int journal_stop (handle_t *handle)
{
transaction_t *transaction = handle->h_transaction;
journal_t *journal = transaction->t_journal;
int force_sync;
if (!handle)
return 0;
J_ASSERT (transaction->t_updates > 0);
J_ASSERT (current->j_handle == handle);
if (--handle->h_ref > 0)
return 0;
jfs_debug(4, "Handle %p going down\n", handle);
current->j_handle = NULL;
// current->fs_locks--;
transaction->t_outstanding_credits -= handle->h_buffer_credits;
transaction->t_updates--;
if (!transaction->t_updates) {
wake_up(&journal->j_wait_updates);
if (journal->j_barrier_count)
wake_up(&journal->j_wait_transaction_locked);
}
/*
* If the journal is marked SYNC, we need to set another commit
* going! We also want to force a commit if the current
* transaction is occupying too much of the log, or if the
* transaction is too old now.
*/
force_sync = (journal->j_flags & JFS_SYNC) || handle->h_sync;
if (force_sync ||
transaction->t_outstanding_credits > journal->j_max_transaction_buffers ||
time_after_eq(jiffies, transaction->t_expires)) {
tid_t tid = transaction->t_tid;
jfs_debug(2, "transaction too old, requesting commit for handle %p\n", handle);
log_start_commit(journal, transaction);
/*
* Special case: JFS_SYNC synchronous updates require us
* to wait for the commit to complete.
*/
if (force_sync)
log_wait_commit(journal, tid);
}
kfree(handle);
return 0;
}
void journal_sync_buffer(struct buffer_head *bh)
{
transaction_t *transaction;
journal_t *journal;
long sequence;
/* If the buffer isn't journaled, this is easy: just sync it to
* disk. */
if (bh->b_transaction == NULL) {
/* If the buffer has already been journaled, then this
* is a noop. */
if (bh->b_cp_transaction == NULL)
return;
ll_rw_block (WRITE, 1, &bh);
wait_on_buffer (bh);
return;
}
/* Otherwise, just wait until the transaction is synced to disk. */
transaction = bh->b_transaction;
journal = transaction->t_journal;
sequence = transaction->t_tid;
jfs_debug(2, "requesting commit for bh %p\n", bh);
log_start_commit (journal, transaction);
while (tid_gt(sequence, journal->j_commit_sequence)) {
wake_up(&journal->j_wait_done_commit);
sleep_on(&journal->j_wait_done_commit);
}
}
struct buffer_head *getblk(kdev_t kdev, blk64_t blocknr, int blocksize)
{
struct buffer_head *bh;
int bufsize = sizeof(*bh) + kdev->k_ctx->fs->blocksize -
sizeof(bh->b_data);
bh = e2fsck_allocate_memory(kdev->k_ctx, bufsize, "block buffer");
if (!bh)
return NULL;
#ifdef CONFIG_JBD_DEBUG
if (journal_enable_debug >= 3)
bh_count++;
#endif
jfs_debug(4, "getblk for block %llu (%d bytes)(total %d)\n",
(unsigned long long) blocknr, blocksize, bh_count);
bh->b_ctx = kdev->k_ctx;
if (kdev->k_dev == K_DEV_FS)
bh->b_io = kdev->k_ctx->fs->io;
else
bh->b_io = kdev->k_ctx->journal_io;
bh->b_size = blocksize;
bh->b_blocknr = blocknr;
return bh;
}
struct buffer_head *getblk(kdev_t kdev, blk64_t blocknr, int blocksize)
{
struct buffer_head *bh;
int bufsize = sizeof(*bh) + kdev->k_fs->blocksize -
sizeof(bh->b_data);
errcode_t retval;
retval = ext2fs_get_memzero(bufsize, &bh);
if (retval)
return NULL;
#ifdef CONFIG_JBD_DEBUG
if (journal_enable_debug >= 3)
bh_count++;
#endif
jfs_debug(4, "getblk for block %llu (%d bytes)(total %d)\n",
(unsigned long long) blocknr, blocksize, bh_count);
bh->b_fs = kdev->k_fs;
if (kdev->k_dev == K_DEV_FS)
bh->b_io = kdev->k_fs->io;
else
bh->b_io = kdev->k_fs->journal_io;
bh->b_size = blocksize;
bh->b_blocknr = blocknr;
return bh;
}
int journal_skip_recovery(journal_t *journal)
{
int err;
journal_superblock_t * sb;
struct recovery_info info;
memset(&info, 0, sizeof(info));
sb = journal->j_superblock;
err = do_one_pass(journal, &info, PASS_SCAN);
if (err) {
printk(KERN_ERR "JFS: error %d scanning journal\n", err);
++journal->j_transaction_sequence;
} else {
int dropped = info.end_transaction - ntohl(sb->s_sequence);
jfs_debug(0,
"JFS: ignoring %d transaction%s from the journal.\n",
dropped, (dropped == 1) ? "" : "s");
journal->j_transaction_sequence = ++info.end_transaction;
}
journal->j_tail = 0;
return err;
}
void brelse(struct buffer_head *bh)
{
if (bh->b_dirty)
ll_rw_block(WRITE, 1, &bh);
jfs_debug(3, "freeing block %lu/%p (total %d)\n",
(unsigned long) bh->b_blocknr, (void *) bh, --bh_count);
ext2fs_free_mem(&bh);
}
void ll_rw_block(int rw, int nr, struct buffer_head *bhp[])
{
int retval;
struct buffer_head *bh;
for (; nr > 0; --nr) {
bh = *bhp++;
if (rw == READ && !bh->b_uptodate) {
jfs_debug(3, "reading block %llu/%p\n",
bh->b_blocknr, (void *) bh);
retval = io_channel_read_blk64(bh->b_io,
bh->b_blocknr,
1, bh->b_data);
if (retval) {
com_err(bh->b_ctx->device_name, retval,
"while reading block %llu\n",
bh->b_blocknr);
bh->b_err = retval;
continue;
}
bh->b_uptodate = 1;
} else if (rw == WRITE && bh->b_dirty) {
jfs_debug(3, "writing block %llu/%p\n",
bh->b_blocknr,
(void *) bh);
retval = io_channel_write_blk64(bh->b_io,
bh->b_blocknr,
1, bh->b_data);
if (retval) {
com_err(bh->b_ctx->device_name, retval,
"while writing block %llu\n",
bh->b_blocknr);
bh->b_err = retval;
continue;
}
bh->b_dirty = 0;
bh->b_uptodate = 1;
} else {
jfs_debug(3, "no-op %s for block %llu\n",
rw == READ ? "read" : "write",
bh->b_blocknr);
}
}
}
int journal_recover(journal_t *journal)
{
int err;
journal_superblock_t * sb;
struct recovery_info info;
memset(&info, 0, sizeof(info));
sb = journal->j_superblock;
/*
* The journal superblock's s_start field (the current log head)
* is always zero if, and only if, the journal was cleanly
* unmounted.
