static void print_journal_information(ext2_filsys fs)
{
errcode_t retval;
char buf[1024];
char str[80];
unsigned int i;
journal_superblock_t *jsb;
/* Get the journal superblock */
if ((retval = io_channel_read_blk64(fs->io,
fs->super->s_first_data_block + 1,
-1024, buf))) {
com_err(program_name, retval, "%s",
_("while reading journal superblock"));
exit(1);
}
jsb = (journal_superblock_t *) buf;
if ((jsb->s_header.h_magic != (unsigned) ntohl(JFS_MAGIC_NUMBER)) ||
(jsb->s_header.h_blocktype !=
(unsigned) ntohl(JFS_SUPERBLOCK_V2))) {
com_err(program_name, 0, "%s",
_("Couldn't find journal superblock magic numbers"));
exit(1);
}
if (jsb->s_feature_compat &
ext2fs_cpu_to_be32(JFS_FEATURE_COMPAT_CHECKSUM))
printf(_("Journal checksum type: crc32\n"));
if (jsb->s_feature_incompat &
ext2fs_cpu_to_be32(JFS_FEATURE_INCOMPAT_CSUM_V2))
printf(_("Journal checksum type: %s\n"
"Journal checksum: 0x%08x\n"),
journal_checksum_type_str(jsb->s_checksum_type),
ext2fs_be32_to_cpu(jsb->s_checksum));
printf(_("\nJournal block size: %u\n"
"Journal length: %u\n"
"Journal first block: %u\n"
"Journal sequence: 0x%08x\n"
"Journal start: %u\n"
"Journal number of users: %u\n"),
(unsigned int)ntohl(jsb->s_blocksize), (unsigned int)ntohl(jsb->s_maxlen),
(unsigned int)ntohl(jsb->s_first), (unsigned int)ntohl(jsb->s_sequence),
(unsigned int)ntohl(jsb->s_start), (unsigned int)ntohl(jsb->s_nr_users));
for (i=0; i < ntohl(jsb->s_nr_users); i++) {
uuid_unparse(&jsb->s_users[i*16], str);
printf(i ? " %s\n"
: _("Journal users: %s\n"),
str);
}
}
static errcode_t journal_close_trans(journal_transaction_t *trans)
{
journal_t *journal;
JOURNAL_CHECK_TRANS_MAGIC(trans);
if (!(trans->flags & J_TRANS_COMMITTED))
return 0;
journal = trans->journal;
if (journal->j_tail == 0) {
/* Update the tail */
journal->j_tail_sequence = trans->tid;
journal->j_tail = trans->start;
journal->j_superblock->s_start = ext2fs_cpu_to_be32(trans->start);
}
/* Update the head */
journal->j_head = trans->end + 1;
journal->j_transaction_sequence = trans->tid + 1;
trans->magic = 0;
/* Mark ourselves as needing recovery */
if (!(EXT2_HAS_INCOMPAT_FEATURE(trans->fs->super,
EXT3_FEATURE_INCOMPAT_RECOVER))) {
trans->fs->super->s_feature_incompat |=
EXT3_FEATURE_INCOMPAT_RECOVER;
ext2fs_mark_super_dirty(trans->fs);
}
return 0;
}
static int jbd2_block_tag_csum_verify(journal_t *j, journal_block_tag_t *tag,
void *buf, __u32 sequence)
{
__u32 provided, calculated;
if (!JFS_HAS_INCOMPAT_FEATURE(j, JFS_FEATURE_INCOMPAT_CSUM_V2))
return 1;
sequence = ext2fs_cpu_to_be32(sequence);
calculated = ext2fs_crc32c_le(~0, j->j_superblock->s_uuid,
sizeof(j->j_superblock->s_uuid));
calculated = ext2fs_crc32c_le(calculated, (__u8 *)&sequence,
sizeof(sequence));
calculated = ext2fs_crc32c_le(calculated, buf, j->j_blocksize) & 0xffff;
provided = ext2fs_be16_to_cpu(tag->t_checksum);
return provided == ext2fs_cpu_to_be32(calculated);
}
void jbd2_block_tag_csum_set(journal_t *j, journal_block_tag_t *tag,
struct buffer_head *bh, __u32 sequence)
{
journal_block_tag3_t *tag3 = (journal_block_tag3_t *)tag;
__u32 csum32;
__be32 seq;
if (!journal_has_csum_v2or3(j))
return;
seq = ext2fs_cpu_to_be32(sequence);
csum32 = jbd2_chksum(j, j->j_csum_seed, (__u8 *)&seq, sizeof(seq));
csum32 = jbd2_chksum(j, csum32, bh->b_data, bh->b_size);
if (jfs_has_feature_csum3(j))
tag3->t_checksum = ext2fs_cpu_to_be32(csum32);
else
tag->t_checksum = ext2fs_cpu_to_be16(csum32);
}
static int jbd2_block_tag_csum_verify(journal_t *j, journal_block_tag_t *tag,
void *buf, __u32 sequence)
{
journal_block_tag3_t *tag3 = (journal_block_tag3_t *)tag;
__u32 csum32;
__u32 seq;
if (!journal_has_csum_v2or3(j))
return 1;
seq = ext2fs_cpu_to_be32(sequence);
