]> git.ipfire.org Git - thirdparty/e2fsprogs.git/blobdiff - e2fsck/pass1.c
e2fsck: remove get_filename_hash() prototype
[thirdparty/e2fsprogs.git] / e2fsck / pass1.c
index 799158ea059f13d8b1e6d47e21ee5751151abedc..7a82536dde57feb7b59e9b6e4c23d6d4279c9125 100644 (file)
@@ -29,7 +29,7 @@
  *
  * Pass 1 is designed to stash away enough information so that the
  * other passes should not need to read in the inode information
- * during the normal course of a filesystem check.  (Althogh if an
+ * during the normal course of a filesystem check.  (Although if an
  * inconsistency is detected, other passes may need to read in an
  * inode to fix it.)
  *
@@ -87,7 +87,7 @@ struct process_block_struct {
                                inode_modified:1;
        blk64_t         num_blocks;
        blk64_t         max_blocks;
-       e2_blkcnt_t     last_block;
+       blk64_t         last_block;
        e2_blkcnt_t     last_init_lblock;
        e2_blkcnt_t     last_db_block;
        int             num_illegal_blocks;
@@ -96,7 +96,7 @@ struct process_block_struct {
        struct problem_context *pctx;
        ext2fs_block_bitmap fs_meta_blocks;
        e2fsck_t        ctx;
-       region_t        region;
+       blk64_t         next_lblock;
        struct extent_tree_info eti;
 };
 
@@ -449,7 +449,7 @@ fix:
 }
 
 static int check_inode_extra_negative_epoch(__u32 xtime, __u32 extra) {
-       return (xtime & (1 << 31)) != 0 &&
+       return (xtime & (1U << 31)) != 0 &&
                (extra & EXT4_EPOCH_MASK) == EXT4_EPOCH_MASK;
 }
 
@@ -488,10 +488,14 @@ static void check_inode_extra_space(e2fsck_t ctx, struct problem_context *pctx)
         * implementations should never allow i_extra_isize to be 0
         */
        if (inode->i_extra_isize &&
-           (inode->i_extra_isize < min || inode->i_extra_isize > max)) {
+           (inode->i_extra_isize < min || inode->i_extra_isize > max ||
+            inode->i_extra_isize & 3)) {
                if (!fix_problem(ctx, PR_1_EXTRA_ISIZE, pctx))
                        return;
-               inode->i_extra_isize = min;
+               if (inode->i_extra_isize < min || inode->i_extra_isize > max)
+                       inode->i_extra_isize = sb->s_want_extra_isize;
+               else
+                       inode->i_extra_isize = (inode->i_extra_isize + 3) & ~3;
                e2fsck_write_inode_full(ctx, pctx->ino, pctx->inode,
                                        EXT2_INODE_SIZE(sb), "pass1");
                return;
@@ -512,7 +516,9 @@ static void check_inode_extra_space(e2fsck_t ctx, struct problem_context *pctx)
         * If the inode's extended atime (ctime, crtime, mtime) is stored in
         * the old, invalid format, repair it.
         */
-       if (sizeof(time_t) > 4 && ctx->now < EXT4_EXTRA_NEGATIVE_DATE_CUTOFF &&
+       if (((sizeof(time_t) <= 4) ||
+            (((sizeof(time_t) > 4) &&
+              ctx->now < EXT4_EXTRA_NEGATIVE_DATE_CUTOFF))) &&
            (CHECK_INODE_EXTRA_NEGATIVE_EPOCH(inode, atime) ||
             CHECK_INODE_EXTRA_NEGATIVE_EPOCH(inode, ctime) ||
             CHECK_INODE_EXTRA_NEGATIVE_EPOCH(inode, crtime) ||
@@ -695,14 +701,17 @@ isdir:
        }
 }
 
-void e2fsck_setup_tdb_icount(e2fsck_t ctx, int flags,
-                            ext2_icount_t *ret)
+extern errcode_t e2fsck_setup_icount(e2fsck_t ctx, const char *icount_name,
+                                    int flags, ext2_icount_t hint,
+                                    ext2_icount_t *ret)
 {
        unsigned int            threshold;
+       unsigned int            save_type;
        ext2_ino_t              num_dirs;
        errcode_t               retval;
        char                    *tdb_dir;
        int                     enable;
+       int                     full_map;
 
