diff options
Diffstat (limited to 'e2fsck/ChangeLog')
| -rw-r--r-- | e2fsck/ChangeLog | 4280 |
1 files changed, 0 insertions, 4280 deletions
diff --git a/e2fsck/ChangeLog b/e2fsck/ChangeLog deleted file mode 100644 index bb03b0a7..00000000 --- a/e2fsck/ChangeLog +++ /dev/null @@ -1,4280 +0,0 @@ -2006-06-24 Theodore Ts'o <tytso@mit.edu> - - * Release of E2fsprogs 1.40 - -2007-06-22 Theodore Tso <tytso@mit.edu> - - * pass1.c (check_ext_attr): Adds an extra validity test in - check_ext_attr(). If an attribute's e_value_size is zero - the current code does not allocate a region for it and as - a result the e_value_offs value is not verified. However, - if e_value_offs is very large then the later call to - ext2fs_ext_attr_hash_entry() can dereference bad memory - and crash e2fsck. - -2007-06-18 Theodore Tso <tytso@mit.edu> - - * journal.c (e2fsck_run_ext3_journal), unix.c (main): Explicitly - add the EXT2_FLAG_MASTER_SB_ONLY flag to make sure we - won't write out the backup superblocks until we're sure - that we want write them out. - -2007-06-04 Theodore Tso <tytso@mit.edu> - - * super.c (check_super_block, e2fsck_get_device_size), - unix.c (main): Move device size logic into the main(), to - support the Linux floppy driver, which disallows another - (non-O_EXCL) open() if it has been opened with O_EXCL. - (Addresses Debian Bug: #410569) - - * e2fsck.h (E2F_FLAG_GOT_DEVSIZE): New flag used in get_device logic - -2007-05-31 Theodore Tso <tytso@mit.edu> - - * util.c (get_backup_sb): Do basic sanity checking to make sure - the blocksize of the superblock passes muster. - - * profile.c (profile_node_iterator): Avoid potential null - dereference bug. - -2007-05-22 Theodore Tso <tytso@mit.edu> - - * message.c (expand_inode_expression, expand_percent_expression): - Fix gcc -Wall warnings on 64-bit systems. - - * pass1.c (e2fsck_pass1_check_device_inode): Mark the fs parameter - as unused. - (e2fsck_setup_tdb_icount): Use profile_get_uint() function - for reading the threshold. Remove unused variables. - - * dirinfo.c (setup_tdb): Use the profile_get_uint() function for - reading the threshold. - (setup_db): Declare it as static - (setup_tdb): #include uuid.h since we use a uuid function - (e2fsck_get_dir_info, e2fsck_dir_info_iter_begin): Remove - unused variables - (e2fsck_dir_info_iter_end): Mark the ctx parameter as - unused. - - * profile.h, profile.c (profile_get_uint): New function which - returns an unsigned integer. - -2007-05-08 Kalpak Shah <kalpak@clusterfs.com> - - * pass1.c (check_ea_in_inode): Remove check that requires in-inode - EA's to be sorted, since they don't have to be. - -2007-05-08 Theodore Tso <tytso@mit.edu> - - * message.c (expand_inode_expression): Use inode_uid() and - inode_gid() macros when printing the uid/gid. - -2007-04-18 Theodore Tso <tytso@mit.edu> - - * e2fsck.conf.5.in (apply): Fix minor typo. - -2007-04-14 Theodore Tso <tytso@mit.edu> - - * pass2.c (e2fsck_process_bad_inode): Remove special kludge that - dealt with long symlinks on big endian systems. It turns - out this was a workaround to a bug described in Red Hat - Bugzilla #232663, with an odd twist. See comment #12 for - more details. - - * pass1.c, pass2.c, util.c: Add better ehandler_operation() - markers so it is clearer what e2fsck was doing when an I/O - error is reported. - - *pass1.c (check_is_really_dir): Fix a typo which could cause - e2fsck to throw an I/O error while doubling checking - whether or not a special device file was really an inode. - Also, don't do this tests on symbolic links since for - filesystems with a large numbers of symlinks it could - degrade performance and increases the risk for false - positives. - -2007-04-10 Jim Garlick <garlick@llnl.gov> - - * pass1b.c (search_dirent_proc): if a file has multiple hard - links, e2fsck pass1c search_dirent_proc() doesn't maintain - its count properly and may return DIRENT_ABORT before it - has found containing directories for all inodes sharing - blocks. - -2007-04-06 Theodore Tso <tytso@mit.edu> - - * e2fsck.conf.5.in: Update man page to document the scratch_files - section. - - * dirinfo.c: If e2fsck.conf configures a scratch_files directory - which is available, and the number of directories exceeds - scratch_files.numdirs_threshold, then try to use the tdb - library to store the directory information