*/
if (!sb->s_start) {
jfs_debug(1, "No recovery required, last transaction %d\n",
ntohl(sb->s_sequence));
journal->j_transaction_sequence = ntohl(sb->s_sequence) + 1;
return 0;
}
err = do_one_pass(journal, &info, PASS_SCAN);
if (!err)
err = do_one_pass(journal, &info, PASS_REVOKE);
if (!err)
err = do_one_pass(journal, &info, PASS_REPLAY);
jfs_debug(0, "JFS: recovery, exit status %d, "
"recovered transactions %u to %u\n",
err, info.start_transaction, info.end_transaction);
jfs_debug(0, "JFS: Replayed %d and revoked %d/%d blocks\n",
info.nr_replays, info.nr_revoke_hits, info.nr_revokes);
/* Restart the log at the next transaction ID, thus invalidating
* any existing commit records in the log. */
journal->j_transaction_sequence = ++info.end_transaction;
journal_clear_revoke(journal);
fsync_dev(journal->j_dev);
return err;
}
int journal_extend (handle_t *handle, int nblocks)
{
transaction_t *transaction = handle->h_transaction;
journal_t *journal = transaction->t_journal;
int result = 1;
int wanted;
lock_journal (journal);
/* Don't extend a locked-down transaction! */
if (handle->h_transaction->t_state != T_RUNNING) {
jfs_debug(3, "denied handle %p %d blocks: "
"transaction not running\n", handle, nblocks);
goto error_out;
}
wanted = transaction->t_outstanding_credits + nblocks;
if (wanted > journal->j_max_transaction_buffers) {
jfs_debug(3, "denied handle %p %d blocks: "
"transaction too large\n", handle, nblocks);
goto error_out;
}
if (wanted > log_space_left(journal)) {
jfs_debug(3, "denied handle %p %d blocks: "
"insufficient log space\n", handle, nblocks);
goto error_out;
}
handle->h_buffer_credits += nblocks;
transaction->t_outstanding_credits += nblocks;
result = 0;
jfs_debug(3, "extended handle %p by %d\n", handle, nblocks);
error_out:
unlock_journal (journal);
return result;
}
struct buffer_head *getblk(kdev_t kdev, blk_t blocknr, int blocksize)
{
struct buffer_head *bh;
bh = e2fsck_allocate_memory(kdev->k_ctx, sizeof(*bh), "block buffer");
if (!bh)
return NULL;
jfs_debug(4, "getblk for block %lu (%d bytes)(total %d)\n",
(unsigned long) blocknr, blocksize, ++bh_count);
bh->b_ctx = kdev->k_ctx;
if (kdev->k_dev == K_DEV_FS)
bh->b_io = kdev->k_ctx->fs->io;
else
bh->b_io = kdev->k_ctx->journal_io;
bh->b_size = blocksize;
bh->b_blocknr = blocknr;
return bh;
}
int journal_restart(handle_t *handle, int nblocks)
{
transaction_t *transaction = handle->h_transaction;
journal_t *journal = transaction->t_journal;
/* First unlink the handle from its current transaction, and
* start the commit on that. */
J_ASSERT (transaction->t_updates > 0);
J_ASSERT (current->j_handle == handle);
transaction->t_outstanding_credits -= handle->h_buffer_credits;
transaction->t_updates--;
if (!transaction->t_updates)
wake_up(&journal->j_wait_updates);
jfs_debug(2, "restarting handle %p\n", handle);
log_start_commit(journal, transaction);
handle->h_buffer_credits = nblocks;
return start_this_handle(journal, handle);
}
int journal_get_create_access (handle_t *handle, struct buffer_head *bh)
{
transaction_t *transaction = handle->h_transaction;
journal_t *journal = transaction->t_journal;
jfs_debug(5, "buffer_head %p\n", bh);
lock_journal(journal);
/* The buffer may already belong to this transaction due to
pre-zeroing in the filesystem's new_block code */
J_ASSERT (bh->b_transaction == transaction || bh->b_transaction == NULL);
J_ASSERT (bh->b_next_transaction == NULL);
J_ASSERT (buffer_locked(bh));
J_ASSERT(handle->h_buffer_credits > 0);
handle->h_buffer_credits--;
bh->b_transaction = transaction;
journal_file_buffer(bh, transaction, BJ_Reserved);
refile_buffer(bh);
unlock_journal(journal);
return 0;
}
int journal_dirty_metadata (handle_t *handle, struct buffer_head *bh)
{
transaction_t *transaction = handle->h_transaction;
journal_t *journal = transaction->t_journal;
jfs_debug(5, "buffer_head %p\n", bh);
lock_journal(journal);
mark_buffer_jdirty(bh);
J_ASSERT(bh->b_transaction != NULL);
/*
* Metadata already on the current transaction list doesn't
* need to be filed. Metadata on another transaction's list must
* be committing, and will be refiled once the commit completes:
* leave it alone for now.