        *ret = 0;
 
@@ -717,13 +726,20 @@ void e2fsck_setup_tdb_icount(e2fsck_t ctx, int flags,
        if (retval)
                num_dirs = 1024;        /* Guess */
 
-       if (!enable || !tdb_dir || access(tdb_dir, W_OK) ||
-           (threshold && num_dirs <= threshold))
-               return;
-
-       retval = ext2fs_create_icount_tdb(ctx->fs, tdb_dir, flags, ret);
-       if (retval)
-               *ret = 0;
+       if (enable && tdb_dir && !access(tdb_dir, W_OK) &&
+           (!threshold || num_dirs > threshold)) {
+               retval = ext2fs_create_icount_tdb(ctx->fs, tdb_dir,
+                                                 flags, ret);
+               if (retval == 0)
+                       return 0;
+       }
+       e2fsck_set_bitmap_type(ctx->fs, EXT2FS_BMAP64_RBTREE, icount_name,
+                              &save_type);
+       if (ctx->options & E2F_OPT_ICOUNT_FULLMAP)
+               flags |= EXT2_ICOUNT_OPT_FULLMAP;
+       retval = ext2fs_create_icount2(ctx->fs, flags, 0, hint, ret);
+       ctx->fs->default_bitmap_type = save_type;
+       return retval;
 }
 
 static errcode_t recheck_bad_inode_checksum(ext2_filsys fs, ext2_ino_t ino,
@@ -1047,10 +1063,10 @@ void e2fsck_pass1(e2fsck_t ctx)
        struct          scan_callback_struct scan_struct;
        struct ext2_super_block *sb = ctx->fs->super;
        const char      *old_op;
-       unsigned int    save_type;
        int             imagic_fs, extent_fs, inlinedata_fs;
        int             low_dtime_check = 1;
        int             inode_size = EXT2_INODE_SIZE(fs->super);
+       int             bufsize;
        int             failed_csum = 0;
        ext2_ino_t      ino_threshold = 0;
        dgrp_t          ra_group = 0;
@@ -1143,22 +1159,18 @@ void e2fsck_pass1(e2fsck_t ctx)
                ctx->flags |= E2F_FLAG_ABORT;
                return;
        }
-       e2fsck_setup_tdb_icount(ctx, 0, &ctx->inode_link_info);
-       if (!ctx->inode_link_info) {
-               e2fsck_set_bitmap_type(fs, EXT2FS_BMAP64_RBTREE,
-                                      "inode_link_info", &save_type);
-               pctx.errcode = ext2fs_create_icount2(fs, 0, 0, 0,
-                                                    &ctx->inode_link_info);
-               fs->default_bitmap_type = save_type;
-       }
-
+       pctx.errcode = e2fsck_setup_icount(ctx, "inode_link_info", 0, NULL,
+                                          &ctx->inode_link_info);
        if (pctx.errcode) {
                fix_problem(ctx, PR_1_ALLOCATE_ICOUNT, &pctx);
                ctx->flags |= E2F_FLAG_ABORT;
                return;
        }
+       bufsize = inode_size;
+       if (bufsize < sizeof(struct ext2_inode_large))
+               bufsize = sizeof(struct ext2_inode_large);
        inode = (struct ext2_inode *)
-               e2fsck_allocate_memory(ctx, inode_size, "scratch inode");
+               e2fsck_allocate_memory(ctx, bufsize, "scratch inode");
 
        inodes_to_process = (struct process_inode_block *)
                e2fsck_allocate_memory(ctx,
@@ -1307,6 +1319,34 @@ void e2fsck_pass1(e2fsck_t ctx)
                }
                failed_csum = pctx.errcode != 0;
 