csum32 = jbd2_chksum(j, j->j_csum_seed, (__u8 *)&seq, sizeof(seq));
csum32 = jbd2_chksum(j, csum32, buf, j->j_blocksize);
if (jfs_has_feature_csum3(j))
return tag3->t_checksum == ext2fs_cpu_to_be32(csum32);
return tag->t_checksum == ext2fs_cpu_to_be16(csum32);
}
static void update_journal_csum(journal_t *journal, int ver)
{
journal_superblock_t *jsb;
if (journal->j_format_version < 2)
return;
if (journal->j_tail != 0 ||
EXT2_HAS_INCOMPAT_FEATURE(journal->j_fs_dev->k_fs->super,
EXT3_FEATURE_INCOMPAT_RECOVER)) {
printf("Journal needs recovery, will not add csums.\n");
return;
}
/* metadata_csum implies journal csum v3 */
jsb = journal->j_superblock;
if (EXT2_HAS_RO_COMPAT_FEATURE(journal->j_fs_dev->k_fs->super,
EXT4_FEATURE_RO_COMPAT_METADATA_CSUM)) {
printf("Setting csum v%d\n", ver);
switch (ver) {
case 2:
journal->j_superblock->s_feature_incompat &=
ext2fs_cpu_to_be32(~JFS_FEATURE_INCOMPAT_CSUM_V3);
journal->j_superblock->s_feature_incompat |=
ext2fs_cpu_to_be32(JFS_FEATURE_INCOMPAT_CSUM_V2);
journal->j_superblock->s_feature_compat &=
ext2fs_cpu_to_be32(~JFS_FEATURE_COMPAT_CHECKSUM);
break;
case 3:
journal->j_superblock->s_feature_incompat &=
ext2fs_cpu_to_be32(~JFS_FEATURE_INCOMPAT_CSUM_V2);
journal->j_superblock->s_feature_incompat |=
ext2fs_cpu_to_be32(JFS_FEATURE_INCOMPAT_CSUM_V3);
journal->j_superblock->s_feature_compat &=
ext2fs_cpu_to_be32(~JFS_FEATURE_COMPAT_CHECKSUM);
break;
default:
printf("Unknown checksum v%d\n", ver);
break;
}
journal->j_superblock->s_checksum_type = JBD2_CRC32C_CHKSUM;
journal->j_csum_seed = jbd2_chksum(journal, ~0, jsb->s_uuid,
sizeof(jsb->s_uuid));
} else {
journal->j_superblock->s_feature_compat |=
ext2fs_cpu_to_be32(JFS_FEATURE_COMPAT_CHECKSUM);
journal->j_superblock->s_feature_incompat &=
ext2fs_cpu_to_be32(~(JFS_FEATURE_INCOMPAT_CSUM_V2 |
JFS_FEATURE_INCOMPAT_CSUM_V3));
}
}
static errcode_t e2fsck_journal_sb_csum_set(journal_t *j,
journal_superblock_t *jsb)
{
__u32 crc;
if (!JFS_HAS_INCOMPAT_FEATURE(j, JFS_FEATURE_INCOMPAT_CSUM_V2))
return 0;
crc = e2fsck_journal_sb_csum(jsb);
jsb->s_checksum = ext2fs_cpu_to_be32(crc);
return 0;
}
static errcode_t e2fsck_journal_sb_csum_set(journal_t *j,
journal_superblock_t *jsb)
{
__u32 crc;
if (!journal_has_csum_v2or3(j))
return 0;
crc = e2fsck_journal_sb_csum(jsb);
jsb->s_checksum = ext2fs_cpu_to_be32(crc);
return 0;
}
void jbd2_descr_block_csum_set(journal_t *j, struct buffer_head *bh)
{
struct journal_block_tail *tail;
__u32 csum;
if (!journal_has_csum_v2or3(j))
return;
tail = (struct journal_block_tail *)(bh->b_data + j->j_blocksize -
sizeof(struct journal_block_tail));
tail->t_checksum = 0;
csum = jbd2_chksum(j, j->j_csum_seed, bh->b_data, j->j_blocksize);
tail->t_checksum = ext2fs_cpu_to_be32(csum);
}
void jbd2_commit_block_csum_set(journal_t *j, struct buffer_head *bh)
{
struct commit_header *h;
__u32 csum;
if (!journal_has_csum_v2or3(j))
return;
h = (struct commit_header *)(bh->b_data);
h->h_chksum_type = 0;
h->h_chksum_size = 0;
h->h_chksum[0] = 0;
csum = jbd2_chksum(j, j->j_csum_seed, bh->b_data, j->j_blocksize);
h->h_chksum[0] = ext2fs_cpu_to_be32(csum);
}
static int jbd2_commit_block_csum_verify(journal_t *j, void *buf)
{
struct commit_header *h;
__u32 provided;
__u32 calculated;
if (!journal_has_csum_v2or3(j))
return 1;
h = buf;
provided = h->h_chksum[0];
h->h_chksum[0] = 0;
calculated = jbd2_chksum(j, j->j_csum_seed, buf, j->j_blocksize);
h->h_chksum[0] = provided;
return provided == ext2fs_cpu_to_be32(calculated);
}
static int jbd2_revoke_block_csum_verify(journal_t *j,
void *buf)
{
struct journal_revoke_tail *tail;
__u32 provided;
__u32 calculated;
if (!journal_has_csum_v2or3(j))
return 1;
tail = (struct journal_revoke_tail *)(buf + j->j_blocksize -