*/
if (bh->b_transaction != transaction) {
J_ASSERT (bh->b_transaction == journal->j_committing_transaction);
J_ASSERT (bh->b_next_transaction == transaction);
/* And this case is illegal: we can't reuse another
* transaction's data buffer, ever. */
J_ASSERT (bh->b_jlist != BJ_Data);
goto done;
}
/* That test should have eliminated the following case: */
J_ASSERT (bh->b_frozen_data == 0);
journal_file_buffer (bh, handle->h_transaction, BJ_Metadata);
done:
unlock_journal(journal);
return 0;
}
int do_get_write_access (handle_t *handle, struct buffer_head *bh,
int force_copy)
{
transaction_t *transaction = handle->h_transaction;
journal_t *journal = transaction->t_journal;
int error = 0;
char *frozen_buffer = NULL;
jfs_debug(5, "buffer_head %p, force_copy %d\n", bh, force_copy);
repeat:
/* @@@ Need to check for errors here at some point. */
/* The caller must make sure that we enter here with the buffer
* unlocked (probably by calling lock_journal_bh_wait). If we
* sleep in this function, we have to wait again for the buffer
* to make sure it is still unlocked. We cannot journal a
* buffer if somebody else may already be in the process of
* writing it to disk! */
J_ASSERT (buffer_uptodate(bh));
J_ASSERT (!buffer_locked(bh));
/* The buffer is already part of this transaction if
* b_transaction or b_next_transaction points to it. */
if (bh->b_transaction == transaction ||
bh->b_next_transaction == transaction)
goto done;
/* If there is already a copy-out version of this buffer, then
* we don't need to make another one. */
if (bh->b_frozen_data) {
J_ASSERT(bh->b_next_transaction == NULL);
bh->b_next_transaction = transaction;
J_ASSERT(handle->h_buffer_credits > 0);
handle->h_buffer_credits--;
goto done;
}
/* Is there data here we need to preserve? */
if (bh->b_transaction && bh->b_transaction != transaction) {
J_ASSERT (bh->b_next_transaction == NULL);
J_ASSERT (bh->b_transaction == journal->j_committing_transaction);
J_ASSERT (bh->b_list == BUF_JOURNAL);
/* There is one case we have to be very careful about.
* If the committing transaction is currently writing
* this buffer out to disk and has NOT made a copy-out,
* then we cannot modify the buffer contents at all
* right now. The essence of copy-out is that it is the
* extra copy, not the primary copy, which gets
* journaled. If the primary copy is already going to
* disk then we cannot do copy-out here. */
if (bh->b_jlist == BJ_Shadow) {
unlock_journal(journal);
/* commit wakes up all shadow buffers after IO */
sleep_on(&bh->b_wait);
lock_journal_bh_wait(bh, journal);
goto repeat;
}
/* Only do the copy if the currently-owning transaction
* still needs it. If it is on the Forget list, the
* committing transaction is past that stage. The
* buffer had better remain locked during the kmalloc,
* but that should be true --- we hold the journal lock
* still and the buffer is already on the BUF_JOURNAL
* list so won't be flushed.
*
* Subtle point, though: if this is a get_undo_access,
* then we will be relying on the frozen_data to contain
* the new value of the committed_data record after the
* transaction, so we HAVE to force the frozen_data copy
* in that case. */
if (bh->b_jlist != BJ_Forget || force_copy) {
if (bh->b_jlist == BJ_Data)
J_ASSERT(test_bit(BH_QuickFree, &bh->b_state));
if (!frozen_buffer) {
unlock_journal(journal);
frozen_buffer = kmalloc(bh->b_size,GFP_KERNEL);
lock_journal_bh_wait(bh, journal);
if (!frozen_buffer) {
error = -ENOMEM;
goto done;
}
goto repeat;
}
bh->b_frozen_data = frozen_buffer;
frozen_buffer = NULL;
memcpy (bh->b_frozen_data, bh->b_data, bh->b_size);
}
bh->b_next_transaction = transaction;
}
J_ASSERT(handle->h_buffer_credits > 0);
handle->h_buffer_credits--;
/* Finally, if the buffer is not journaled right now, we need to
* make sure it doesn't get written to disk before the caller
* actually commits the new data. */
if (!bh->b_transaction) {
J_ASSERT (!bh->b_next_transaction);
bh->b_transaction = transaction;
journal_file_buffer(bh, transaction, BJ_Reserved);
}
done:
if (bh->b_list != BUF_JOURNAL)
refile_buffer(bh);
clear_bit(BH_QuickFree, &bh->b_state);
/* If we are about to journal a buffer, then any revoke pending
on it is no longer valid. */
journal_cancel_revoke(handle, bh);
if (frozen_buffer)
kfree(frozen_buffer);
return error;
}
static errcode_t e2fsck_get_journal(e2fsck_t ctx, journal_t **ret_journal)
{
struct ext2_super_block *sb = ctx->fs->super;
struct ext2_super_block jsuper;
struct problem_context pctx;
struct buffer_head *bh;
struct inode *j_inode = NULL;
struct kdev_s *dev_fs = NULL, *dev_journal;
const char *journal_name = 0;
journal_t *journal = NULL;
errcode_t retval = 0;
io_manager io_ptr = 0;
unsigned long start = 0;
blk_t blk;
int ext_journal = 0;
int tried_backup_jnl = 0;
int i;
clear_problem_context(&pctx);
journal = e2fsck_allocate_memory(ctx, sizeof(journal_t), "journal");
if (!journal) {
return EXT2_ET_NO_MEMORY;
}
dev_fs = e2fsck_allocate_memory(ctx, 2*sizeof(struct kdev_s), "kdev");
if (!dev_fs) {
retval = EXT2_ET_NO_MEMORY;
goto errout;
}
dev_journal = dev_fs+1;
dev_fs->k_ctx = dev_journal->k_ctx = ctx;
dev_fs->k_dev = K_DEV_FS;
dev_journal->k_dev = K_DEV_JOURNAL;