+               /*
+                * Check for inodes who might have been part of the
+                * orphaned list linked list.  They should have gotten
+                * dealt with by now, unless the list had somehow been
+                * corrupted.
+                *
+                * FIXME: In the future, inodes which are still in use
+                * (and which are therefore) pending truncation should
+                * be handled specially.  Right now we just clear the
+                * dtime field, and the normal e2fsck handling of
+                * inodes where i_size and the inode blocks are
+                * inconsistent is to fix i_size, instead of releasing
+                * the extra blocks.  This won't catch the inodes that
+                * was at the end of the orphan list, but it's better
+                * than nothing.  The right answer is that there
+                * shouldn't be any bugs in the orphan list handling.  :-)
+                */
+               if (inode->i_dtime && low_dtime_check &&
+                   inode->i_dtime < ctx->fs->super->s_inodes_count) {
+                       if (fix_problem(ctx, PR_1_LOW_DTIME, &pctx)) {
+                               inode->i_dtime = inode->i_links_count ?
+                                       0 : ctx->now;
+                               e2fsck_write_inode(ctx, ino, inode,
+                                                  "pass1");
+                               failed_csum = 0;
+                       }
+               }
+
                if (inode->i_links_count) {
                        pctx.errcode = ext2fs_icount_store(ctx->inode_link_info,
                                           ino, inode->i_links_count);
@@ -1316,6 +1356,19 @@ void e2fsck_pass1(e2fsck_t ctx)
                                ctx->flags |= E2F_FLAG_ABORT;
                                goto endit;
                        }
+               } else if ((ino >= EXT2_FIRST_INODE(fs->super)) &&
+                          !quota_inum_is_reserved(fs, ino)) {
+                       if (!inode->i_dtime && inode->i_mode) {
+                               if (fix_problem(ctx,
+                                           PR_1_ZERO_DTIME, &pctx)) {
+                                       inode->i_dtime = ctx->now;
+                                       e2fsck_write_inode(ctx, ino, inode,
+                                                          "pass1");
+                                       failed_csum = 0;
+                               }
+                       }
+                       FINISH_INODE_LOOP(ctx, ino, &pctx, failed_csum);
+                       continue;
                }
 
                /* Conflicting inlinedata/extents inode flags? */
@@ -1350,7 +1403,6 @@ void e2fsck_pass1(e2fsck_t ctx)
 
                /* Test for inline data flag but no attr */
                if ((inode->i_flags & EXT4_INLINE_DATA_FL) && inlinedata_fs &&
-                   EXT2_I_SIZE(inode) > EXT4_MIN_INLINE_DATA_SIZE &&
                    (ino >= EXT2_FIRST_INODE(fs->super))) {
                        size_t size = 0;
                        errcode_t err;
@@ -1384,15 +1436,15 @@ void e2fsck_pass1(e2fsck_t ctx)
                                /* broken EA or no system.data EA; truncate */
                                if (fix_problem(ctx, PR_1_INLINE_DATA_NO_ATTR,
                                                &pctx)) {
-                                       err = ext2fs_inode_size_set(fs, inode,
-                                                       sizeof(inode->i_block));
+                                       err = ext2fs_inode_size_set(fs, inode, 0);
                                        if (err) {
                                                pctx.errcode = err;
                                                ctx->flags |= E2F_FLAG_ABORT;
                                                goto endit;
                                        }
-                                       if (LINUX_S_ISLNK(inode->i_mode))
-                                               inode->i_flags &= ~EXT4_INLINE_DATA_FL;
+                                       inode->i_flags &= ~EXT4_INLINE_DATA_FL;
+                                       memset(&inode->i_block, 0,
+                                              sizeof(inode->i_block));
                                        e2fsck_write_inode(ctx, ino, inode,
                                                           "pass1");
                                        failed_csum = 0;
@@ -1529,7 +1581,7 @@ void e2fsck_pass1(e2fsck_t ctx)
                        /*
                         * Make sure the root inode is a directory; if
                         * not, offer to clear it.  It will be
-                        * regnerated in pass #3.
+                        * regenerated in pass #3.
                         */
                        if (!LINUX_S_ISDIR(inode->i_mode)) {
                                if (fix_problem(ctx, PR_1_ROOT_NO_DIR, &pctx))
@@ -1633,48 +1685,7 @@ void e2fsck_pass1(e2fsck_t ctx)
                        continue;
                }
 
