Whamcloud - gitweb
e4defrag: adjust total count when files change during the run
[tools/e2fsprogs.git] / e2fsck / journal.c
index dd0aa6a..e83f3a9 100644 (file)
@@ -12,7 +12,9 @@
  * any later version.
  */
 
+#include "config.h"
 #ifdef HAVE_SYS_MOUNT_H
+#include <sys/param.h>
 #include <sys/mount.h>
 #define MNT_FL (MS_MGC_VAL | MS_RDONLY)
 #endif
 #include "problem.h"
 #include "uuid/uuid.h"
 
-#ifdef JFS_DEBUG
+#ifdef CONFIG_JBD_DEBUG                /* Enabled by configure --enable-jfs-debug */
 static int bh_count = 0;
-int journal_enable_debug = 0;
 #endif
 
+/*
+ * Define USE_INODE_IO to use the inode_io.c / fileio.c codepaths.
+ * This creates a larger static binary, and a smaller binary using
+ * shared libraries.  It's also probably slightly less CPU-efficient,
+ * which is why it's not on by default.  But, it's a good way of
+ * testing the functions in inode_io.c and fileio.c.
+ */
+#undef USE_INODE_IO
+
+/* Checksumming functions */
+static int e2fsck_journal_verify_csum_type(journal_t *j,
+                                          journal_superblock_t *jsb)
+{
+       if (!journal_has_csum_v2or3(j))
+               return 1;
+
+       return jsb->s_checksum_type == JBD2_CRC32C_CHKSUM;
+}
+
+static __u32 e2fsck_journal_sb_csum(journal_superblock_t *jsb)
+{
+       __u32 crc, old_crc;
+
+       old_crc = jsb->s_checksum;
+       jsb->s_checksum = 0;
+       crc = ext2fs_crc32c_le(~0, (unsigned char *)jsb,
+                              sizeof(journal_superblock_t));
+       jsb->s_checksum = old_crc;
+
+       return crc;
+}
+
+static int e2fsck_journal_sb_csum_verify(journal_t *j,
+                                        journal_superblock_t *jsb)
+{
+       __u32 provided, calculated;
+
+       if (!journal_has_csum_v2or3(j))
+               return 1;
+
+       provided = ext2fs_be32_to_cpu(jsb->s_checksum);
+       calculated = e2fsck_journal_sb_csum(jsb);
+
+       return provided == calculated;
+}
+
+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;
+}
+
 /* Kernel compatibility functions for handling the journal.  These allow us
  * 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 bmap(struct inode *inode, int block)
+int journal_bmap(journal_t *journal, blk64_t block, unsigned long long *phys)
 {
-       int retval;
-       blk_t phys;
-
-       retval = ext2fs_bmap(inode->i_ctx->fs, inode->i_ino, &inode->i_ext2,
-                            NULL, 0, block, &phys);
+#ifdef USE_INODE_IO
+       *phys = block;
+       return 0;
+#else
+       struct inode    *inode = journal->j_inode;
+       errcode_t       retval;
+       blk64_t         pblk;
 
-       if (retval)
-               com_err(inode->i_ctx->device_name, retval,
-                       _("bmap journal inode %ld, block %d\n"),
-                       inode->i_ino, block);
+       if (!inode) {
+               *phys = block;
+               return 0;
+       }
 
-       return phys;
+       retval= ext2fs_bmap2(inode->i_ctx->fs, inode->i_ino,
+                            &inode->i_ext2, NULL, 0, block, 0, &pblk);
+       *phys = pblk;
+       return -1 * ((int) retval);
+#endif
 }
 
-struct buffer_head *getblk(e2fsck_t ctx, blk_t blocknr, int blocksize)
+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(ctx, sizeof(*bh), "block buffer");
+       bh = e2fsck_allocate_memory(kdev->k_ctx, bufsize, "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 = ctx;
+#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;
 }
 
+int sync_blockdev(kdev_t kdev)
+{
+       io_channel      io;
+
+       if (kdev->k_dev == K_DEV_FS)
+               io = kdev->k_ctx->fs->io;
+       else
+               io = kdev->k_ctx->journal_io;
+
+       return io_channel_flush(io) ? -EIO : 0;
+}
+
 void ll_rw_block(int rw, int nr, struct buffer_head *bhp[])
 {
-       int retval;
+       errcode_t retval;
        struct buffer_head *bh;
 
        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);
-                       retval = io_channel_read_blk(bh->b_ctx->fs->io, 
+                       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 %ld\n", 
+                                       "while reading block %llu\n",
                                        bh->b_blocknr);
-                               bh->b_err = retval;
+                               bh->b_err = (int) 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);
-                       retval = io_channel_write_blk(bh->b_ctx->fs->io, 
+                       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 %ld\n", 
+                                       "while writing block %llu\n",
                                        bh->b_blocknr);
-                               bh->b_err = retval;
+                               bh->b_err = (int) retval;
                                continue;
                        }
                        bh->b_dirty = 0;
                        bh->b_uptodate = 1;
-               } else
-                       jfs_debug(3, "no-op %s for block %lu\n",
-                                 rw == READ ? "read" : "write", 
-                                 (unsigned long) bh->b_blocknr);
+               } else {
+                       jfs_debug(3, "no-op %s for block %llu\n",
+                                 rw == READ ? "read" : "write",
+                                 bh->b_blocknr);
+               }
        }
 }
 
-void mark_buffer_dirty(struct buffer_head *bh, int dummy)
+void mark_buffer_dirty(struct buffer_head *bh)
 {
-       bh->b_dirty = dummy | 1; /* use dummy to avoid unused variable */
+       bh->b_dirty = 1;
 }
 
 static void mark_buffer_clean(struct buffer_head * bh)
@@ -125,9 +215,9 @@ 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((void **) &bh);
+       jfs_debug(3, "freeing block %llu/%p (total %d)\n",
+                 bh->b_blocknr, (void *) bh, --bh_count);
+       ext2fs_free_mem(&bh);
 }
 
 int buffer_uptodate(struct buffer_head *bh)
@@ -135,6 +225,11 @@ int buffer_uptodate(struct buffer_head *bh)
        return bh->b_uptodate;
 }
 