sizeof(struct journal_revoke_tail));
provided = tail->r_checksum;
tail->r_checksum = 0;
calculated = jbd2_chksum(j, j->j_csum_seed, buf, j->j_blocksize);
tail->r_checksum = provided;
return provided == ext2fs_cpu_to_be32(calculated);
}
static errcode_t journal_write(journal_t *journal,
int flags, blk64_t *block_list,
size_t block_len, blk64_t *revoke_list,
size_t revoke_len, FILE *fp)
{
blk64_t blocks;
journal_transaction_t trans;
errcode_t err;
if (revoke_len > 0) {
journal->j_superblock->s_feature_incompat |=
ext2fs_cpu_to_be32(JFS_FEATURE_INCOMPAT_REVOKE);
mark_buffer_dirty(journal->j_sb_buffer);
}
blocks = journal_guess_blocks(journal, block_len, revoke_len);
err = journal_open_trans(journal, &trans, blocks);
if (err)
goto error;
err = journal_add_blocks_to_trans(&trans, block_list, block_len, fp);
if (err)
goto error;
err = journal_add_revoke_to_trans(&trans, revoke_list, revoke_len);
if (err)
goto error;
if (!(flags & JOURNAL_WRITE_NO_COMMIT)) {
err = journal_commit_trans(&trans);
if (err)
goto error;
}
err = journal_close_trans(&trans);
if (err)
goto error;
error:
return err;
}
static void update_64bit_flag(journal_t *journal)
{
if (journal->j_format_version < 2)
return;
if (!EXT2_HAS_INCOMPAT_FEATURE(journal->j_fs_dev->k_fs->super,
EXT4_FEATURE_INCOMPAT_64BIT))
return;
if (JFS_HAS_INCOMPAT_FEATURE(journal, JFS_FEATURE_INCOMPAT_64BIT) &&
EXT2_HAS_INCOMPAT_FEATURE(journal->j_fs_dev->k_fs->super,
EXT4_FEATURE_INCOMPAT_64BIT))
return;
if (journal->j_tail != 0 ||
EXT2_HAS_INCOMPAT_FEATURE(journal->j_fs_dev->k_fs->super,
EXT3_FEATURE_INCOMPAT_RECOVER)) {
printf("Journal needs recovery, will not set 64bit.\n");
return;
}
journal->j_superblock->s_feature_incompat |=
ext2fs_cpu_to_be32(JFS_FEATURE_INCOMPAT_64BIT);
}
static void print_inline_journal_information(ext2_filsys fs)
{
journal_superblock_t *jsb;
struct ext2_inode inode;
ext2_file_t journal_file;
errcode_t retval;
ino_t ino = fs->super->s_journal_inum;
char buf[1024];
__u32 *mask_ptr, mask, m;
int i, j, size, printed = 0;
if (fs->flags & EXT2_FLAG_IMAGE_FILE)
return;
retval = ext2fs_read_inode(fs, ino, &inode);
if (retval) {
com_err(program_name, retval, "%s",
_("while reading journal inode"));
exit(1);
}
retval = ext2fs_file_open2(fs, ino, &inode, 0, &journal_file);
if (retval) {
com_err(program_name, retval, "%s",
_("while opening journal inode"));
exit(1);
}
retval = ext2fs_file_read(journal_file, buf, sizeof(buf), 0);
if (retval) {
com_err(program_name, retval, "%s",
_("while reading journal super block"));
exit(1);
}
ext2fs_file_close(journal_file);
jsb = (journal_superblock_t *) buf;
if (be32_to_cpu(jsb->s_header.h_magic) != JFS_MAGIC_NUMBER) {
fprintf(stderr, "%s",
_("Journal superblock magic number invalid!\n"));
exit(1);
}
printf("%s", _("Journal features: "));
for (i=0, mask_ptr=&jsb->s_feature_compat; i <3; i++,mask_ptr++) {
mask = be32_to_cpu(*mask_ptr);
for (j=0,m=1; j < 32; j++, m<<=1) {
if (mask & m) {
printf(" %s", e2p_jrnl_feature2string(i, m));
printed++;
}
}
}
if (printed == 0)
printf(" (none)");
printf("\n");
fputs(_("Journal size: "), stdout);
if ((fs->super->s_feature_ro_compat &
EXT4_FEATURE_RO_COMPAT_HUGE_FILE) &&
(inode.i_flags & EXT4_HUGE_FILE_FL))
size = inode.i_blocks / (fs->blocksize / 1024);
else
size = inode.i_blocks >> 1;
if (size < 8192)
printf("%uk\n", size);
else
printf("%uM\n", size >> 10);
printf(_("Journal length: %u\n"
"Journal sequence: 0x%08x\n"
"Journal start: %u\n"),
(unsigned int)ntohl(jsb->s_maxlen),
(unsigned int)ntohl(jsb->s_sequence),
(unsigned int)ntohl(jsb->s_start));
if (jsb->s_feature_compat &
ext2fs_cpu_to_be32(JFS_FEATURE_COMPAT_CHECKSUM))
printf(_("Journal checksum type: crc32\n"));