-               /*
-                * Check for inodes who might have been part of the
-                * orphaned list linked list.  They should have gotten
-                * dealt with by now, unless the list had somehow been
-                * corrupted.
-                *
-                * FIXME: In the future, inodes which are still in use
-                * (and which are therefore) pending truncation should
-                * be handled specially.  Right now we just clear the
-                * dtime field, and the normal e2fsck handling of
-                * inodes where i_size and the inode blocks are
-                * inconsistent is to fix i_size, instead of releasing
-                * the extra blocks.  This won't catch the inodes that
-                * was at the end of the orphan list, but it's better
-                * than nothing.  The right answer is that there
-                * shouldn't be any bugs in the orphan list handling.  :-)
-                */
-               if (inode->i_dtime && low_dtime_check &&
-                   inode->i_dtime < ctx->fs->super->s_inodes_count) {
-                       if (fix_problem(ctx, PR_1_LOW_DTIME, &pctx)) {
-                               inode->i_dtime = inode->i_links_count ?
-                                       0 : ctx->now;
-                               e2fsck_write_inode(ctx, ino, inode,
-                                                  "pass1");
-                               failed_csum = 0;
-                       }
-               }
-
-               /*
-                * This code assumes that deleted inodes have
-                * i_links_count set to 0.
-                */
                if (!inode->i_links_count) {
-                       if (!inode->i_dtime && inode->i_mode) {
-                               if (fix_problem(ctx,
-                                           PR_1_ZERO_DTIME, &pctx)) {
-                                       inode->i_dtime = ctx->now;
-                                       e2fsck_write_inode(ctx, ino, inode,
-                                                          "pass1");
-                                       failed_csum = 0;
-                               }
-                       }
                        FINISH_INODE_LOOP(ctx, ino, &pctx, failed_csum);
                        continue;
                }
@@ -1709,11 +1720,11 @@ void e2fsck_pass1(e2fsck_t ctx)
                if (inode->i_faddr || frag || fsize ||
                    (LINUX_S_ISDIR(inode->i_mode) && inode->i_dir_acl))
                        mark_inode_bad(ctx, ino);
-               if ((fs->super->s_creator_os == EXT2_OS_LINUX) &&
+               if ((fs->super->s_creator_os != EXT2_OS_HURD) &&
                    !ext2fs_has_feature_64bit(fs->super) &&
                    inode->osd2.linux2.l_i_file_acl_high != 0)
                        mark_inode_bad(ctx, ino);
-               if ((fs->super->s_creator_os == EXT2_OS_LINUX) &&
+               if ((fs->super->s_creator_os != EXT2_OS_HURD) &&
                    !ext2fs_has_feature_huge_file(fs->super) &&
                    (inode->osd2.linux2.l_i_blocks_hi != 0))
                        mark_inode_bad(ctx, ino);
@@ -1814,9 +1825,14 @@ void e2fsck_pass1(e2fsck_t ctx)
                     inode->i_block[EXT2_DIND_BLOCK] ||
                     inode->i_block[EXT2_TIND_BLOCK] ||
                     ext2fs_file_acl_block(fs, inode))) {
+                       struct ext2_inode_large *ip;
+
                        inodes_to_process[process_inode_count].ino = ino;
-                       inodes_to_process[process_inode_count].inode =
-                                      *(struct ext2_inode_large *)inode;
+                       ip = &inodes_to_process[process_inode_count].inode;
+                       if (inode_size < sizeof(struct ext2_inode_large))
+                               memcpy(ip, inode, inode_size);
+                       else
+                               memcpy(ip, inode, sizeof(*ip));
                        process_inode_count++;
                } else
                        check_blocks(ctx, &pctx, block_buf);
@@ -2512,7 +2528,7 @@ void e2fsck_clear_inode(e2fsck_t ctx, ext2_ino_t ino,
  * line up.
  */
 static int has_unaligned_cluster_map(e2fsck_t ctx,
-                                    blk64_t last_pblk, e2_blkcnt_t last_lblk,
+                                    blk64_t last_pblk, blk64_t last_lblk,
                                     blk64_t pblk, blk64_t lblk)
 {
        blk64_t cluster_mask;
@@ -2612,12 +2628,21 @@ static void scan_extent_node(e2fsck_t ctx, struct problem_context *pctx,
                        problem = PR_1_EXTENT_ENDS_BEYOND;
                else if (is_leaf && is_dir &&
                         ((extent.e_lblk + extent.e_len) >
-                         (1 << (21 - ctx->fs->super->s_log_block_size))))
+                         (1U << (21 - ctx->fs->super->s_log_block_size))))
                        problem = PR_1_TOOBIG_DIR;
 