+void mark_buffer_uptodate(struct buffer_head *bh, int val)
+{
+       bh->b_uptodate = val;
+}
+
 void wait_on_buffer(struct buffer_head *bh)
 {
        if (!bh->b_uptodate)
@@ -144,7 +239,7 @@ void wait_on_buffer(struct buffer_head *bh)
 
 static void e2fsck_clear_recover(e2fsck_t ctx, int error)
 {
-       ctx->fs->super->s_feature_incompat &= ~EXT3_FEATURE_INCOMPAT_RECOVER;
+       ext2fs_clear_feature_journal_needs_recovery(ctx->fs->super);
 
        /* if we had an error doing journal recovery, we need a full fsck */
        if (error)
@@ -152,142 +247,313 @@ static void e2fsck_clear_recover(e2fsck_t ctx, int error)
        ext2fs_mark_super_dirty(ctx->fs);
 }
 
-static errcode_t e2fsck_journal_init_inode(e2fsck_t ctx,
-                                          struct ext2_super_block *s,
-                                          ext2_ino_t journal_inum,
-                                          journal_t **journal)
+/*
+ * This is a helper function to check the validity of the journal.
+ */
+struct process_block_struct {
+       e2_blkcnt_t     last_block;
+};
+
+static int process_journal_block(ext2_filsys fs,
+                                blk64_t        *block_nr,
+                                e2_blkcnt_t blockcnt,
+                                blk64_t ref_block EXT2FS_ATTR((unused)),
+                                int ref_offset EXT2FS_ATTR((unused)),
+                                void *priv_data)
 {
-       struct inode *inode;
-       struct buffer_head *bh;
-       blk_t start;
-       int retval;
+       struct process_block_struct *p;
+       blk64_t blk = *block_nr;
+
+       p = (struct process_block_struct *) priv_data;
+
+       if (!blk || blk < fs->super->s_first_data_block ||
+           blk >= ext2fs_blocks_count(fs->super))
+               return BLOCK_ABORT;
+
+       if (blockcnt >= 0)
+               p->last_block = blockcnt;
+       return 0;
+}
 
-       jfs_debug(1, "Using journal inode %u\n", journal_inum);
-       *journal = e2fsck_allocate_memory(ctx, sizeof(journal_t), "journal");
-       if (!*journal) {
+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                     ret;
+       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;
        }
 
-       inode = e2fsck_allocate_memory(ctx, sizeof(*inode), "journal inode");
-       if (!inode) {
+       dev_fs = e2fsck_allocate_memory(ctx, 2*sizeof(struct kdev_s), "kdev");
+       if (!dev_fs) {
                retval = EXT2_ET_NO_MEMORY;
-               goto exit_journal;
+               goto errout;
        }
+       dev_journal = dev_fs+1;
 
-       inode->i_ctx = ctx;
-       inode->i_ino = journal_inum;
-       retval = ext2fs_read_inode(ctx->fs, journal_inum, &inode->i_ext2);
-       if (retval)
-               goto exit_inode;
-
-       (*journal)->j_dev = ctx;
-       (*journal)->j_inode = inode;
-       (*journal)->j_blocksize = ctx->fs->blocksize;
-       (*journal)->j_maxlen = inode->i_ext2.i_size / (*journal)->j_blocksize;
-
-       if (!inode->i_ext2.i_links_count ||
-           !LINUX_S_ISREG(inode->i_ext2.i_mode) ||
-           (*journal)->j_maxlen < JFS_MIN_JOURNAL_BLOCKS ||
-           (start = bmap(inode, 0)) == 0) {
-               retval = EXT2_ET_BAD_INODE_NUM;
-               goto exit_inode;
-       }
+       dev_fs->k_ctx = dev_journal->k_ctx = ctx;
+       dev_fs->k_dev = K_DEV_FS;
+       dev_journal->k_dev = K_DEV_JOURNAL;
 
-       bh = getblk(ctx, start, (*journal)->j_blocksize);
-       if (!bh) {
-               retval = EXT2_ET_NO_MEMORY;
-               goto exit_inode;
-       }
-       (*journal)->j_sb_buffer = bh;
-       (*journal)->j_superblock = (journal_superblock_t *)bh->b_data;
+       journal->j_dev = dev_journal;
+       journal->j_fs_dev = dev_fs;
+       journal->j_inode = NULL;
+       journal->j_blocksize = ctx->fs->blocksize;
 
-       return 0;
+       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;
+               }
 
-exit_inode:
-       ext2fs_free_mem((void **)&inode);
-exit_journal:
-       ext2fs_free_mem((void **)journal);
+               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 ((ret = journal_bmap(journal, 0, &start)) != 0) {
+                       retval = (errcode_t) (-1 * ret);
+                       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;
 
-       return retval;
-}
+               if (!journal_name) {
+                       fix_problem(ctx, PR_0_CANT_FIND_JOURNAL, &pctx);
+                       retval = EXT2_ET_LOAD_EXT_JOURNAL;
+                       goto errout;
+               }
 
-static errcode_t e2fsck_get_journal(e2fsck_t ctx, journal_t **journal)
-{
-       char uuid_str[40];
-       struct problem_context pctx;
-       struct ext2_super_block *sb = ctx->fs->super;
+               jfs_debug(1, "Using journal file %s\n", journal_name);
+               io_ptr = unix_io_manager;
+       }
 
-       clear_problem_context(&pctx);
+#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;
 