journal->j_dev = dev_journal;
journal->j_fs_dev = dev_fs;
journal->j_inode = NULL;
journal->j_blocksize = ctx->fs->blocksize;
if (uuid_is_null(sb->s_journal_uuid)) {
if (!sb->s_journal_inum)
return EXT2_ET_BAD_INODE_NUM;
j_inode = e2fsck_allocate_memory(ctx, sizeof(*j_inode),
"journal inode");
if (!j_inode) {
retval = EXT2_ET_NO_MEMORY;
goto errout;
}
j_inode->i_ctx = ctx;
j_inode->i_ino = sb->s_journal_inum;
if ((retval = ext2fs_read_inode(ctx->fs,
sb->s_journal_inum,
&j_inode->i_ext2))) {
try_backup_journal:
if (sb->s_jnl_backup_type != EXT3_JNL_BACKUP_BLOCKS ||
tried_backup_jnl)
goto errout;
memset(&j_inode->i_ext2, 0, sizeof(struct ext2_inode));
memcpy(&j_inode->i_ext2.i_block[0], sb->s_jnl_blocks,
EXT2_N_BLOCKS*4);
j_inode->i_ext2.i_size = sb->s_jnl_blocks[16];
j_inode->i_ext2.i_links_count = 1;
j_inode->i_ext2.i_mode = LINUX_S_IFREG | 0600;
tried_backup_jnl++;
}
if (!j_inode->i_ext2.i_links_count ||
!LINUX_S_ISREG(j_inode->i_ext2.i_mode)) {
retval = EXT2_ET_NO_JOURNAL;
goto try_backup_journal;
}
if (j_inode->i_ext2.i_size / journal->j_blocksize <
JFS_MIN_JOURNAL_BLOCKS) {
retval = EXT2_ET_JOURNAL_TOO_SMALL;
goto try_backup_journal;
}
for (i=0; i < EXT2_N_BLOCKS; i++) {
blk = j_inode->i_ext2.i_block[i];
if (!blk) {
if (i < EXT2_NDIR_BLOCKS) {
retval = EXT2_ET_JOURNAL_TOO_SMALL;
goto try_backup_journal;
}
continue;
}
if (blk < sb->s_first_data_block ||
blk >= sb->s_blocks_count) {
retval = EXT2_ET_BAD_BLOCK_NUM;
goto try_backup_journal;
}
}
journal->j_maxlen = j_inode->i_ext2.i_size / journal->j_blocksize;
#ifdef USE_INODE_IO
retval = ext2fs_inode_io_intern2(ctx->fs, sb->s_journal_inum,
&j_inode->i_ext2,
&journal_name);
if (retval)
goto errout;
io_ptr = inode_io_manager;
#else
journal->j_inode = j_inode;
ctx->journal_io = ctx->fs->io;
if ((retval = journal_bmap(journal, 0, &start)) != 0)
goto errout;
#endif
} else {
ext_journal = 1;
if (!ctx->journal_name) {
char uuid[37];
uuid_unparse(sb->s_journal_uuid, uuid);
ctx->journal_name = blkid_get_devname(ctx->blkid,
"UUID", uuid);
if (!ctx->journal_name)
ctx->journal_name = blkid_devno_to_devname(sb->s_journal_dev);
}
journal_name = ctx->journal_name;
if (!journal_name) {
fix_problem(ctx, PR_0_CANT_FIND_JOURNAL, &pctx);
return EXT2_ET_LOAD_EXT_JOURNAL;
}
jfs_debug(1, "Using journal file %s\n", journal_name);
io_ptr = unix_io_manager;
}
#if 0
test_io_backing_manager = io_ptr;
io_ptr = test_io_manager;
#endif
#ifndef USE_INODE_IO
if (ext_journal)
#endif
retval = io_ptr->open(journal_name, IO_FLAG_RW,
&ctx->journal_io);
if (retval)
goto errout;
io_channel_set_blksize(ctx->journal_io, ctx->fs->blocksize);
if (ext_journal) {
if (ctx->fs->blocksize == 1024)
start = 1;
bh = getblk(dev_journal, start, ctx->fs->blocksize);
if (!bh) {
retval = EXT2_ET_NO_MEMORY;
goto errout;
}
ll_rw_block(READ, 1, &bh);
if ((retval = bh->b_err) != 0) {
brelse(bh);
goto errout;
}
memcpy(&jsuper, start ? bh->b_data : bh->b_data + 1024,
sizeof(jsuper));
brelse(bh);
#ifdef EXT2FS_ENABLE_SWAPFS
if (jsuper.s_magic == ext2fs_swab16(EXT2_SUPER_MAGIC))
ext2fs_swap_super(&jsuper);
#endif
if (jsuper.s_magic != EXT2_SUPER_MAGIC ||
!(jsuper.s_feature_incompat & EXT3_FEATURE_INCOMPAT_JOURNAL_DEV)) {
fix_problem(ctx, PR_0_EXT_JOURNAL_BAD_SUPER, &pctx);
retval = EXT2_ET_LOAD_EXT_JOURNAL;
goto errout;
}
/* Make sure the journal UUID is correct */
if (memcmp(jsuper.s_uuid, ctx->fs->super->s_journal_uuid,
sizeof(jsuper.s_uuid))) {
fix_problem(ctx, PR_0_JOURNAL_BAD_UUID, &pctx);
retval = EXT2_ET_LOAD_EXT_JOURNAL;
goto errout;
}
journal->j_maxlen = jsuper.s_blocks_count;
start++;
}
if (!(bh = getblk(dev_journal, start, journal->j_blocksize))) {
retval = EXT2_ET_NO_MEMORY;
goto errout;
}
journal->j_sb_buffer = bh;
journal->j_superblock = (journal_superblock_t *)bh->b_data;
#ifdef USE_INODE_IO
if (j_inode)
ext2fs_free_mem(&j_inode);
#endif
*ret_journal = journal;
return 0;
errout:
if (dev_fs)
ext2fs_free_mem(&dev_fs);
if (j_inode)
ext2fs_free_mem(&j_inode);
if (journal)
ext2fs_free_mem(&journal);
return retval;
}
static int do_one_pass(journal_t *journal, struct recovery_info *info,
enum passtype pass)
{
unsigned int first_commit_ID, next_commit_ID;
unsigned long next_log_block;
int err, success = 0;
journal_superblock_t * sb;
journal_header_t * tmp;
struct buffer_head * bh;
unsigned int sequence;
int blocktype;
/* Precompute the maximum metadata descriptors in a descriptor block */
int MAX_BLOCKS_PER_DESC;
MAX_BLOCKS_PER_DESC = ((journal->j_blocksize-sizeof(journal_header_t))
/ sizeof(journal_block_tag_t));
/*
* First thing is to establish what we expect to find in the log
* (in terms of transaction IDs), and where (in terms of log
* block offsets): query the superblock.
*/
sb = journal->j_superblock;
next_commit_ID = ntohl(sb->s_sequence);
next_log_block = ntohl(sb->s_start);
first_commit_ID = next_commit_ID;
if (pass == PASS_SCAN)
info->start_transaction = first_commit_ID;
jfs_debug(1, "Starting recovery pass %d\n", pass);
/*
* Now we walk through the log, transaction by transaction,
* making sure that each transaction has a commit block in the
* expected place. Each complete transaction gets replayed back
* into the main filesystem.