-               if (is_leaf && problem == 0 && extent.e_len > 0 &&
-                   region_allocate(pb->region, extent.e_lblk, extent.e_len))
-                       problem = PR_1_EXTENT_COLLISION;
+               if (is_leaf && problem == 0 && extent.e_len > 0) {
+#if 0
+                       printf("extent_region(ino=%u, expect=%llu, "
+                              "lblk=%llu, len=%u)\n",
+                              pb->ino, pb->next_lblock,
+                              extent.e_lblk, extent.e_len);
+#endif
+                       if (extent.e_lblk < pb->next_lblock)
+                               problem = PR_1_EXTENT_COLLISION;
+                       else if (extent.e_lblk + extent.e_len > pb->next_lblock)
+                               pb->next_lblock = extent.e_lblk + extent.e_len;
+               }
 
                /*
                 * Uninitialized blocks in a directory?  Clear the flag and
@@ -2697,6 +2722,7 @@ report_problem:
                         * will reallocate the block; then we can try again.
                         */
                        if (pb->ino != EXT2_RESIZE_INO &&
+                           extent.e_pblk < ctx->fs->super->s_blocks_count &&
                            ext2fs_test_block_bitmap2(ctx->block_metadata_map,
                                                      extent.e_pblk)) {
                                next_try_repairs = 0;
@@ -2704,7 +2730,8 @@ report_problem:
                                fix_problem(ctx,
                                            PR_1_CRITICAL_METADATA_COLLISION,
                                            pctx);
-                               ctx->flags |= E2F_FLAG_RESTART_LATER;
+                               if ((ctx->options & E2F_OPT_NO) == 0)
+                                       ctx->flags |= E2F_FLAG_RESTART_LATER;
                        }
                        pctx->errcode = ext2fs_extent_get(ehandle,
                                                  EXT2_EXTENT_DOWN, &extent);
@@ -2781,7 +2808,7 @@ report_problem:
                 * pass 3 allocating empty directory blocks to fill the hole.
                 */
                if (try_repairs && is_dir &&
-                   pb->last_block + 1 < (e2_blkcnt_t)extent.e_lblk) {
+                   pb->last_block + 1 < extent.e_lblk) {
                        blk64_t new_lblk;
 
                        new_lblk = pb->last_block + 1;
@@ -2948,13 +2975,7 @@ static void check_blocks_extents(e2fsck_t ctx, struct problem_context *pctx,
        memset(pb->eti.ext_info, 0, sizeof(pb->eti.ext_info));
        pb->eti.ino = pb->ino;
 
-       pb->region = region_create(0, info.max_lblk);
-       if (!pb->region) {
-               ext2fs_extent_free(ehandle);
-               fix_problem(ctx, PR_1_EXTENT_ALLOC_REGION_ABORT, pctx);
-               ctx->flags |= E2F_FLAG_ABORT;
-               return;
-       }
+       pb->next_lblock = 0;
 
        eof_lblk = ((EXT2_I_SIZE(inode) + fs->blocksize - 1) >>
                EXT2_BLOCK_SIZE_BITS(fs->super)) - 1;
@@ -2967,8 +2988,6 @@ static void check_blocks_extents(e2fsck_t ctx, struct problem_context *pctx,
                                   "check_blocks_extents");
                pctx->errcode = 0;
        }
-       region_free(pb->region);
-       pb->region = NULL;
        ext2fs_extent_free(ehandle);
 
        /* Rebuild unless it's a dir and we're rehashing it */
@@ -3049,7 +3068,7 @@ static void check_blocks(e2fsck_t ctx, struct problem_context *pctx,
 