if (jsb->s_feature_incompat &
ext2fs_cpu_to_be32(JFS_FEATURE_INCOMPAT_CSUM_V2))
printf(_("Journal checksum type: %s\n"
"Journal checksum: 0x%08x\n"),
journal_checksum_type_str(jsb->s_checksum_type),
ext2fs_be32_to_cpu(jsb->s_checksum));
if (jsb->s_errno != 0)
printf(_("Journal errno: %d\n"),
(int) ntohl(jsb->s_errno));
}
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;
int tag_bytes = journal_tag_bytes(journal);
__u32 crc32_sum = ~0; /* Transactional Checksums */
int descr_csum_size = 0;
int block_error = 0;
/*
* 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 = ext2fs_be32_to_cpu(sb->s_sequence);
next_log_block = ext2fs_be32_to_cpu(sb->s_start);
first_commit_ID = next_commit_ID;
if (pass == PASS_SCAN)
info->start_transaction = first_commit_ID;
jbd_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;
cond_resched();
/* 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;
jbd_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. */
jbd_debug(3, "JBD2: 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 != ext2fs_cpu_to_be32(JFS_MAGIC_NUMBER)) {
brelse(bh);
break;
}
blocktype = ext2fs_be32_to_cpu(tmp->h_blocktype);
sequence = ext2fs_be32_to_cpu(tmp->h_sequence);
jbd_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:
/* Verify checksum first */
if (journal_has_csum_v2or3(journal))
descr_csum_size =
sizeof(struct journal_block_tail);
if (descr_csum_size > 0 &&
!jbd2_descr_block_csum_verify(journal,
bh->b_data)) {
err = -EFSBADCRC;
brelse(bh);
goto failed;
}
/* If it is a valid descriptor block, replay it
* in pass REPLAY; if journal_checksums enabled, then
* calculate checksums in PASS_SCAN, otherwise,
* just skip over the blocks it describes. */
if (pass != PASS_REPLAY) {
if (pass == PASS_SCAN &&
jfs_has_feature_checksum(journal) &&
!info->end_transaction) {
if (calc_chksums(journal, bh,
&next_log_block,
&crc32_sum)) {
put_bh(bh);
break;
}
put_bh(bh);
continue;
}
next_log_block += count_tags(journal, bh);
wrap(journal, next_log_block);
put_bh(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 + tag_bytes)
<= journal->j_blocksize - descr_csum_size) {
unsigned long io_block;
tag = (journal_block_tag_t *) tagp;
flags = ext2fs_be16_to_cpu(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
"JBD2: IO error %d recovering "
"block %ld in log\n",
err, io_block);
} else {
unsigned long long blocknr;
J_ASSERT(obh != NULL);
blocknr = read_tag_block(journal,
tag);
/* 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;
}
/* Look for block corruption */
if (!jbd2_block_tag_csum_verify(
journal, tag, obh->b_data,
ext2fs_be32_to_cpu(tmp->h_sequence))) {
brelse(obh);
success = -EFSBADCRC;
printk(KERN_ERR "JBD2: Invalid "
"checksum recovering "
"block %llu in log\n",
blocknr);
block_error = 1;
goto skip_write;
}
/* Find a buffer for the new
* data being restored */
nbh = __getblk(journal->j_fs_dev,
blocknr,
journal->j_blocksize);
if (nbh == NULL) {
printk(KERN_ERR
"JBD2: Out of memory "
"during recovery.\n");
err = -ENOMEM;
brelse(bh);
brelse(obh);
goto failed;
}
lock_buffer(nbh);
memcpy(nbh->b_data, obh->b_data,
journal->j_blocksize);
if (flags & JFS_FLAG_ESCAPE) {
*((__u32 *)nbh->b_data) =
ext2fs_cpu_to_be32(JFS_MAGIC_NUMBER);
}
BUFFER_TRACE(nbh, "marking dirty");
set_buffer_uptodate(nbh);
mark_buffer_dirty(nbh);
BUFFER_TRACE(nbh, "marking uptodate");
++info->nr_replays;
/* ll_rw_block(WRITE, 1, &nbh); */
unlock_buffer(nbh);
brelse(obh);
brelse(nbh);
}
skip_write:
tagp += tag_bytes;
if (!(flags & JFS_FLAG_SAME_UUID))
tagp += 16;
if (flags & JFS_FLAG_LAST_TAG)
break;
}
brelse(bh);
continue;
case JFS_COMMIT_BLOCK:
/* How to differentiate between interrupted commit
* and journal corruption ?