*/
while (1) {
int flags;
char * tagp;
journal_block_tag_t * tag;
struct buffer_head * obh;
struct buffer_head * nbh;
/* If we already know where to stop the log traversal,
* check right now that we haven't gone past the end of
* the log. */
if (pass != PASS_SCAN)
if (tid_geq(next_commit_ID, info->end_transaction))
break;
jfs_debug(2, "Scanning for sequence ID %u at %lu/%lu\n",
next_commit_ID, next_log_block, journal->j_last);
/* Skip over each chunk of the transaction looking
* either the next descriptor block or the final commit
* record. */
jfs_debug(3, "JFS: checking block %ld\n", next_log_block);
err = jread(&bh, journal, next_log_block);
if (err)
goto failed;
next_log_block++;
wrap(journal, next_log_block);
/* What kind of buffer is it?
*
* If it is a descriptor block, check that it has the
* expected sequence number. Otherwise, we're all done
* here. */
tmp = (journal_header_t *) bh->b_data;
if (tmp->h_magic != htonl(JFS_MAGIC_NUMBER)) {
brelse(bh);
break;
}
blocktype = ntohl(tmp->h_blocktype);
sequence = ntohl(tmp->h_sequence);
jfs_debug(3, "Found magic %d, sequence %d\n",
blocktype, sequence);
if (sequence != next_commit_ID) {
brelse(bh);
break;
}
/* OK, we have a valid descriptor block which matches
* all of the sequence number checks. What are we going
* to do with it? That depends on the pass... */
switch(blocktype) {
case JFS_DESCRIPTOR_BLOCK:
/* If it is a valid descriptor block, replay it
* in pass REPLAY; otherwise, just skip over the
* blocks it describes. */
if (pass != PASS_REPLAY) {
next_log_block += count_tags(bh, journal->j_blocksize);
wrap(journal, next_log_block);
brelse(bh);
continue;
}
/* A descriptor block: we can now write all of
* the data blocks. Yay, useful work is finally
* getting done here! */
tagp = &bh->b_data[sizeof(journal_header_t)];
while ((tagp - bh->b_data +sizeof(journal_block_tag_t))
<= journal->j_blocksize) {
unsigned long io_block;
tag = (journal_block_tag_t *) tagp;
flags = ntohl(tag->t_flags);
io_block = next_log_block++;
wrap(journal, next_log_block);
err = jread(&obh, journal, io_block);
if (err) {
/* Recover what we can, but
* report failure at the end. */
success = err;
printk (KERN_ERR
"JFS: IO error %d recovering "
"block %ld in log\n",
err, io_block);
} else {
unsigned long blocknr;
J_ASSERT(obh != NULL);
blocknr = ntohl(tag->t_blocknr);
/* If the block has been
* revoked, then we're all done
* here. */
if (journal_test_revoke
(journal, blocknr,
next_commit_ID)) {
brelse(obh);
++info->nr_revoke_hits;
goto skip_write;
}
/* Find a buffer for the new
* data being restored */
nbh = getblk(journal->j_dev, blocknr,
journal->j_blocksize);
if (nbh == NULL) {
printk(KERN_ERR
"JFS: Out of memory "
"during recovery.\n");
err = -ENOMEM;
brelse(bh);
brelse(obh);
goto failed;
}
memcpy(nbh->b_data, obh->b_data,
journal->j_blocksize);
if (flags & JFS_FLAG_ESCAPE) {
* ((unsigned int *) bh->b_data) = htonl(JFS_MAGIC_NUMBER);
}
mark_buffer_dirty(nbh, 1);
mark_buffer_uptodate(nbh, 1);
++info->nr_replays;
/* ll_rw_block(WRITE, 1, &nbh); */
brelse(obh);
brelse(nbh);
}
skip_write:
tagp += sizeof(journal_block_tag_t);
if (!(flags & JFS_FLAG_SAME_UUID))
tagp += 16;
if (flags & JFS_FLAG_LAST_TAG)
break;
}
brelse(bh);
continue;
case JFS_COMMIT_BLOCK:
/* Found an expected commit block: not much to
* do other than move on to the next sequence
* number. */
brelse(bh);
next_commit_ID++;
continue;
case JFS_REVOKE_BLOCK:
/* If we aren't in the REVOKE pass, then we can
* just skip over this block. */
if (pass != PASS_REVOKE) {
brelse(bh);
continue;
}
err = scan_revoke_records(journal, bh,
next_commit_ID, info);
brelse(bh);
if (err)
goto failed;
continue;
default:
jfs_debug(3, "Unrecognised magic %d, end of scan.\n",
blocktype);
goto done;
}
}
done:
/*
* We broke out of the log scan loop: either we came to the
* known end of the log or we found an unexpected block in the
* log. If the latter happened, then we know that the "current"
* transaction marks the end of the valid log.
*/
if (pass == PASS_SCAN)
info->end_transaction = next_commit_ID;
else {
/* It's really bad news if different passes end up at
* different places (but possible due to IO errors). */
if (info->end_transaction != next_commit_ID) {
printk (KERN_ERR "JFS: recovery pass %d ended at "
"transaction %u, expected %u\n",
pass, next_commit_ID, info->end_transaction);
if (!success)
success = -EIO;
}
}
return success;
failed:
return err;
}
static errcode_t ext2fs_get_journal(ext2_filsys fs, journal_t **ret_journal)
{
struct process_block_struct pb;
struct ext2_super_block *sb = fs->super;
struct ext2_super_block jsuper;
struct buffer_head *bh;
struct inode *j_inode = NULL;
struct kdev_s *dev_fs = NULL, *dev_journal;
const char *journal_name = 0;
journal_t *journal = NULL;
errcode_t retval = 0;
