaction, + BJ_Locked); + jbd_unlock_bh_state(bh); + if (lock_need_resched(&journal->j_list_lock)) { + spin_unlock(&journal->j_list_lock); + goto write_out_data; + } + } else { + if (buffer_dirty(bh)) { + BUFFER_TRACE(bh, "start journal writeout"); + get_bh(bh); + wbuf[bufs++] = bh; + if (bufs == journal->j_wbufsize) { + jbd_debug(2, "submit %d writes\n", + bufs); + spin_unlock(&journal->j_list_lock); + ll_rw_block(WRITE, bufs, wbuf); + journal_brelse_array(wbuf, bufs); + bufs = 0; + goto write_out_data; + } + } else { + BUFFER_TRACE(bh, "writeout complete: unfile"); + if (!inverted_lock(journal, bh)) + goto write_out_data; + __journal_unfile_buffer(jh); + jbd_unlock_bh_state(bh); + journal_remove_journal_head(bh); + put_bh(bh); + if (lock_need_resched(&journal->j_list_lock)) { + spin_unlock(&journal->j_list_lock); + goto write_out_data; + } + } + } + } + + if (bufs) { + spin_unlock(&journal->j_list_lock); + ll_rw_block(WRITE, bufs, wbuf); + journal_brelse_array(wbuf, bufs); + spin_lock(&journal->j_list_lock); + } + + /* + * Wait for all previously submitted IO to complete. + */ + 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(&jou |