*
* {nth transaction}
* Checksum Verification Failed
* |
* ____________________
* | |
* async_commit sync_commit
* | |
* | GO TO NEXT "Journal Corruption"
* | TRANSACTION
* |
* {(n+1)th transanction}
* |
* _______|______________
* | |
* Commit block found Commit block not found
* | |
* "Journal Corruption" |
* _____________|_________
* | |
* nth trans corrupt OR nth trans
* and (n+1)th interrupted interrupted
* before commit block
* could reach the disk.
* (Cannot find the difference in above
* mentioned conditions. Hence assume
* "Interrupted Commit".)
*/
/* Found an expected commit block: if checksums
* are present verify them in PASS_SCAN; else not
* much to do other than move on to the next sequence
* number. */
if (pass == PASS_SCAN &&
jfs_has_feature_checksum(journal)) {
int chksum_err, chksum_seen;
struct commit_header *cbh =
(struct commit_header *)bh->b_data;
unsigned found_chksum =
ext2fs_be32_to_cpu(cbh->h_chksum[0]);
chksum_err = chksum_seen = 0;
if (info->end_transaction) {
journal->j_failed_commit =
info->end_transaction;
brelse(bh);
break;
}
if (crc32_sum == found_chksum &&
cbh->h_chksum_type == JFS_CRC32_CHKSUM &&
cbh->h_chksum_size ==
JFS_CRC32_CHKSUM_SIZE)
chksum_seen = 1;
else if (!(cbh->h_chksum_type == 0 &&
cbh->h_chksum_size == 0 &&
found_chksum == 0 &&
!chksum_seen))
/*
* If fs is mounted using an old kernel and then
* kernel with journal_chksum is used then we
* get a situation where the journal flag has
* checksum flag set but checksums are not
* present i.e chksum = 0, in the individual
* commit blocks.
* Hence to avoid checksum failures, in this
* situation, this extra check is added.
*/
chksum_err = 1;
if (chksum_err) {
info->end_transaction = next_commit_ID;
if (!jfs_has_feature_async_commit(journal)){
journal->j_failed_commit =
next_commit_ID;
brelse(bh);
break;
}
}
crc32_sum = ~0;
}
if (pass == PASS_SCAN &&
!jbd2_commit_block_csum_verify(journal,
bh->b_data)) {
info->end_transaction = next_commit_ID;
if (!jfs_has_feature_async_commit(journal)) {
journal->j_failed_commit =
next_commit_ID;
brelse(bh);
break;
}
}
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:
jbd_debug(3, "Unrecognised magic %d, end of scan.\n",
blocktype);
brelse(bh);
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) {
if (!info->end_transaction)
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 "JBD2: recovery pass %d ended at "
"transaction %u, expected %u\n",
pass, next_commit_ID, info->end_transaction);
if (!success)
success = -EIO;
}
}
if (block_error && success == 0)
success = -EIO;
return success;
failed:
return err;
}
static errcode_t journal_add_blocks_to_trans(journal_transaction_t *trans,
blk64_t *block_list, size_t block_len,
FILE *fp)
{
blk64_t curr_blk, jdb_blk;
size_t i, j;
int csum_size = 0;
journal_header_t *jdb;
journal_block_tag_t *jdbt;
int tag_bytes;
void *buf = NULL, *jdb_buf = NULL;
struct buffer_head *bh = NULL, *data_bh;
errcode_t err;
JOURNAL_CHECK_TRANS_MAGIC(trans);
if ((trans->flags & J_TRANS_COMMITTED) ||
!(trans->flags & J_TRANS_OPEN))
return EXT2_ET_INVALID_ARGUMENT;
if (block_len == 0)
return 0;
/* Do we need to leave space at the end for a checksum? */
if (journal_has_csum_v2or3(trans->journal))
csum_size = sizeof(struct journal_block_tail);