io_manager io_ptr = 0;
unsigned long long start = 0;
int ext_journal = 0;
int tried_backup_jnl = 0;
retval = ext2fs_get_memzero(sizeof(journal_t), &journal);
if (retval)
return retval;
retval = ext2fs_get_memzero(2 * sizeof(struct kdev_s), &dev_fs);
if (retval)
goto errout;
dev_journal = dev_fs+1;
dev_fs->k_fs = dev_journal->k_fs = fs;
dev_fs->k_dev = K_DEV_FS;
dev_journal->k_dev = K_DEV_JOURNAL;
journal->j_dev = dev_journal;
journal->j_fs_dev = dev_fs;
journal->j_inode = NULL;
journal->j_blocksize = fs->blocksize;
if (uuid_is_null(sb->s_journal_uuid)) {
if (!sb->s_journal_inum) {
retval = EXT2_ET_BAD_INODE_NUM;
goto errout;
}
retval = ext2fs_get_memzero(sizeof(*j_inode), &j_inode);
if (retval)
goto errout;
j_inode->i_fs = fs;
j_inode->i_ino = sb->s_journal_inum;
retval = ext2fs_read_inode(fs, sb->s_journal_inum,
&j_inode->i_ext2);
if (retval) {
try_backup_journal:
if (sb->s_jnl_backup_type != EXT3_JNL_BACKUP_BLOCKS ||
tried_backup_jnl)
goto errout;
memset(&j_inode->i_ext2, 0, sizeof(struct ext2_inode));
memcpy(&j_inode->i_ext2.i_block[0], sb->s_jnl_blocks,
EXT2_N_BLOCKS*4);
j_inode->i_ext2.i_size_high = sb->s_jnl_blocks[15];
j_inode->i_ext2.i_size = sb->s_jnl_blocks[16];
j_inode->i_ext2.i_links_count = 1;
j_inode->i_ext2.i_mode = LINUX_S_IFREG | 0600;
tried_backup_jnl++;
}
if (!j_inode->i_ext2.i_links_count ||
!LINUX_S_ISREG(j_inode->i_ext2.i_mode)) {
retval = EXT2_ET_NO_JOURNAL;
goto try_backup_journal;
}
if (EXT2_I_SIZE(&j_inode->i_ext2) / journal->j_blocksize <
JFS_MIN_JOURNAL_BLOCKS) {
retval = EXT2_ET_JOURNAL_TOO_SMALL;
goto try_backup_journal;
}
pb.last_block = -1;
retval = ext2fs_block_iterate3(fs, j_inode->i_ino,
BLOCK_FLAG_HOLE, 0,
process_journal_block, &pb);
if ((pb.last_block + 1) * fs->blocksize <
(int) EXT2_I_SIZE(&j_inode->i_ext2)) {
retval = EXT2_ET_JOURNAL_TOO_SMALL;
goto try_backup_journal;
}
if (tried_backup_jnl && (fs->flags & EXT2_FLAG_RW)) {
retval = ext2fs_write_inode(fs, sb->s_journal_inum,
&j_inode->i_ext2);
if (retval)
goto errout;
}
journal->j_maxlen = EXT2_I_SIZE(&j_inode->i_ext2) /
journal->j_blocksize;
#ifdef USE_INODE_IO
retval = ext2fs_inode_io_intern2(fs, sb->s_journal_inum,
&j_inode->i_ext2,
&journal_name);
if (retval)
goto errout;
io_ptr = inode_io_manager;
#else
journal->j_inode = j_inode;
fs->journal_io = fs->io;
retval = (errcode_t)journal_bmap(journal, 0, &start);
if (retval)
goto errout;
#endif
} else {
ext_journal = 1;
if (!fs->journal_name) {
char uuid[37];
blkid_cache blkid;
blkid_get_cache(&blkid, NULL);
uuid_unparse(sb->s_journal_uuid, uuid);
fs->journal_name = blkid_get_devname(blkid,
"UUID", uuid);
if (!fs->journal_name)
fs->journal_name = blkid_devno_to_devname(sb->s_journal_dev);
blkid_put_cache(blkid);
}
journal_name = fs->journal_name;
if (!journal_name) {
retval = EXT2_ET_LOAD_EXT_JOURNAL;
goto errout;
}
jfs_debug(1, "Using journal file %s\n", journal_name);
io_ptr = unix_io_manager;
}
#if 0
test_io_backing_manager = io_ptr;
io_ptr = test_io_manager;
#endif
#ifndef USE_INODE_IO
if (ext_journal)
#endif
{
retval = io_ptr->open(journal_name, fs->flags & EXT2_FLAG_RW,
&fs->journal_io);
}
if (retval)
goto errout;
io_channel_set_blksize(fs->journal_io, fs->blocksize);
if (ext_journal) {
blk64_t maxlen;
start = ext2fs_journal_sb_start(fs->blocksize) - 1;
bh = getblk(dev_journal, start, fs->blocksize);
if (!bh) {
retval = EXT2_ET_NO_MEMORY;
goto errout;
}
ll_rw_block(READ, 1, &bh);
retval = bh->b_err;
if (retval) {
brelse(bh);
goto errout;
}
memcpy(&jsuper, start ? bh->b_data :
bh->b_data + SUPERBLOCK_OFFSET,
sizeof(jsuper));
#ifdef WORDS_BIGENDIAN
if (jsuper.s_magic == ext2fs_swab16(EXT2_SUPER_MAGIC))
ext2fs_swap_super(&jsuper);
#endif
if (jsuper.s_magic != EXT2_SUPER_MAGIC ||
!ext2fs_has_feature_journal_dev(&jsuper)) {
retval = EXT2_ET_LOAD_EXT_JOURNAL;
brelse(bh);
goto errout;
}
/* Make sure the journal UUID is correct */
if (memcmp(jsuper.s_uuid, fs->super->s_journal_uuid,
sizeof(jsuper.s_uuid))) {
retval = EXT2_ET_LOAD_EXT_JOURNAL;
brelse(bh);
goto errout;
}
/* Check the superblock checksum */
if (ext2fs_has_feature_metadata_csum(&jsuper)) {
struct struct_ext2_filsys fsx;
struct ext2_super_block superx;
void *p;
p = start ? bh->b_data : bh->b_data + SUPERBLOCK_OFFSET;
memcpy(&fsx, fs, sizeof(fsx));
memcpy(&superx, fs->super, sizeof(superx));
fsx.super = &superx;
ext2fs_set_feature_metadata_csum(fsx.super);
if (!ext2fs_superblock_csum_verify(&fsx, p)) {
retval = EXT2_ET_LOAD_EXT_JOURNAL;
brelse(bh);
goto errout;
}
}
brelse(bh);
maxlen = ext2fs_blocks_count(&jsuper);
journal->j_maxlen = (maxlen < 1ULL << 32) ? maxlen :
(1ULL << 32) - 1;
start++;
}
bh = getblk(dev_journal, start, journal->j_blocksize);
if (!bh) {
retval = EXT2_ET_NO_MEMORY;
goto errout;
}
journal->j_sb_buffer = bh;
journal->j_superblock = (journal_superblock_t *)bh->b_data;