-       if (sb->s_feature_compat & EXT3_FEATURE_COMPAT_HAS_JOURNAL) {
-               /* FIXME: check if dev is valid block dev, has a journal */
-               if (sb->s_journal_dev) {
-                       pctx.num = sb->s_journal_dev;
-                       /* this problem aborts on -y, -p, unsupported on -n */
-                       if (!fix_problem(ctx, PR_0_JOURNAL_UNSUPP_DEV, &pctx))
-                               return EXT2_ET_UNSUPP_FEATURE;
-                       sb->s_journal_dev = 0;
-                       sb->s_state &= ~EXT2_VALID_FS;
-                       ext2fs_mark_super_dirty(ctx->fs);
+       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;
                }
-               /* FIXME: check if UUID is valid block dev, has a journal */
-               if (!uuid_is_null(sb->s_journal_uuid)) {
-                       uuid_unparse(sb->s_journal_uuid, uuid_str);
-                       pctx.str = uuid_str;
-                       /* this problem aborts on -y, -p, unsupported on -n */
-                       if (!fix_problem(ctx, PR_0_JOURNAL_UNSUPP_UUID, &pctx))
-                               return EXT2_ET_UNSUPP_FEATURE;
-                       uuid_clear(sb->s_journal_uuid);
-                       sb->s_state &= ~EXT2_VALID_FS;
-                       ext2fs_mark_super_dirty(ctx->fs);
+               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 ||
+                   !ext2fs_has_feature_journal_dev(&jsuper)) {
+                       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;
                }
-               if (!sb->s_journal_inum)
-                       return EXT2_ET_BAD_INODE_NUM;
-       }
 
-       if (sb->s_journal_dev) {
-               pctx.num = sb->s_journal_dev;
-               if (!fix_problem(ctx, PR_0_JOURNAL_BAD_DEV, &pctx))
-                       return EXT2_ET_UNSUPP_FEATURE;
-               sb->s_journal_dev = 0;
-               sb->s_state &= ~EXT2_VALID_FS;
-               ext2fs_mark_super_dirty(ctx->fs);
+               /* 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, ctx->fs, sizeof(fsx));
+                       memcpy(&superx, ctx->fs->super, sizeof(superx));
+                       fsx.super = &superx;
+                       ext2fs_set_feature_metadata_csum(fsx.super);
+                       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 (!uuid_is_null(sb->s_journal_uuid)) {
-               uuid_unparse(sb->s_journal_uuid, uuid_str);
-               pctx.str = uuid_str;
-               if (!fix_problem(ctx, PR_0_JOURNAL_BAD_UUID, &pctx))
-                       return EXT2_ET_UNSUPP_FEATURE;
-               uuid_clear(sb->s_journal_uuid);
-               sb->s_state &= ~EXT2_VALID_FS;
-               ext2fs_mark_super_dirty(ctx->fs);
+
+       if (!(bh = getblk(dev_journal, start, journal->j_blocksize))) {
+               retval = EXT2_ET_NO_MEMORY;
+               goto errout;
        }
 
-       return e2fsck_journal_init_inode(ctx, sb, sb->s_journal_inum, journal);
+       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 errcode_t e2fsck_journal_fix_bad_inode(e2fsck_t ctx,
                                              struct problem_context *pctx)
 {
        struct ext2_super_block *sb = ctx->fs->super;
-       int recover = ctx->fs->super->s_feature_incompat &
-               EXT3_FEATURE_INCOMPAT_RECOVER;
-       int has_journal = ctx->fs->super->s_feature_compat &
-               EXT3_FEATURE_COMPAT_HAS_JOURNAL;
+       int recover = ext2fs_has_feature_journal_needs_recovery(ctx->fs->super);
+       int has_journal = ext2fs_has_feature_journal(ctx->fs->super);
 
        if (has_journal || sb->s_journal_inum) {
                /* The journal inode is bogus, remove and force full fsck */
                pctx->ino = sb->s_journal_inum;
                if (fix_problem(ctx, PR_0_JOURNAL_BAD_INODE, pctx)) {
                        if (has_journal && sb->s_journal_inum)
-                               printf("*** ext3 journal has been deleted - "
-                                      "filesystem is now ext2 only ***\n\n");
-                       sb->s_feature_compat &= ~EXT3_FEATURE_COMPAT_HAS_JOURNAL;
+                               printf("*** journal has been deleted ***\n\n");
+                       ext2fs_clear_feature_journal(sb);
                        sb->s_journal_inum = 0;
+                       memset(sb->s_jnl_blocks, 0, sizeof(sb->s_jnl_blocks));
+                       ctx->flags |= E2F_FLAG_JOURNAL_INODE;
+                       ctx->fs->flags &= ~EXT2_FLAG_MASTER_SB_ONLY;
                        e2fsck_clear_recover(ctx, 1);
                        return 0;
                }
-               return EXT2_ET_BAD_INODE_NUM;
+               return EXT2_ET_CORRUPT_JOURNAL_SB;
        } else if (recover) {
                if (fix_problem(ctx, PR_0_JOURNAL_RECOVER_SET, pctx)) {
                        e2fsck_clear_recover(ctx, 1);
@@ -298,9 +564,27 @@ static errcode_t e2fsck_journal_fix_bad_inode(e2fsck_t ctx,
        return 0;
 }
 
+#define V1_SB_SIZE     0x0024
+static void clear_v2_journal_fields(journal_t *journal)
+{
+       e2fsck_t ctx = journal->j_dev->k_ctx;
+       struct problem_context pctx;
+
+       clear_problem_context(&pctx);
+
+       if (!fix_problem(ctx, PR_0_CLEAR_V2_JOURNAL, &pctx))
+               return;
+
+       ctx->flags |= E2F_FLAG_PROBLEMS_FIXED;
+       memset(((char *) journal->j_superblock) + V1_SB_SIZE, 0,
+              ctx->fs->blocksize-V1_SB_SIZE);
+       mark_buffer_dirty(journal->j_sb_buffer);
+}
+
+
 static errcode_t e2fsck_journal_load(journal_t *journal)
 {
-       e2fsck_t ctx = journal->j_dev;
+       e2fsck_t ctx = journal->j_dev->k_ctx;
        journal_superblock_t *jsb;
        struct buffer_head *jbh = journal->j_sb_buffer;
        struct problem_context pctx;
@@ -309,7 +593,7 @@ static errcode_t e2fsck_journal_load(journal_t *journal)
 