curr_blk = jdb_blk = trans->block;
data_bh = getblk(trans->journal->j_dev, curr_blk,
trans->journal->j_blocksize);
if (data_bh == NULL)
return ENOMEM;
buf = data_bh->b_data;
/* write the descriptor block header */
bh = getblk(trans->journal->j_dev, curr_blk,
trans->journal->j_blocksize);
if (bh == NULL) {
err = ENOMEM;
goto error;
}
jdb = jdb_buf = bh->b_data;
jdb->h_magic = ext2fs_cpu_to_be32(JFS_MAGIC_NUMBER);
jdb->h_blocktype = ext2fs_cpu_to_be32(JFS_DESCRIPTOR_BLOCK);
jdb->h_sequence = ext2fs_cpu_to_be32(trans->tid);
jdbt = (journal_block_tag_t *)(jdb + 1);
curr_blk++;
for (i = 0; i < block_len; i++) {
j = fread(data_bh->b_data, trans->journal->j_blocksize, 1, fp);
if (j != 1) {
err = errno;
goto error;
}
tag_bytes = journal_tag_bytes(trans->journal);
/* No space left in descriptor block, write it out */
if ((char *)jdbt + tag_bytes >
(char *)jdb_buf + trans->journal->j_blocksize - csum_size) {
jbd2_descr_block_csum_set(trans->journal, bh);
err = journal_bmap(trans->journal, jdb_blk,
&bh->b_blocknr);
if (err)
goto error;
dbg_printf("Writing descriptor block at %llu:%llu\n",
jdb_blk, bh->b_blocknr);
mark_buffer_dirty(bh);
ll_rw_block(WRITE, 1, &bh);
err = bh->b_err;
if (err)
goto error;
jdbt = (journal_block_tag_t *)(jdb + 1);
jdb_blk = curr_blk;
curr_blk++;
}
if (block_list[i] >=
ext2fs_blocks_count(trans->journal->j_fs_dev->k_fs->super)) {
err = EXT2_ET_BAD_BLOCK_NUM;
goto error;
}
/* Fill out the block tag */
jdbt->t_blocknr = ext2fs_cpu_to_be32(block_list[i] & 0xFFFFFFFF);
jdbt->t_flags = 0;
if (jdbt != (journal_block_tag_t *)(jdb + 1))
jdbt->t_flags |= ext2fs_cpu_to_be16(JFS_FLAG_SAME_UUID);
else {
memcpy(jdbt + tag_bytes,
trans->journal->j_superblock->s_uuid,
sizeof(trans->journal->j_superblock->s_uuid));
tag_bytes += 16;
}
if (i == block_len - 1)
jdbt->t_flags |= ext2fs_cpu_to_be16(JFS_FLAG_LAST_TAG);
if (*((__u32 *)buf) == ext2fs_cpu_to_be32(JFS_MAGIC_NUMBER)) {
*((__u32 *)buf) = 0;
jdbt->t_flags |= ext2fs_cpu_to_be16(JFS_FLAG_ESCAPE);
}
if (JFS_HAS_INCOMPAT_FEATURE(trans->journal,
JFS_FEATURE_INCOMPAT_64BIT))
jdbt->t_blocknr_high = ext2fs_cpu_to_be32(block_list[i] >> 32);
jbd2_block_tag_csum_set(trans->journal, jdbt, data_bh,
trans->tid);
/* Write the data block */
err = journal_bmap(trans->journal, curr_blk,
&data_bh->b_blocknr);
if (err)
goto error;
dbg_printf("Writing data block %llu at %llu:%llu tag %d\n",
block_list[i], curr_blk, data_bh->b_blocknr,
tag_bytes);
mark_buffer_dirty(data_bh);
ll_rw_block(WRITE, 1, &data_bh);
err = data_bh->b_err;
if (err)
goto error;
curr_blk++;
jdbt = (journal_block_tag_t *)(((char *)jdbt) + tag_bytes);
}
/* Write out the last descriptor block */
if (jdbt != (journal_block_tag_t *)(jdb + 1)) {
jbd2_descr_block_csum_set(trans->journal, bh);
err = journal_bmap(trans->journal, jdb_blk, &bh->b_blocknr);
if (err)
goto error;
dbg_printf("Writing descriptor block at %llu:%llu\n",
jdb_blk, bh->b_blocknr);
mark_buffer_dirty(bh);
ll_rw_block(WRITE, 1, &bh);
err = bh->b_err;
if (err)
goto error;
}
error:
trans->block = curr_blk;
if (bh)
brelse(bh);
brelse(data_bh);
return err;
}
errcode_t journal_find_head(journal_t *journal)
{
unsigned int next_commit_ID;
blk64_t next_log_block, head_block;
int err;
journal_superblock_t *sb;