#ifdef USE_INODE_IO
if (j_inode)
ext2fs_free_mem(&j_inode);
#endif
*ret_journal = journal;
return 0;
errout:
if (dev_fs)
ext2fs_free_mem(&dev_fs);
if (j_inode)
ext2fs_free_mem(&j_inode);
if (journal)
ext2fs_free_mem(&journal);
return retval;
}
static errcode_t e2fsck_get_journal(e2fsck_t ctx, journal_t **ret_journal)
{
struct process_block_struct pb;
struct ext2_super_block *sb = ctx->fs->super;
struct ext2_super_block jsuper;
struct problem_context pctx;
struct buffer_head *bh;
struct inode *j_inode = NULL;
struct kdev_s *dev_fs = NULL, *dev_journal;
const char *journal_name = 0;
journal_t *journal = NULL;
errcode_t retval = 0;
io_manager io_ptr = 0;
unsigned long long start = 0;
int ext_journal = 0;
int tried_backup_jnl = 0;
clear_problem_context(&pctx);
journal = e2fsck_allocate_memory(ctx, sizeof(journal_t), "journal");
if (!journal) {
return EXT2_ET_NO_MEMORY;
}
dev_fs = e2fsck_allocate_memory(ctx, 2*sizeof(struct kdev_s), "kdev");
if (!dev_fs) {
retval = EXT2_ET_NO_MEMORY;
goto errout;
}
dev_journal = dev_fs+1;
dev_fs->k_ctx = dev_journal->k_ctx = ctx;
dev_fs->k_dev = K_DEV_FS;
dev_journal->k_dev = K_DEV_JOURNAL;
journal->j_dev = dev_journal;
journal->j_fs_dev = dev_fs;
journal->j_inode = NULL;
journal->j_blocksize = ctx->fs->blocksize;
if (uuid_is_null(sb->s_journal_uuid)) {
if (!sb->s_journal_inum) {
retval = EXT2_ET_BAD_INODE_NUM;
goto errout;
}
j_inode = e2fsck_allocate_memory(ctx, sizeof(*j_inode),
"journal inode");
if (!j_inode) {
retval = EXT2_ET_NO_MEMORY;
goto errout;
}
j_inode->i_ctx = ctx;
j_inode->i_ino = sb->s_journal_inum;
if ((retval = ext2fs_read_inode(ctx->fs,
sb->s_journal_inum,
&j_inode->i_ext2))) {
try_backup_journal:
if (sb->s_jnl_backup_type != EXT3_JNL_BACKUP_BLOCKS ||
tried_backup_jnl)
goto errout;
memset(&j_inode->i_ext2, 0, sizeof(struct ext2_inode));
memcpy(&j_inode->i_ext2.i_block[0], sb->s_jnl_blocks,
EXT2_N_BLOCKS*4);
j_inode->i_ext2.i_size_high = sb->s_jnl_blocks[15];
j_inode->i_ext2.i_size = sb->s_jnl_blocks[16];
j_inode->i_ext2.i_links_count = 1;
j_inode->i_ext2.i_mode = LINUX_S_IFREG | 0600;
e2fsck_use_inode_shortcuts(ctx, 1);
ctx->stashed_ino = j_inode->i_ino;
ctx->stashed_inode = &j_inode->i_ext2;
tried_backup_jnl++;
}
if (!j_inode->i_ext2.i_links_count ||
!LINUX_S_ISREG(j_inode->i_ext2.i_mode)) {
retval = EXT2_ET_NO_JOURNAL;
goto try_backup_journal;
}
if (EXT2_I_SIZE(&j_inode->i_ext2) / journal->j_blocksize <
JFS_MIN_JOURNAL_BLOCKS) {
retval = EXT2_ET_JOURNAL_TOO_SMALL;
goto try_backup_journal;
}
pb.last_block = -1;
retval = ext2fs_block_iterate3(ctx->fs, j_inode->i_ino,
BLOCK_FLAG_HOLE, 0,
process_journal_block, &pb);
if ((pb.last_block + 1) * ctx->fs->blocksize <
(int) EXT2_I_SIZE(&j_inode->i_ext2)) {
retval = EXT2_ET_JOURNAL_TOO_SMALL;
goto try_backup_journal;
}
if (tried_backup_jnl && !(ctx->options & E2F_OPT_READONLY)) {
retval = ext2fs_write_inode(ctx->fs, sb->s_journal_inum,
&j_inode->i_ext2);
if (retval)
goto errout;
}
journal->j_maxlen = EXT2_I_SIZE(&j_inode->i_ext2) /
journal->j_blocksize;
#ifdef USE_INODE_IO
retval = ext2fs_inode_io_intern2(ctx->fs, sb->s_journal_inum,
&j_inode->i_ext2,
&journal_name);
if (retval)
goto errout;
io_ptr = inode_io_manager;
#else
journal->j_inode = j_inode;
ctx->journal_io = ctx->fs->io;
if ((retval = (errcode_t) journal_bmap(journal, 0, &start)) != 0)
goto errout;
#endif
} else {
ext_journal = 1;
if (!ctx->journal_name) {
char uuid[37];
uuid_unparse(sb->s_journal_uuid, uuid);
ctx->journal_name = blkid_get_devname(ctx->blkid,
"UUID", uuid);
if (!ctx->journal_name)
ctx->journal_name = blkid_devno_to_devname(sb->s_journal_dev);
}
journal_name = ctx->journal_name;
if (!journal_name) {
fix_problem(ctx, PR_0_CANT_FIND_JOURNAL, &pctx);
retval = EXT2_ET_LOAD_EXT_JOURNAL;
goto errout;
}
jfs_debug(1, "Using journal file %s\n", journal_name);
io_ptr = unix_io_manager;
}
#if 0
test_io_backing_manager = io_ptr;
io_ptr = test_io_manager;
#endif
#ifndef USE_INODE_IO
if (ext_journal)
#endif
{
int flags = IO_FLAG_RW;
if (!(ctx->mount_flags & EXT2_MF_ISROOT &&
ctx->mount_flags & EXT2_MF_READONLY))
flags |= IO_FLAG_EXCLUSIVE;
if ((ctx->mount_flags & EXT2_MF_READONLY) &&
(ctx->options & E2F_OPT_FORCE))
flags &= ~IO_FLAG_EXCLUSIVE;
retval = io_ptr->open(journal_name, flags,
&ctx->journal_io);
}
if (retval)
goto errout;
io_channel_set_blksize(ctx->journal_io, ctx->fs->blocksize);
if (ext_journal) {
blk64_t maxlen;
start = ext2fs_journal_sb_start(ctx->fs->blocksize) - 1;
bh = getblk(dev_journal, start, ctx->fs->blocksize);
if (!bh) {
retval = EXT2_ET_NO_MEMORY;
goto errout;
}
ll_rw_block(READ, 1, &bh);
if ((retval = bh->b_err) != 0) {
brelse(bh);
goto errout;
}
memcpy(&jsuper, start ? bh->b_data : bh->b_data + SUPERBLOCK_OFFSET,
sizeof(jsuper));
#ifdef WORDS_BIGENDIAN