        ll_rw_block(READ, 1, &jbh);
        if (jbh->b_err) {
-               com_err(ctx->device_name, jbh->b_err,
+               com_err(ctx->device_name, jbh->b_err, "%s",
                        _("reading journal superblock\n"));
                return jbh->b_err;
        }
@@ -322,53 +606,79 @@ static errcode_t e2fsck_journal_load(journal_t *journal)
        switch (ntohl(jsb->s_header.h_blocktype)) {
        case JFS_SUPERBLOCK_V1:
                journal->j_format_version = 1;
+               if (jsb->s_feature_compat ||
+                   jsb->s_feature_incompat ||
+                   jsb->s_feature_ro_compat ||
+                   jsb->s_nr_users)
+                       clear_v2_journal_fields(journal);
                break;
-               
+
        case JFS_SUPERBLOCK_V2:
                journal->j_format_version = 2;
+               if (ntohl(jsb->s_nr_users) > 1 &&
+                   uuid_is_null(ctx->fs->super->s_journal_uuid))
+                       clear_v2_journal_fields(journal);
+               if (ntohl(jsb->s_nr_users) > 1) {
+                       fix_problem(ctx, PR_0_JOURNAL_UNSUPP_MULTIFS, &pctx);
+                       return EXT2_ET_JOURNAL_UNSUPP_VERSION;
+               }
                break;
-               
+
+       /*
+        * These should never appear in a journal super block, so if
+        * they do, the journal is badly corrupted.
+        */
+       case JFS_DESCRIPTOR_BLOCK:
+       case JFS_COMMIT_BLOCK:
+       case JFS_REVOKE_BLOCK:
+               return EXT2_ET_CORRUPT_JOURNAL_SB;
+
        /* If we don't understand the superblock major type, but there
         * is a magic number, then it is likely to be a new format we
         * just don't understand, so leave it alone. */
        default:
-               com_err(ctx->program_name, EXT2_ET_UNSUPP_FEATURE,
-                       _("%s: journal has unrecognised format\n"),
-                       ctx->device_name);
-               return EXT2_ET_UNSUPP_FEATURE;
+               return EXT2_ET_JOURNAL_UNSUPP_VERSION;
        }
 
-       if (JFS_HAS_INCOMPAT_FEATURE(journal, ~JFS_KNOWN_INCOMPAT_FEATURES)) {
-               com_err(ctx->program_name, EXT2_ET_UNSUPP_FEATURE,
-                       _("%s: journal has incompatible features\n"),
-                       ctx->device_name);
+       if (JFS_HAS_INCOMPAT_FEATURE(journal, ~JFS_KNOWN_INCOMPAT_FEATURES))
                return EXT2_ET_UNSUPP_FEATURE;
-       }
-               
-       if (JFS_HAS_RO_COMPAT_FEATURE(journal, ~JFS_KNOWN_ROCOMPAT_FEATURES)) {
-               com_err(ctx->program_name, EXT2_ET_UNSUPP_FEATURE,
-                       _("%s: journal has readonly-incompatible features\n"),
-                       ctx->device_name);
+
+       if (JFS_HAS_RO_COMPAT_FEATURE(journal, ~JFS_KNOWN_ROCOMPAT_FEATURES))
                return EXT2_ET_RO_UNSUPP_FEATURE;
-       }
+
+       /* Checksum v1-3 are mutually exclusive features. */
+       if (jfs_has_feature_csum2(journal) && jfs_has_feature_csum3(journal))
+               return EXT2_ET_CORRUPT_JOURNAL_SB;
+
+       if (journal_has_csum_v2or3(journal) &&
+           jfs_has_feature_checksum(journal))
+               return EXT2_ET_CORRUPT_JOURNAL_SB;
+
+       if (!e2fsck_journal_verify_csum_type(journal, jsb) ||
+           !e2fsck_journal_sb_csum_verify(journal, jsb))
+               return EXT2_ET_CORRUPT_JOURNAL_SB;
+
+       if (journal_has_csum_v2or3(journal))
+               journal->j_csum_seed = jbd2_chksum(journal, ~0, jsb->s_uuid,
+                                                  sizeof(jsb->s_uuid));
 
        /* We have now checked whether we know enough about the journal
         * format to be able to proceed safely, so any other checks that
         * fail we should attempt to recover from. */
        if (jsb->s_blocksize != htonl(journal->j_blocksize)) {
-               com_err(ctx->program_name, EXT2_ET_CORRUPT_SUPERBLOCK,
+               com_err(ctx->program_name, EXT2_ET_CORRUPT_JOURNAL_SB,
                        _("%s: no valid journal superblock found\n"),
                        ctx->device_name);
-               return EXT2_ET_CORRUPT_SUPERBLOCK;
+               return EXT2_ET_CORRUPT_JOURNAL_SB;
        }
 
        if (ntohl(jsb->s_maxlen) < journal->j_maxlen)
                journal->j_maxlen = ntohl(jsb->s_maxlen);
        else if (ntohl(jsb->s_maxlen) > journal->j_maxlen) {
-               com_err(ctx->program_name, EXT2_ET_CORRUPT_SUPERBLOCK,
+               com_err(ctx->program_name, EXT2_ET_CORRUPT_JOURNAL_SB,
                        _("%s: journal too short\n"),
                        ctx->device_name);
-               return EXT2_ET_CORRUPT_SUPERBLOCK;
+               return EXT2_ET_CORRUPT_JOURNAL_SB;
        }
 
        journal->j_tail_sequence = ntohl(jsb->s_sequence);
@@ -384,11 +694,17 @@ static void e2fsck_journal_reset_super(e2fsck_t ctx, journal_superblock_t *jsb,
                                       journal_t *journal)
 {
        char *p;
-       
+       union {
+               uuid_t uuid;
+               __u32 val[4];
+       } u;
+       __u32 new_seq = 0;
+       int i;
+
        /* Leave a valid existing V1 superblock signature alone.
         * Anything unrecognisable we overwrite with a new V2
         * signature. */
-       
+
        if (jsb->s_header.h_magic != htonl(JFS_MAGIC_NUMBER) ||
            jsb->s_header.h_blocktype != htonl(JFS_SUPERBLOCK_V1)) {
                jsb->s_header.h_magic = htonl(JFS_MAGIC_NUMBER);
@@ -396,19 +712,26 @@ static void e2fsck_journal_reset_super(e2fsck_t ctx, journal_superblock_t *jsb,
        }
 
        /* Zero out everything else beyond the superblock header */
-       
+
        p = ((char *) jsb) + sizeof(journal_header_t);
        memset (p, 0, ctx->fs->blocksize-sizeof(journal_header_t));
 
        jsb->s_blocksize = htonl(ctx->fs->blocksize);
        jsb->s_maxlen = htonl(journal->j_maxlen);
        jsb->s_first = htonl(1);
-       jsb->s_sequence = htonl(1);
 