journal_header_t *tmp;
struct buffer_head *bh;
unsigned int sequence;
int blocktype;
/*
* 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 = ext2fs_be32_to_cpu(sb->s_sequence);
next_log_block = ext2fs_be32_to_cpu(sb->s_start);
head_block = next_log_block;
if (next_log_block == 0)
return 0;
bh = getblk(journal->j_dev, 0, journal->j_blocksize);
if (bh == NULL)
return ENOMEM;
/*
* 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) {
dbg_printf("Scanning for sequence ID %u at %lu/%lu\n",
next_commit_ID, (unsigned long)next_log_block,
journal->j_last);
/* Skip over each chunk of the transaction looking
* either the next descriptor block or the final commit
* record. */
err = journal_bmap(journal, next_log_block, &bh->b_blocknr);
if (err)
goto err;
mark_buffer_uptodate(bh, 0);
ll_rw_block(READ, 1, &bh);
err = bh->b_err;
if (err)
goto err;
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 != ext2fs_cpu_to_be32(JFS_MAGIC_NUMBER)) {
dbg_printf("JBD2: wrong magic 0x%x\n", tmp->h_magic);
goto err;
}
blocktype = ext2fs_be32_to_cpu(tmp->h_blocktype);
sequence = ext2fs_be32_to_cpu(tmp->h_sequence);
dbg_printf("Found magic %d, sequence %d\n",
blocktype, sequence);
if (sequence != next_commit_ID) {
dbg_printf("JBD2: Wrong sequence %d (wanted %d)\n",
sequence, next_commit_ID);
goto err;
}
/* 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:
next_log_block += count_tags(journal, bh->b_data);
wrap(journal, next_log_block);
continue;
case JFS_COMMIT_BLOCK:
head_block = next_log_block;
next_commit_ID++;
continue;
case JFS_REVOKE_BLOCK:
continue;
default:
dbg_printf("Unrecognised magic %d, end of scan.\n",
blocktype);
err = -EINVAL;
goto err;
}
}
err:
if (err == 0) {
dbg_printf("head seq=%d blk=%llu\n", next_commit_ID,
head_block);
journal->j_transaction_sequence = next_commit_ID;
journal->j_head = head_block;
}
brelse(bh);
return err;
}
static errcode_t journal_add_revoke_to_trans(journal_transaction_t *trans,
blk64_t *revoke_list,
size_t revoke_len)
{
journal_revoke_header_t *jrb;
void *buf;
size_t i, offset;
blk64_t curr_blk;
int sz, csum_size = 0;
struct buffer_head *bh;
errcode_t err;
JOURNAL_CHECK_TRANS_MAGIC(trans);
if ((trans->flags & J_TRANS_COMMITTED) ||
!(trans->flags & J_TRANS_OPEN))
return EXT2_ET_INVALID_ARGUMENT;
if (revoke_len == 0)
return 0;
/* Do we need to leave space at the end for a checksum? */
if (journal_has_csum_v2or3(trans->journal))
csum_size = sizeof(struct journal_revoke_tail);
curr_blk = trans->block;
bh = getblk(trans->journal->j_dev, curr_blk,
trans->journal->j_blocksize);
if (bh == NULL)
return ENOMEM;
jrb = buf = bh->b_data;
jrb->r_header.h_magic = ext2fs_cpu_to_be32(JFS_MAGIC_NUMBER);
jrb->r_header.h_blocktype = ext2fs_cpu_to_be32(JFS_REVOKE_BLOCK);
jrb->r_header.h_sequence = ext2fs_cpu_to_be32(trans->tid);
offset = sizeof(*jrb);
if (JFS_HAS_INCOMPAT_FEATURE(trans->journal,
JFS_FEATURE_INCOMPAT_64BIT))
sz = 8;
else
sz = 4;
for (i = 0; i < revoke_len; i++) {
/* Block full, write to journal */
if (offset + sz > trans->journal->j_blocksize - csum_size) {
jrb->r_count = ext2fs_cpu_to_be32(offset);
jbd2_revoke_csum_set(trans->journal, bh);
err = journal_bmap(trans->journal, curr_blk,
&bh->b_blocknr);
if (err)
goto error;
dbg_printf("Writing revoke block at %llu:%llu\n",
