On 2012-05-22, at 9:24 AM, Eric Sandeen wrote:
> On 5/21/12 8:42 PM, Theodore Ts'o wrote:
>> I just found what a somewhat disturbing bug which is still present in
>> 1.42.3 --- namely, that the journal replay code in e2fsck was not
>> properly handling 64-bit file systems correctly, by dropping the high
>> bits of the block number from the jbd2 descriptor blocks.
Hmm, we did have a bug report that would be explained by this, but it wasn't easily reproduced (dependent on metadata blocks beyond 16TB in
hindsight) and only on a test filesystem, so I hadn't had a chance to
dig into it yet.
We did do a 128TB full-filesystem data write + verification + e2fsck,
but I guess that didn't validate journal recovery. Most filesystems
only have a fraction of data beyond 16TB, and the allocators prefer
to locate inodes/blocks at the beginning of the filesystem, so it
hasn't shown up in real usage very much.
Good catch.
>> I have not had a chance to fully test to make sure we don't have other
>> bugs hiding in the 64-bit code, but it's clear no one else has had a
>> lot of time to test it either -- at least not to the extent of doing
>> power fail testing of > 16TB file systems! :-(
>
> Whoops. :(
>
> How do you decide between using "unsigned long long" and "blk64_t" below?
>
> If they'd all been blk_t's they'd have been easier to spot, if anyone went looking, I suppose.
Makes sense. Better to be consistent with these things, so that the
compiler and users (grep, etc) can find type mismatches and such.
>> - Ted
>>
>> From 3b693d0b03569795d04920a04a0a21e5f64ffedc Mon Sep 17 00:00:00 2001
>> From: Theodore Ts'o <tytso@xxxxxxx>
>> Date: Mon, 21 May 2012 21:30:45 -0400
>> Subject: [PATCH] e2fsck: fix 64-bit journal support
>>
>> 64-bit journal support was broken; we weren't using the high bits from
>> the journal descriptor blocks! We were also using "unsigned long" for
>> the journal block numbers, which would be a problem on 32-bit systems.
>>
>> Signed-off-by: "Theodore Ts'o" <tytso@xxxxxxx>
>> ---
>> e2fsck/jfs_user.h | 4 ++--
>> e2fsck/journal.c | 33 +++++++++++++++++----------------
>> e2fsck/recovery.c | 25 ++++++++++++-------------
>> 3 files changed, 31 insertions(+), 31 deletions(-)
>>
>> diff --git a/e2fsck/jfs_user.h b/e2fsck/jfs_user.h
>> index 9e33306..92f8ae2 100644
>> --- a/e2fsck/jfs_user.h
>> +++ b/e2fsck/jfs_user.h
>> @@ -18,7 +18,7 @@ struct buffer_head {
>> e2fsck_t b_ctx;
>> io_channel b_io;
>> int b_size;
>> - blk_t b_blocknr;
>> + unsigned long long b_blocknr;
>> int b_dirty;
>> int b_uptodate;
>> int b_err;
>> @@ -121,7 +121,7 @@ _INLINE_ size_t journal_tag_bytes(journal_t *journal)
>> /*
>> * Kernel compatibility functions are defined in journal.c
>> */
>> -int journal_bmap(journal_t *journal, blk64_t block, unsigned long *phys);
>> +int journal_bmap(journal_t *journal, blk64_t block, unsigned long long *phys);
>> struct buffer_head *getblk(kdev_t ctx, blk64_t blocknr, int blocksize);
>> void sync_blockdev(kdev_t kdev);
>> void ll_rw_block(int rw, int dummy, struct buffer_head *bh[]);
>> diff --git a/e2fsck/journal.c b/e2fsck/journal.c
>> index 915b8bb..bada028 100644
>> --- a/e2fsck/journal.c
>> +++ b/e2fsck/journal.c
>> @@ -44,7 +44,7 @@ static int bh_count = 0;
>> * to use the recovery.c file virtually unchanged from the kernel, so we
>> * don't have to do much to keep kernel and user recovery in sync.