-       /* In theory we should also re-zero the entire journal here.
-        * Initialising s_sequence to a random value would be a
-        * reasonable compromise. */
+       /* Initialize the journal sequence number so that there is "no"
+        * chance we will find old "valid" transactions in the journal.
+        * This avoids the need to zero the whole journal (slow to do,
+        * and risky when we are just recovering the filesystem).
+        */
+       uuid_generate(u.uuid);
+       for (i = 0; i < 4; i ++)
+               new_seq ^= u.val[i];
+       jsb->s_sequence = htonl(new_seq);
+       e2fsck_journal_sb_csum_set(journal, jsb);
 
+       mark_buffer_dirty(journal->j_sb_buffer);
        ll_rw_block(WRITE, 1, &journal->j_sb_buffer);
 }
 
@@ -417,12 +740,9 @@ static errcode_t e2fsck_journal_fix_corrupt_super(e2fsck_t ctx,
                                                  struct problem_context *pctx)
 {
        struct ext2_super_block *sb = ctx->fs->super;
-       int recover = ctx->fs->super->s_feature_incompat &
-               EXT3_FEATURE_INCOMPAT_RECOVER;
+       int recover = ext2fs_has_feature_journal_needs_recovery(ctx->fs->super);
 
-       pctx->num = journal->j_inode->i_ino;
-
-       if (sb->s_feature_compat & EXT3_FEATURE_COMPAT_HAS_JOURNAL) {
+       if (ext2fs_has_feature_journal(sb)) {
                if (fix_problem(ctx, PR_0_JOURNAL_BAD_SUPER, pctx)) {
                        e2fsck_journal_reset_super(ctx, journal->j_superblock,
                                                   journal);
@@ -430,9 +750,9 @@ static errcode_t e2fsck_journal_fix_corrupt_super(e2fsck_t ctx,
                        e2fsck_clear_recover(ctx, recover);
                        return 0;
                }
-               return EXT2_ET_CORRUPT_SUPERBLOCK;
+               return EXT2_ET_CORRUPT_JOURNAL_SB;
        } else if (e2fsck_journal_fix_bad_inode(ctx, pctx))
-               return EXT2_ET_CORRUPT_SUPERBLOCK;
+               return EXT2_ET_CORRUPT_JOURNAL_SB;
 
        return 0;
 }
@@ -446,54 +766,73 @@ static void e2fsck_journal_release(e2fsck_t ctx, journal_t *journal,
                mark_buffer_clean(journal->j_sb_buffer);
        else if (!(ctx->options & E2F_OPT_READONLY)) {
                jsb = journal->j_superblock;
-               jsb->s_sequence = htonl(journal->j_transaction_sequence);
+               jsb->s_sequence = htonl(journal->j_tail_sequence);
                if (reset)
                        jsb->s_start = 0; /* this marks the journal as empty */
-               mark_buffer_dirty(journal->j_sb_buffer, 1);
+               e2fsck_journal_sb_csum_set(journal, jsb);
+               mark_buffer_dirty(journal->j_sb_buffer);
        }
        brelse(journal->j_sb_buffer);
 
+       if (ctx->journal_io) {
+               if (ctx->fs && ctx->fs->io != ctx->journal_io)
+                       io_channel_close(ctx->journal_io);
+               ctx->journal_io = 0;
+       }
+
+#ifndef USE_INODE_IO
        if (journal->j_inode)
-               ext2fs_free_mem((void **)&journal->j_inode);
-       ext2fs_free_mem((void **)&journal);
+               ext2fs_free_mem(&journal->j_inode);
+#endif
+       if (journal->j_fs_dev)
+               ext2fs_free_mem(&journal->j_fs_dev);
+       ext2fs_free_mem(&journal);
 }
 
 /*
  * This function makes sure that the superblock fields regarding the
  * journal are consistent.
  */
-int e2fsck_check_ext3_journal(e2fsck_t ctx)
+errcode_t e2fsck_check_ext3_journal(e2fsck_t ctx)
 {
        struct ext2_super_block *sb = ctx->fs->super;
        journal_t *journal;
-       int recover = ctx->fs->super->s_feature_incompat &
-               EXT3_FEATURE_INCOMPAT_RECOVER;
+       int recover = ext2fs_has_feature_journal_needs_recovery(ctx->fs->super);
        struct problem_context pctx;
+       problem_t problem;
        int reset = 0, force_fsck = 0;
-       int retval;
+       errcode_t retval;
 
        /* If we don't have any journal features, don't do anything more */
-       if (!(sb->s_feature_compat & EXT3_FEATURE_COMPAT_HAS_JOURNAL) &&
+       if (!ext2fs_has_feature_journal(sb) &&
            !recover && sb->s_journal_inum == 0 && sb->s_journal_dev == 0 &&
            uuid_is_null(sb->s_journal_uuid))
                return 0;
 
-#ifdef JFS_DEBUG               /* Enabled by configure --enable-jfs-debug */
-       journal_enable_debug = 2;
-#endif
        clear_problem_context(&pctx);
        pctx.num = sb->s_journal_inum;
 
        retval = e2fsck_get_journal(ctx, &journal);
        if (retval) {
-               if (retval == EXT2_ET_BAD_INODE_NUM)
+               if ((retval == EXT2_ET_BAD_INODE_NUM) ||
+                   (retval == EXT2_ET_BAD_BLOCK_NUM) ||
+                   (retval == EXT2_ET_JOURNAL_TOO_SMALL) ||
+                   (retval == EXT2_ET_NO_JOURNAL))
                        return e2fsck_journal_fix_bad_inode(ctx, &pctx);
                return retval;
        }
 