        pb.ino = ino;
        pb.num_blocks = 0;
-       pb.last_block = -1;
+       pb.last_block = ~0;
        pb.last_init_lblock = -1;
        pb.last_db_block = -1;
        pb.num_illegal_blocks = 0;
@@ -3059,7 +3078,7 @@ static void check_blocks(e2fsck_t ctx, struct problem_context *pctx,
        pb.previous_block = 0;
        pb.is_dir = LINUX_S_ISDIR(inode->i_mode);
        pb.is_reg = LINUX_S_ISREG(inode->i_mode);
-       pb.max_blocks = 1 << (31 - fs->super->s_log_block_size);
+       pb.max_blocks = 1U << (31 - fs->super->s_log_block_size);
        pb.inode = inode;
        pb.pctx = pctx;
        pb.ctx = ctx;
@@ -3166,7 +3185,8 @@ static void check_blocks(e2fsck_t ctx, struct problem_context *pctx,
        if (ino != quota_type2inum(PRJQUOTA, fs->super) &&
            (ino == EXT2_ROOT_INO || ino >= EXT2_FIRST_INODE(ctx->fs->super))) {
                quota_data_add(ctx->qctx, (struct ext2_inode_large *) inode,
-                              ino, pb.num_blocks * fs->blocksize);
+                              ino,
+                              pb.num_blocks * EXT2_CLUSTER_SIZE(fs->super));
                quota_data_inodes(ctx->qctx, (struct ext2_inode_large *) inode,
                                  ino, +1);
        }
@@ -3176,29 +3196,28 @@ static void check_blocks(e2fsck_t ctx, struct problem_context *pctx,
                pb.num_blocks *= (fs->blocksize / 512);
        pb.num_blocks *= EXT2FS_CLUSTER_RATIO(fs);
 #if 0
-       printf("inode %u, i_size = %u, last_block = %lld, i_blocks=%llu, num_blocks = %llu\n",
+       printf("inode %u, i_size = %u, last_block = %llu, i_blocks=%llu, num_blocks = %llu\n",
               ino, inode->i_size, pb.last_block, ext2fs_inode_i_blocks(fs, inode),
               pb.num_blocks);
 #endif
        if (pb.is_dir) {
-               int nblock = inode->i_size >> EXT2_BLOCK_SIZE_BITS(fs->super);
+               unsigned nblock = inode->i_size >> EXT2_BLOCK_SIZE_BITS(fs->super);
                if (inode->i_flags & EXT4_INLINE_DATA_FL) {
                        int flags;
-                       size_t size;
+                       size_t sz = 0;
                        errcode_t err;
 
-                       size = 0;
                        flags = ctx->fs->flags;
                        ctx->fs->flags |= EXT2_FLAG_IGNORE_CSUM_ERRORS;
                        err = ext2fs_inline_data_size(ctx->fs, pctx->ino,
-                                                     &size);
+                                                     &sz);
                        ctx->fs->flags = (flags &
                                          EXT2_FLAG_IGNORE_CSUM_ERRORS) |
                                         (ctx->fs->flags &
                                          ~EXT2_FLAG_IGNORE_CSUM_ERRORS);
-                       if (err || size != inode->i_size) {
+                       if (err || sz != inode->i_size) {
                                bad_size = 7;
-                               pctx->num = size;
+                               pctx->num = sz;
                        }
                } else if (inode->i_size & (fs->blocksize - 1))
                        bad_size = 5;
@@ -3253,7 +3272,7 @@ static void check_blocks(e2fsck_t ctx, struct problem_context *pctx,
        if (LINUX_S_ISREG(inode->i_mode) &&
            ext2fs_needs_large_file_feature(EXT2_I_SIZE(inode)))
                ctx->large_files++;
-       if ((fs->super->s_creator_os == EXT2_OS_LINUX) &&
+       if ((fs->super->s_creator_os != EXT2_OS_HURD) &&
            ((pb.num_blocks != ext2fs_inode_i_blocks(fs, inode)) ||
             (ext2fs_has_feature_huge_file(fs->super) &&
              (inode->i_flags & EXT4_HUGE_FILE_FL) &&
@@ -3435,10 +3454,12 @@ static int process_block(ext2_filsys fs,
         */
        if (blockcnt < 0 &&
            p->ino != EXT2_RESIZE_INO &&
+           blk < ctx->fs->super->s_blocks_count &&
            ext2fs_test_block_bitmap2(ctx->block_metadata_map, blk)) {
                pctx->blk = blk;
                fix_problem(ctx, PR_1_CRITICAL_METADATA_COLLISION, pctx);
-               ctx->flags |= E2F_FLAG_RESTART_LATER;
+               if ((ctx->options & E2F_OPT_NO) == 0)
+                       ctx->flags |= E2F_FLAG_RESTART_LATER;
        }
 
        if (problem) {
@@ -3927,7 +3948,7 @@ static void mark_table_blocks(e2fsck_t ctx)
 }
 
 /*
- * Thes subroutines short circuits ext2fs_get_blocks and
+ * These subroutines short circuits ext2fs_get_blocks and
  * ext2fs_check_directory; we use them since we already have the inode
  * structure, so there's no point in letting the ext2fs library read
  * the inode again.