if (jsuper.s_magic == ext2fs_swab16(EXT2_SUPER_MAGIC))
ext2fs_swap_super(&jsuper);
#endif
if (jsuper.s_magic != EXT2_SUPER_MAGIC ||
!(jsuper.s_feature_incompat & EXT3_FEATURE_INCOMPAT_JOURNAL_DEV)) {
fix_problem(ctx, PR_0_EXT_JOURNAL_BAD_SUPER, &pctx);
retval = EXT2_ET_LOAD_EXT_JOURNAL;
brelse(bh);
goto errout;
}
/* Make sure the journal UUID is correct */
if (memcmp(jsuper.s_uuid, ctx->fs->super->s_journal_uuid,
sizeof(jsuper.s_uuid))) {
fix_problem(ctx, PR_0_JOURNAL_BAD_UUID, &pctx);
retval = EXT2_ET_LOAD_EXT_JOURNAL;
brelse(bh);
goto errout;
}
/* Check the superblock checksum */
if (jsuper.s_feature_ro_compat &
EXT4_FEATURE_RO_COMPAT_METADATA_CSUM) {
struct struct_ext2_filsys fsx;
struct ext2_super_block superx;
void *p;
p = start ? bh->b_data : bh->b_data + SUPERBLOCK_OFFSET;
memcpy(&fsx, ctx->fs, sizeof(fsx));
memcpy(&superx, ctx->fs->super, sizeof(superx));
fsx.super = &superx;
fsx.super->s_feature_ro_compat |=
EXT4_FEATURE_RO_COMPAT_METADATA_CSUM;
if (!ext2fs_superblock_csum_verify(&fsx, p) &&
fix_problem(ctx, PR_0_EXT_JOURNAL_SUPER_CSUM_INVALID,
&pctx)) {
ext2fs_superblock_csum_set(&fsx, p);
mark_buffer_dirty(bh);
}
}
brelse(bh);
maxlen = ext2fs_blocks_count(&jsuper);
journal->j_maxlen = (maxlen < 1ULL << 32) ? maxlen : (1ULL << 32) - 1;
start++;
}
if (!(bh = getblk(dev_journal, start, journal->j_blocksize))) {
retval = EXT2_ET_NO_MEMORY;
goto errout;
}
journal->j_sb_buffer = bh;
journal->j_superblock = (journal_superblock_t *)bh->b_data;
#ifdef USE_INODE_IO
if (j_inode)
ext2fs_free_mem(&j_inode);
#endif
*ret_journal = journal;
e2fsck_use_inode_shortcuts(ctx, 0);
return 0;
errout:
e2fsck_use_inode_shortcuts(ctx, 0);
if (dev_fs)
ext2fs_free_mem(&dev_fs);
if (j_inode)
ext2fs_free_mem(&j_inode);
if (journal)
ext2fs_free_mem(&journal);
return retval;
}
static int start_this_handle(journal_t *journal, handle_t *handle)
{
transaction_t *transaction;
int needed;
int nblocks = handle->h_buffer_credits;
jfs_debug(4, "New handle %p going live.\n", handle);
repeat:
lock_journal(journal);
if ((journal->j_flags & JFS_ABORT) ||
(journal->j_errno != 0 && !(journal->j_flags & JFS_ACK_ERR))) {
unlock_journal(journal);
return -EROFS;
}
/* Wait on the journal's transaction barrier if necessary */
if (journal->j_barrier_count) {
unlock_journal(journal);
sleep_on(&journal->j_wait_transaction_locked);
goto repeat;
}
repeat_locked:
if (!journal->j_running_transaction)
get_transaction(journal);
/* @@@ Error? */
J_ASSERT(journal->j_running_transaction);
transaction = journal->j_running_transaction;
/* If the current transaction is locked down for commit, wait
* for the lock to be released. */
if (transaction->t_state == T_LOCKED) {
unlock_journal(journal);
jfs_debug(3, "Handle %p stalling...\n", handle);
sleep_on(&journal->j_wait_transaction_locked);
goto repeat;
}
/* If there is not enough space left in the log to write all
* potential buffers requested by this operation, we need to
* stall pending a log checkpoint to free some more log
* space. */
needed = transaction->t_outstanding_credits + nblocks;
if (needed > journal->j_max_transaction_buffers) {
/* If the current transaction is already too large, then
* start to commit it: we can then go back and attach
* this handle to a new transaction. */
jfs_debug(2, "Handle %p starting new commit...\n", handle);
log_start_commit(journal, transaction);
unlock_journal(journal);
sleep_on(&journal->j_wait_transaction_locked);
lock_journal(journal);
goto repeat_locked;
}
/*
* The commit code assumes that it can get enough log space
* without forcing a checkpoint. This is *critical* for
* correctness: a checkpoint of a buffer which is also
* associated with a committing transaction creates a deadlock,
* so commit simply cannot force through checkpoints.
*
* We must therefore ensure the necessary space in the journal
* *before* starting to dirty potentially checkpointed buffers
* in the new transaction.
*
* The worst part is, any transaction currently committing can
* reduce the free space arbitrarily. Be careful to account for
* those buffers when checkpointing.
*/
needed = journal->j_max_transaction_buffers;
if (journal->j_committing_transaction)
needed += journal->j_committing_transaction->t_outstanding_credits;
if (log_space_left(journal) < needed) {
jfs_debug(2, "Handle %p waiting for checkpoint...\n", handle);
log_wait_for_space(journal, needed);
goto repeat_locked;
}
/* OK, account for the buffers that this operation expects to
* use and add the handle to the running transaction. */
handle->h_transaction = transaction;
transaction->t_outstanding_credits += nblocks;
transaction->t_updates++;
jfs_debug(4, "Handle %p given %d credits (total %d, free %d)\n",
handle, nblocks, transaction->t_outstanding_credits,
log_space_left(journal));
unlock_journal(journal);
return 0;
}