        retval = e2fsck_journal_load(journal);
        if (retval) {
-               if (retval == EXT2_ET_CORRUPT_SUPERBLOCK)
+               if ((retval == EXT2_ET_CORRUPT_JOURNAL_SB) ||
+                   ((retval == EXT2_ET_UNSUPP_FEATURE) &&
+                   (!fix_problem(ctx, PR_0_JOURNAL_UNSUPP_INCOMPAT,
+                                 &pctx))) ||
+                   ((retval == EXT2_ET_RO_UNSUPP_FEATURE) &&
+                   (!fix_problem(ctx, PR_0_JOURNAL_UNSUPP_ROCOMPAT,
+                                 &pctx))) ||
+                   ((retval == EXT2_ET_JOURNAL_UNSUPP_VERSION) &&
+                   (!fix_problem(ctx, PR_0_JOURNAL_UNSUPP_VERSION, &pctx))))
                        retval = e2fsck_journal_fix_corrupt_super(ctx, journal,
                                                                  &pctx);
                e2fsck_journal_release(ctx, journal, 0, 1);
@@ -506,9 +845,8 @@ int e2fsck_check_ext3_journal(e2fsck_t ctx)
         * with -y, -n, or -p, only if a user isn't making up their mind.
         */
 no_has_journal:
-       if (!(sb->s_feature_compat & EXT3_FEATURE_COMPAT_HAS_JOURNAL)) {
-               recover = sb->s_feature_incompat & EXT3_FEATURE_INCOMPAT_RECOVER;
-               pctx.str = "inode";
+       if (!ext2fs_has_feature_journal(sb)) {
+               recover = ext2fs_has_feature_journal_needs_recovery(sb);
                if (fix_problem(ctx, PR_0_JOURNAL_HAS_JOURNAL, &pctx)) {
                        if (recover &&
                            !fix_problem(ctx, PR_0_JOURNAL_RECOVER_SET, &pctx))
@@ -526,15 +864,27 @@ no_has_journal:
                               sizeof(sb->s_journal_uuid));
                        e2fsck_clear_recover(ctx, force_fsck);
                } else if (!(ctx->options & E2F_OPT_READONLY)) {
-                       sb->s_feature_compat |= EXT3_FEATURE_COMPAT_HAS_JOURNAL;
+                       ext2fs_set_feature_journal(sb);
+                       ctx->fs->flags &= ~EXT2_FLAG_MASTER_SB_ONLY;
                        ext2fs_mark_super_dirty(ctx->fs);
                }
        }
 
-       if (sb->s_feature_compat & EXT3_FEATURE_COMPAT_HAS_JOURNAL &&
-           !(sb->s_feature_incompat & EXT3_FEATURE_INCOMPAT_RECOVER) &&
+       if (ext2fs_has_feature_journal(sb) &&
+           !ext2fs_has_feature_journal_needs_recovery(sb) &&
            journal->j_superblock->s_start != 0) {
-               if (fix_problem(ctx, PR_0_JOURNAL_RESET_JOURNAL, &pctx)) {
+               /* Print status information */
+               fix_problem(ctx, PR_0_JOURNAL_RECOVERY_CLEAR, &pctx);
+               if (ctx->superblock)
+                       problem = PR_0_JOURNAL_RUN_DEFAULT;
+               else
+                       problem = PR_0_JOURNAL_RUN;
+               if (fix_problem(ctx, problem, &pctx)) {
+                       ctx->options |= E2F_OPT_FORCE;
+                       ext2fs_set_feature_journal_needs_recovery(sb);
+                       ext2fs_mark_super_dirty(ctx->fs);
+               } else if (fix_problem(ctx,
+                                      PR_0_JOURNAL_RESET_JOURNAL, &pctx)) {
                        reset = 1;
                        sb->s_state &= ~EXT2_VALID_FS;
                        ext2fs_mark_super_dirty(ctx->fs);
@@ -544,7 +894,7 @@ no_has_journal:
                 * ignore the fact that journal apparently has data;
                 * accidentally replaying over valid data would be far
                 * worse than skipping a questionable recovery.
-                * 
+                *
                 * XXX should we abort with a fatal error here?  What
                 * will the ext3 kernel code do if a filesystem with
                 * !NEEDS_RECOVERY but with a non-zero
@@ -552,15 +902,33 @@ no_has_journal:
                 */
        }
 
+       /*
+        * If we don't need to do replay the journal, check to see if
+        * the journal's errno is set; if so, we need to mark the file
+        * system as being corrupt and clear the journal's s_errno.
+        */
+       if (!ext2fs_has_feature_journal_needs_recovery(sb) &&
+           journal->j_superblock->s_errno) {
+               ctx->fs->super->s_state |= EXT2_ERROR_FS;
+               ext2fs_mark_super_dirty(ctx->fs);
+               journal->j_superblock->s_errno = 0;
+               e2fsck_journal_sb_csum_set(journal, journal->j_superblock);
+               mark_buffer_dirty(journal->j_sb_buffer);
+       }
+
        e2fsck_journal_release(ctx, journal, reset, 0);
        return retval;
 }
 
 static errcode_t recover_ext3_journal(e2fsck_t ctx)
 {
+       struct problem_context  pctx;
        journal_t *journal;
-       int retval;
+       errcode_t retval;
 
+       clear_problem_context(&pctx);
+
+       journal_init_revoke_caches();
        retval = e2fsck_get_journal(ctx, &journal);
        if (retval)
                return retval;
@@ -572,121 +940,34 @@ static errcode_t recover_ext3_journal(e2fsck_t ctx)
        retval = journal_init_revoke(journal, 1024);
        if (retval)
                goto errout;
-       
-       retval = -journal_recover(journal);
-errout:
-       e2fsck_journal_release(ctx, journal, 1, 0);
-       return retval;
-}
-
 