curr_blk, bh->b_blocknr);
mark_buffer_dirty(bh);
ll_rw_block(WRITE, 1, &bh);
err = bh->b_err;
if (err)
goto error;
offset = sizeof(*jrb);
curr_blk++;
}
if (revoke_list[i] >=
ext2fs_blocks_count(trans->journal->j_fs_dev->k_fs->super)) {
err = EXT2_ET_BAD_BLOCK_NUM;
goto error;
}
if (JFS_HAS_INCOMPAT_FEATURE(trans->journal,
JFS_FEATURE_INCOMPAT_64BIT))
*((__u64 *)(&((char *)buf)[offset])) =
ext2fs_cpu_to_be64(revoke_list[i]);
else
*((__u32 *)(&((char *)buf)[offset])) =
ext2fs_cpu_to_be32(revoke_list[i]);
offset += sz;
}
if (offset > 0) {
jrb->r_count = ext2fs_cpu_to_be32(offset);
jbd2_revoke_csum_set(trans->journal, bh);
err = journal_bmap(trans->journal, curr_blk, &bh->b_blocknr);
if (err)
goto error;
dbg_printf("Writing revoke block at %llu:%llu\n",
curr_blk, bh->b_blocknr);
mark_buffer_dirty(bh);
ll_rw_block(WRITE, 1, &bh);
err = bh->b_err;
if (err)
goto error;
curr_blk++;
}
error:
trans->block = curr_blk;
brelse(bh);
return err;
}
static errcode_t journal_commit_trans(journal_transaction_t *trans)
{
struct buffer_head *bh, *cbh = NULL;
struct commit_header *commit;
#ifdef HAVE_SYS_TIME_H
struct timeval tv;
#endif
errcode_t err;
JOURNAL_CHECK_TRANS_MAGIC(trans);
if ((trans->flags & J_TRANS_COMMITTED) ||
!(trans->flags & J_TRANS_OPEN))
return EXT2_ET_INVALID_ARGUMENT;
bh = getblk(trans->journal->j_dev, 0, trans->journal->j_blocksize);
if (bh == NULL)
return ENOMEM;
/* write the descriptor block header */
commit = (struct commit_header *)bh->b_data;
commit->h_magic = ext2fs_cpu_to_be32(JFS_MAGIC_NUMBER);
commit->h_blocktype = ext2fs_cpu_to_be32(JFS_COMMIT_BLOCK);
commit->h_sequence = ext2fs_cpu_to_be32(trans->tid);
if (JFS_HAS_COMPAT_FEATURE(trans->journal,
JFS_FEATURE_COMPAT_CHECKSUM)) {
__u32 csum_v1 = ~0;
blk64_t cblk;
cbh = getblk(trans->journal->j_dev, 0,
trans->journal->j_blocksize);
if (cbh == NULL) {
err = ENOMEM;
goto error;
}
for (cblk = trans->start; cblk < trans->block; cblk++) {
err = journal_bmap(trans->journal, cblk,
&cbh->b_blocknr);
if (err)
goto error;
mark_buffer_uptodate(cbh, 0);
ll_rw_block(READ, 1, &cbh);
err = cbh->b_err;
if (err)
goto error;
csum_v1 = ext2fs_crc32_be(csum_v1,
(unsigned char const *)cbh->b_data,
cbh->b_size);
}
commit->h_chksum_type = JFS_CRC32_CHKSUM;
commit->h_chksum_size = JFS_CRC32_CHKSUM_SIZE;
commit->h_chksum[0] = ext2fs_cpu_to_be32(csum_v1);
} else {
commit->h_chksum_type = 0;
commit->h_chksum_size = 0;
commit->h_chksum[0] = 0;
}
#ifdef HAVE_SYS_TIME_H
gettimeofday(&tv, NULL);
commit->h_commit_sec = ext2fs_cpu_to_be32(tv.tv_sec);
commit->h_commit_nsec = ext2fs_cpu_to_be32(tv.tv_usec * 1000);
#else
commit->h_commit_sec = 0;
commit->h_commit_nsec = 0;
#endif
/* Write block */
jbd2_commit_block_csum_set(trans->journal, bh);
err = journal_bmap(trans->journal, trans->block, &bh->b_blocknr);
if (err)
goto error;
dbg_printf("Writing commit block at %llu:%llu\n", trans->block,
bh->b_blocknr);
mark_buffer_dirty(bh);
ll_rw_block(WRITE, 1, &bh);
err = bh->b_err;
if (err)
goto error;
trans->flags |= J_TRANS_COMMITTED;
trans->flags &= ~J_TRANS_OPEN;
trans->block++;
trans->fs->super->s_feature_incompat |= EXT3_FEATURE_INCOMPAT_RECOVER;
ext2fs_mark_super_dirty(trans->fs);
error:
if (cbh)
brelse(cbh);
brelse(bh);
return err;
}