>> */
>> -int journal_bmap(journal_t *journal, blk64_t block, unsigned long *phys)
>> +int journal_bmap(journal_t *journal, blk64_t block, unsigned long long *phys)
>> {
>> #ifdef USE_INODE_IO
>> *phys = block;
>> @@ -80,8 +80,8 @@ struct buffer_head *getblk(kdev_t kdev, blk64_t blocknr, int blocksize)
>> if (journal_enable_debug >= 3)
>> bh_count++;
>> #endif
>> - jfs_debug(4, "getblk for block %lu (%d bytes)(total %d)\n",
>> - (unsigned long) blocknr, blocksize, bh_count);
>> + 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)
>> @@ -114,38 +114,39 @@ void ll_rw_block(int rw, int nr, struct buffer_head *bhp[])
>> for (; nr > 0; --nr) {
>> bh = *bhp++;
>> if (rw == READ && !bh->b_uptodate) {
>> - jfs_debug(3, "reading block %lu/%p\n",
>> - (unsigned long) bh->b_blocknr, (void *) bh);
>> + 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 %lu\n",
>> - (unsigned long) bh->b_blocknr);
>> + "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 %lu/%p\n",
>> - (unsigned long) bh->b_blocknr, (void *) bh);
>> + 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 %lu\n",
>> - (unsigned long) bh->b_blocknr);
>> + "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 %lu\n",
>> + jfs_debug(3, "no-op %s for block %llu\n",
>> rw == READ ? "read" : "write",
>> - (unsigned long) bh->b_blocknr);
>> + bh->b_blocknr);
>> }
>> }
>> }
>> @@ -164,8 +165,8 @@ 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);
>> + jfs_debug(3, "freeing block %llu/%p (total %d)\n",
>> + bh->b_blocknr, (void *) bh, --bh_count);
>> ext2fs_free_mem(&bh);
>> }
>>
>> @@ -237,7 +238,7 @@ static errcode_t e2fsck_get_journal(e2fsck_t ctx, journal_t **ret_journal)
>> journal_t *journal = NULL;
>> errcode_t retval = 0;
>> io_manager io_ptr = 0;
>> - unsigned long start = 0;
>> + unsigned long long start = 0;
>> int ext_journal = 0;
>> int tried_backup_jnl = 0;
>>
>> diff --git a/e2fsck/recovery.c b/e2fsck/recovery.c
>> index b669941..e94ef4e 100644
>> --- a/e2fsck/recovery.c
>> +++ b/e2fsck/recovery.c
>> @@ -71,7 +71,7 @@ static int do_readahead(journal_t *journal, unsigned int start)
>> {
>> int err;
>> unsigned int max, nbufs, next;
>> - unsigned long blocknr;
>> + unsigned long long blocknr;
>> struct buffer_head *bh;
>>
>> struct buffer_head * bufs[MAXBUF];
>> @@ -133,7 +133,7 @@ static int jread(struct buffer_head **bhp, journal_t *journal,
>> unsigned int offset)
>> {
>> int err;
>> - unsigned long blocknr;
>> + unsigned long long blocknr;
>> struct buffer_head *bh;
>>
>> *bhp = NULL;
>> @@ -309,7 +309,6 @@ int journal_skip_recovery(journal_t *journal)
>> return err;
>> }
>>
>> -#if 0
>> static inline unsigned long long read_tag_block(int tag_bytes, journal_block_tag_t *tag)
>> {
>> unsigned long long block = be32_to_cpu(tag->t_blocknr);
>> @@ -317,17 +316,16 @@ static inline unsigned long long read_tag_block(int tag_bytes, journal_block_tag
>> block |= (__u64)be32_to_cpu(tag->t_blocknr_high) << 32;
>> return block;
>> }
>> -#endif
>>
>> /*
>> * calc_chksums calculates the checksums for the blocks described in the
>> * descriptor block.
>> */
>> static int calc_chksums(journal_t *journal, struct buffer_head *bh,
>> - unsigned long *next_log_block, __u32 *crc32_sum)
>> + unsigned long long *next_log_block, __u32 *crc32_sum)
>> {
>> int i, num_blks, err;
>> - unsigned long io_block;
>> + unsigned long long io_block;
>> struct buffer_head *obh;
>>
>> num_blks = count_tags(journal, bh);
>> @@ -340,7 +338,7 @@ static int calc_chksums(journal_t *journal, struct buffer_head *bh,
>> err = jread(&obh, journal, io_block);
>> if (err) {
>> printk(KERN_ERR "JBD: IO error %d recovering block "
>> - "%lu in log\n", err, io_block);
>> + "%llu in log\n", err, io_block);
>> return 1;
>> } else {
>> *crc32_sum = crc32_be(*crc32_sum, (void *)obh->b_data,
>> @@ -355,7 +353,7 @@ 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;
>> + unsigned long long next_log_block;
>> int err, success = 0;
>> journal_superblock_t * sb;
>> journal_header_t * tmp;
>> @@ -485,7 +483,7 @@ static int do_one_pass(journal_t *journal,
>> tagp = &bh->b_data[sizeof(journal_header_t)];
>> while ((tagp - bh->b_data + tag_bytes)
>> <= journal->j_blocksize) {
>> - unsigned long io_block;
>> + unsigned long long io_block;
>>
>> tag = (journal_block_tag_t *) tagp;
>> flags = be32_to_cpu(tag->t_flags);
>> @@ -499,13 +497,14 @@ static int do_one_pass(journal_t *journal,
>> success = err;
>> printk (KERN_ERR
>> "JBD: IO error %d recovering "
>> - "block %ld in log\n",
>> + "block %llu in log\n",
>> err, io_block);
>> } else {
>> - unsigned long blocknr;
>> + unsigned long long blocknr;
>>
>> J_ASSERT(obh != NULL);
>> - blocknr = be32_to_cpu(tag->t_blocknr);
>> + blocknr = read_tag_block(tag_bytes,
>> + tag);
>>
>> /* If the block has been
>> * revoked, then we're all done
>> @@ -733,7 +732,7 @@ static int scan_revoke_records(journal_t *journal, struct buffer_head *bh,
>> record_len = 8;
>>
>> while (offset < max) {
>> - unsigned long blocknr;
>> + unsigned long long blocknr;
>> int err;
>>
>> if (record_len == 4)
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
Cheers, Andreas
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