-#if 0
-#define TEMPLATE "/tmp/ext3.XXXXXX"
-
-/*
- * This function attempts to mount and unmount an ext3 filesystem,
- * which is a cheap way to force the kernel to run the journal and
- * handle the recovery for us.
- */
-static errcode_t recover_ext3_journal_via_mount(e2fsck_t ctx)
-{
-       ext2_filsys fs = ctx->fs;
-       char    *dirlist[] = {"/mnt","/lost+found","/tmp","/root","/boot",0};
-       errcode_t        retval, retval2;
-       int      count = 0;
-       char     template[] = TEMPLATE;
-       struct stat buf;
-       char    *tmpdir;
-
-       if (ctx->options & E2F_OPT_READONLY) {
-               printf("%s: won't do journal recovery while read-only\n",
-                      ctx->device_name);
-               return EXT2_ET_FILE_RO;
-       }
-
-       printf(_("%s: trying for ext3 kernel journal recovery\n"),
-              ctx->device_name);
-       /*
-        * First try to make a temporary directory.  This may fail if
-        * the root partition is still mounted read-only.
-        */
-newtemp:
-       tmpdir = mktemp(template);
-       if (tmpdir) {
-               jfs_debug(2, "trying %s as ext3 temp mount point\n", tmpdir);
-               if (mkdir(template, 0700)) {
-                       if (errno == EROFS) {
-                               tmpdir = NULL;
-                               template[0] = '\0';
-                       } else if (errno == EEXIST && count++ < 10) {
-                               strcpy(template, TEMPLATE);
-                               goto newtemp;
-                       }
-                       return errno;
-               }
-       }
-
-       /*
-        * OK, creating a temporary directory didn't work.
-        * Let's try a list of possible temporary mountpoints.
-        */
-       if (!tmpdir) {
-               dev_t   rootdev;
-               char    **cpp, *dir;
-
-               if (stat("/", &buf))
-                       return errno;
-
-               rootdev = buf.st_dev;
+       retval = -journal_recover(journal);
+       if (retval)
+               goto errout;
 
-               /*
-                * Check that dir is on the same device as root (no other
-                * filesystem is mounted there), and it's a directory.
-                */
-               for (cpp = dirlist; (dir = *cpp); cpp++)
-                       if (stat(dir, &buf) == 0 && buf.st_dev == rootdev &&
-                           S_ISDIR(buf.st_mode)) {
-                               tmpdir = dir;
-                               break;
-                       }
+       if (journal->j_failed_commit) {
+               pctx.ino = journal->j_failed_commit;
+               fix_problem(ctx, PR_0_JNL_TXN_CORRUPT, &pctx);
+               journal->j_superblock->s_errno = -EINVAL;
+               mark_buffer_dirty(journal->j_sb_buffer);
        }
 
-       if (tmpdir) {
-               io_manager      io_ptr = fs->io->manager;
-               int             blocksize = fs->blocksize;
-
-               jfs_debug(2, "using %s for ext3 mount\n", tmpdir);
-               /* FIXME - need to handle loop devices here */
-               if (mount(ctx->device_name, tmpdir, "ext3", MNT_FL, NULL)) {
-                       retval = errno;
-                       com_err(ctx->program_name, errno,
-                               "when mounting %s", ctx->device_name);
-                       if (template[0])
-                               rmdir(tmpdir);
-                       return retval;
-               }
-               /*
-                * Now that it mounted cleanly, the filesystem will have been
-                * recovered, so we can now unmount it.
-                */
-               if (umount(tmpdir))
-                       return errno;
+       journal->j_tail_sequence = journal->j_transaction_sequence;
 
-               /*
-                * Remove the temporary directory, if it was created.
-                */
-               if (template[0])
-                       rmdir(tmpdir);
-               return 0;
-       }
+errout:
+       journal_destroy_revoke(journal);
+       journal_destroy_revoke_caches();
+       e2fsck_journal_release(ctx, journal, 1, 0);
+       return retval;
 }
-#endif
 
-int e2fsck_run_ext3_journal(e2fsck_t ctx)
+errcode_t e2fsck_run_ext3_journal(e2fsck_t ctx)
 {
        io_manager io_ptr = ctx->fs->io->manager;
        int blocksize = ctx->fs->blocksize;
        errcode_t       retval, recover_retval;
+       io_stats        stats = 0;
+       unsigned long long kbytes_written = 0;
 
        printf(_("%s: recovering journal\n"), ctx->device_name);
        if (ctx->options & E2F_OPT_READONLY) {
@@ -695,19 +976,26 @@ int e2fsck_run_ext3_journal(e2fsck_t ctx)
                return EXT2_ET_FILE_RO;
        }
 
-       ext2fs_flush(ctx->fs);  /* Force out any modifications */
+       if (ctx->fs->flags & EXT2_FLAG_DIRTY)
+               ext2fs_flush(ctx->fs);  /* Force out any modifications */
 
        recover_retval = recover_ext3_journal(ctx);
-       
+
        /*
         * Reload the filesystem context to get up-to-date data from disk
         * because journal recovery will change the filesystem under us.
         */
-       ext2fs_close(ctx->fs);
-       retval = ext2fs_open(ctx->filesystem_name, EXT2_FLAG_RW,
+       if (ctx->fs->super->s_kbytes_written &&
+           ctx->fs->io->manager->get_stats)
+               ctx->fs->io->manager->get_stats(ctx->fs->io, &stats);
+       if (stats && stats->bytes_written)
+               kbytes_written = stats->bytes_written >> 10;
+
+       ext2fs_mmp_stop(ctx->fs);
+       ext2fs_free(ctx->fs);
+       retval = ext2fs_open(ctx->filesystem_name, ctx->openfs_flags,
                             ctx->superblock, blocksize, io_ptr,
                             &ctx->fs);
-
        if (retval) {
                com_err(ctx->program_name, retval,
                        _("while trying to re-open %s"),
@@ -715,8 +1003,183 @@ int e2fsck_run_ext3_journal(e2fsck_t ctx)
                fatal_error(ctx, 0);
        }
        ctx->fs->priv_data = ctx;
+       ctx->fs->now = ctx->now;
+       ctx->fs->flags |= EXT2_FLAG_MASTER_SB_ONLY;
+       ctx->fs->super->s_kbytes_written += kbytes_written;
 
        /* Set the superblock flags */
-       e2fsck_clear_recover(ctx, recover_retval);
-       return recover_retval;
+       e2fsck_clear_recover(ctx, recover_retval != 0);
+
+       /*
+        * Do one last sanity check, and propagate journal->s_errno to
+        * the EXT2_ERROR_FS flag in the fs superblock if needed.
+        */
+       retval = e2fsck_check_ext3_journal(ctx);
+       return retval ? retval : recover_retval;
+}
+
+/*
+ * This function will move the journal inode from a visible file in
+ * the filesystem directory hierarchy to the reserved inode if necessary.
+ */
+static const char * const journal_names[] = {
+       ".journal", "journal", ".journal.dat", "journal.dat", 0 };
+
+void e2fsck_move_ext3_journal(e2fsck_t ctx)
+{
+       struct ext2_super_block *sb = ctx->fs->super;
+       struct problem_context  pctx;
+       struct ext2_inode       inode;
+       ext2_filsys             fs = ctx->fs;
+       ext2_ino_t              ino;
+       errcode_t               retval;
+       const char * const *    cpp;
+       dgrp_t                  group;
+       int                     mount_flags;
+
+       clear_problem_context(&pctx);
+
+       /*
+        * If the filesystem is opened read-only, or there is no
+        * journal, then do nothing.
+        */
+       if ((ctx->options & E2F_OPT_READONLY) ||
+           (sb->s_journal_inum == 0) ||
+           !ext2fs_has_feature_journal(sb))
+               return;
+
+       /*
+        * Read in the journal inode
+        */
+       if (ext2fs_read_inode(fs, sb->s_journal_inum, &inode) != 0)
+               return;
+
+       /*
+        * If it's necessary to backup the journal inode, do so.
+        */
+       if ((sb->s_jnl_backup_type == 0) ||
+           ((sb->s_jnl_backup_type == EXT3_JNL_BACKUP_BLOCKS) &&
+            memcmp(inode.i_block, sb->s_jnl_blocks, EXT2_N_BLOCKS*4))) {
+               if (fix_problem(ctx, PR_0_BACKUP_JNL, &pctx)) {
+                       memcpy(sb->s_jnl_blocks, inode.i_block,
+                              EXT2_N_BLOCKS*4);
+                       sb->s_jnl_blocks[15] = inode.i_size_high;
+                       sb->s_jnl_blocks[16] = inode.i_size;
+                       sb->s_jnl_backup_type = EXT3_JNL_BACKUP_BLOCKS;
+                       ext2fs_mark_super_dirty(fs);
+                       fs->flags &= ~EXT2_FLAG_MASTER_SB_ONLY;
+               }
+       }
+
+       /*
+        * If the journal is already the hidden inode, then do nothing
+        */
+       if (sb->s_journal_inum == EXT2_JOURNAL_INO)
+               return;
+
+       /*
+        * The journal inode had better have only one link and not be readable.
+        */
+       if (inode.i_links_count != 1)
+               return;
+
+       /*
+        * If the filesystem is mounted, or we can't tell whether
+        * or not it's mounted, do nothing.
+        */
+       retval = ext2fs_check_if_mounted(ctx->filesystem_name, &mount_flags);
+       if (retval || (mount_flags & EXT2_MF_MOUNTED))
+               return;
+
+       /*
+        * If we can't find the name of the journal inode, then do
+        * nothing.
+        */
+       for (cpp = journal_names; *cpp; cpp++) {
+               retval = ext2fs_lookup(fs, EXT2_ROOT_INO, *cpp,
+                                      strlen(*cpp), 0, &ino);
+               if ((retval == 0) && (ino == sb->s_journal_inum))
+                       break;
+       }
+       if (*cpp == 0)
+               return;
+
+       /* We need the inode bitmap to be loaded */
+       retval = ext2fs_read_bitmaps(fs);
+       if (retval)
+               return;
+
+       pctx.str = *cpp;
+       if (!fix_problem(ctx, PR_0_MOVE_JOURNAL, &pctx))
+               return;
+
+       /*
+        * OK, we've done all the checks, let's actually move the
+        * journal inode.  Errors at this point mean we need to force
+        * an ext2 filesystem check.
+        */
+       if ((retval = ext2fs_unlink(fs, EXT2_ROOT_INO, *cpp, ino, 0)) != 0)
+               goto err_out;
+       if ((retval = ext2fs_write_inode(fs, EXT2_JOURNAL_INO, &inode)) != 0)
+               goto err_out;
+       sb->s_journal_inum = EXT2_JOURNAL_INO;
+       ext2fs_mark_super_dirty(fs);
+       fs->flags &= ~EXT2_FLAG_MASTER_SB_ONLY;
+       inode.i_links_count = 0;
+       inode.i_dtime = ctx->now;
+       if ((retval = ext2fs_write_inode(fs, ino, &inode)) != 0)
+               goto err_out;
+
+       group = ext2fs_group_of_ino(fs, ino);
+       ext2fs_unmark_inode_bitmap2(fs->inode_map, ino);
+       ext2fs_mark_ib_dirty(fs);
+       ext2fs_bg_free_inodes_count_set(fs, group, ext2fs_bg_free_inodes_count(fs, group) + 1);
+       ext2fs_group_desc_csum_set(fs, group);
+       fs->super->s_free_inodes_count++;
+       return;
+
+err_out:
+       pctx.errcode = retval;
+       fix_problem(ctx, PR_0_ERR_MOVE_JOURNAL, &pctx);
+       fs->super->s_state &= ~EXT2_VALID_FS;
+       ext2fs_mark_super_dirty(fs);
+       return;
+}
+
+/*
+ * This function makes sure the superblock hint for the external
+ * journal is correct.
+ */
+int e2fsck_fix_ext3_journal_hint(e2fsck_t ctx)
+{
+       struct ext2_super_block *sb = ctx->fs->super;
+       struct problem_context pctx;
+       char uuid[37], *journal_name;
+       struct stat st;
+
+       if (!ext2fs_has_feature_journal(sb) ||
+           uuid_is_null(sb->s_journal_uuid))
+               return 0;
+
+       uuid_unparse(sb->s_journal_uuid, uuid);
+       journal_name = blkid_get_devname(ctx->blkid, "UUID", uuid);
+       if (!journal_name)
+               return 0;
+
+       if (stat(journal_name, &st) < 0) {
+               free(journal_name);
+               return 0;
+       }
+
+       if (st.st_rdev != sb->s_journal_dev) {
+               clear_problem_context(&pctx);
+               pctx.num = st.st_rdev;
+               if (fix_problem(ctx, PR_0_EXTERNAL_JOURNAL_HINT, &pctx)) {
+                       sb->s_journal_dev = st.st_rdev;
+                       ext2fs_mark_super_dirty(ctx->fs);
+               }
+       }
+
+       free(journal_name);
+       return 0;
 }