abstraction. - This allows us to check very large filesystems without - needing as much physical memory. - - * e2fsck.h, pass1.c, pass2.c: If e2fsck.conf configures a - scratch_files directory which is available, and the number - of directories exceeds scratch_files.numdirs_threshold, - then try to use the tdb library to store the inode count - abstraction. This allows us to check very large - filesystems without needing as much physical memory. - -2007-04-04 Theodore Tso <tytso@mit.edu> - - * dirinfo.c, pass2.c, pass3.c, rehash.c, e2fsck.h: Change the - iterator abstraction and replace e2fsck_get_dir_info() - with e2fsck_dir_info_{set,get}_{parent,dotdot} so that we - can support an on-disk dirinfo implementation. This - allows e2fsck to check very large filesystems on systems - with smaller amounts of memory and/or address space. - - * pass1.c (check_ext_attr): Make sure the region variable is - initialized to zero to avoid checking and then freeing an - uninitialized pointer if there is a corrupted EA block. - -2007-03-31 Theodore Tso <tytso@mit.edu> - - * pass1.c (e2fsck_pass1, check_is_really_dir): Check for an edge - condition where the mode of a directory is incorrect, and - looks like a special device, but it is really a directory. - We can't do this for regular files because of the - performance hit, but this will catch directories which - have their i_mode bits mutated so they looks like a - special device. - - * problem.c, problem.h (PR_1_TREAT_AS_DIRECTORY): New problem code - - * message.c (expand_percent_expression): Add support for %It, which - will print the type of the inode. - - * pass3.c (fix_dotdot_proc): Fix the filetype of the '..' entry to - be EXT2_FT_DIR. (Addresses Lustre BZ #11645) - - * pass1.c (e2fsck_pass1_check_device_inode): Don't assume that a - special device is bogus just because i_blocks is non-zero. - The i_blocks field could get adjusted later, and if this - happens it will confuse the e2fsck_process_bad_inode() in - pass 2. In practice true garbage inodes will have random - non-zero values in i_blocks[4..15], so there's no point - doing the check for an illegal i_blocks value. - -2007-03-28 Theodore Tso <tytso@mit.edu> - - * pass1.c (e2fsck_pass1, check_ext_attr), - pass5.c (check_block_bitmaps, check_inode_bitmaps), - pass4.c (e2fsck_pass4), swapfs.c (swap_inodes), - journal.c (e2fsck_get_journal), - unix.c (parse_extended_opts): Fix memory leaks - -2007-03-21 Theodore Tso <tytso@mit.edu> - - * profile.c (profile_open_file): Fix memory leak if malloc() fails - while setting up the profile data structure. - -2007-03-19 Theodore Tso <tytso@mit.edu> - - * pass3.c (check_directory): Add error check in case - e2fsck_get_dir_info() returns NULL. Also fix another - error check for e2fsck_get_dir_info() to display the - correct inode number in case of this internal (should - never happen) error. - - * pass1b.c (clone_file): Fix a coverity-found bug; add error - checking in case dict_lookup() returns NULL when looking up - an block or inode record after cloning the EA block. - - * profile.c (profile_init, get_dirlist): Fix bug where if a - profile directory is completely empty, the profile library - would segfault. - -2006-12-22 Theodore Tso <tytso@mit.edu> - - * unix.c (PRS, main): Use the new {add,remove}_error_table comerr - interfaces instead of initialize_*_error_table. - -2006-11-14 Theodore Tso <tytso@mit.edu> - - * unix.c (PRS): Always allocate the replacement PATH environment - passed to putenv() to avoid gcc -Wall warning. - - * pass1.c, pass2.c, profile.c, super.c: Remove unused variables - and fixed signed vs unsigned and const gcc -Wall warning. - - * message.c (expand_inode_expression): Fix const gcc -Wall warning. - - * journal.c (e2fsck_fix_ext3_journal_hint): Remove unusued - variables retval and problem. - -2006-11-11 Theodore Tso <tytso@mit.edu> - - * super.c (e2fsck_fix_dirhash_hint, check_super_block): If neither - the signed or unsigned dirhash hint, set it based on - default signed vs. unsigned character type in use by the - platform. - - * problem.c, problem.h (PR_0_DIRHASH_HINT): Add new problem code. - - * pass2.c (check_dir_block), rehash.c (fill_dir_block): Check the - superblock flags to determine whether to use the signed or - unsigned version of the hash should be used. - - * problem.c, problem.h (PR_2_BLOCKS_HI_ZERO): Add new problem code. - - * pass1.c (e2fsck_pass1), pass2.c (e2fsck_process_bad_inode): - Replace check for l_i_frag and l_i_fsize with one for - i_blocks_hi. - -2006-11-08 Theodore Tso <tytso@mit.edu> - - * badblocks.c (read_bad_blocks_file): Change the last_block - argument passed to the badblocks program to be consistent - with the change to badblocks. - -2006-10-21 Theodore Tso <tytso@mit.edu> - - * pass2.c (parse_int_node): Don't core dump if there is a corrupt - htree interior node. If the block number is larger than - the number of blocks in the directory, don't write past - the end of malloc'ed memory. (Addresses SourceForge Bug: - #1512778) - -2006-10-02 Theodore Tso <tytso@mit.edu> - - * e2fsck.conf.5.in: Minor correction to man page. - (Addresses Debian bug #369761) - -2006-10-01 Theodore Tso <tytso@mit.edu> - - * Makefile.in (DEPLIBBLKID): Use DEPLIBBLKID not LIBBLKID to - define the dependencies used by the blkid library. - -2006-08-30 Eric Sandeen <esandeen@redhat.com> - - * pass1.c (handle_bad_fs_blocks): use blk_t, not int for first_block. - -2006-08-30 Eric Sandeen <esandeen@redhat.com> - - * unix.c (show_stats): use ext2_ino_t for inode containers. - -2006-08-30 Eric Sandeen <esandeen@redhat.com> - - * pass1.c (new_table_block, handle_fs_bad_blocks): - * super.c (check_super_block): - Use new inlines to calculate group first & last blocks. - -2006-08-30 Eric Sandeen <esandeen@redhat.com> - - * e2fsck.h (e2fsck): Use unsigned types for filesystem counters. - * emptydir.c (add_empty_dirblock): - * iscan.c (main): - * unix.c (show_stats, check_if_skip): Fix printf formats. - -2006-08-30 Eric Sandeen <esandeen@redhat.com> - - * pass1.c (handle_fs_bad_blocks): Remove unused variables. - -2006-08-30 Eric Sandeen <esandeen@redhat.com> - - * pass1b.c (check_if_fs_block): Change block group loop to use - a common pattern of first_block/last_block, etc. - - * super.c (check_super_block): Avoid overflows when iterating over - group descriptors on very large filesystems - -2006-08-30 Theodore Tso <tytso@mit.edu> - - * pass5.c (check_inode_bitmaps, check_inode_end, check_block_end): - * pass4.c (e2fsck_pass4): Fix potential overflow problems when the - number of blocks is close to 2**31. - -2006-08-29 Theodore Tso <tytso@mit.edu> - - * super.c (release_inode_blocks): Fix silly spelling error. - (Addresses SourceForge Bug: #1531372) - -2006-05-29 Theodore Tso <tytso@mit.edu> - - * pass1b.c: Add missing semicolon when HAVE_INTPTR_T is not defined - -2006-05-22 Theodore Tso <tytso@mit.edu> - - * e2fsck.8.in: Fixed spelling mistake. (Addresses Debian Bug: - #368392) - -2006-05-14 Theodore Tso <tytso@mit.edu> - - * recovery.c (do_one_pass), journal.c (e2fsck_get_journal): Add - missing brelse() calls to avoid memory leaks in error - paths. (Thanks to Michael C. Thompson for pointing these - out; they were originally found using Coverity.) - -2006-05-13 Theodore Tso <tytso@mit.edu> - - * badblocks.c (read_bad_blocks_file): Pass the private option -X - to badblocks so that e2fsck -cc will work correctly. - -2006-05-08 Theodore Tso <tytso@mit.edu> - - * pass5.c (check_block_bitmaps, check_inode_bitmaps): Add support - for the lazy_bg feature; if the block group does not have - an initialized block or inode bitmaps/table, emulate what - the allocation bitmap would look like if no blocks or - inodes have been allocated. - -2006-03-27 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.8.in: Add badblocks(8) to the See Also section. - (Addresses Debian Bug: #347295) - - * Makefile.in, e2fsck.conf.5.in: Add man page documenting the - configuration file /etc/e2fsck.conf. - -2006-03-22 Theodore Ts'o <tytso@mit.edu> - - * profile.c (profile_update_file, profile_parse_file, parse_line, - parse_std_line): Integrate profile_std_line() into - parse_line(), and profile_parse_file() into - profile_update_file() to make the code tighter. - -2006-03-18 Theodore Ts'o <tytso@mit.edu> - - * unix.c (show_stats): Use blk_t instead of int when referring to - block numbers. - - * emptydir.c, message.c, pass1b.c, pass2.c, recovery.c, unix.c: - Change printf statements to use %u instead of %d when - printing block numbers. - - * unix.c (main): If the filesystem is not mounted, open the - filesystem in exclusive mode (to prevent 2.6 kernels from - mounting the filesystem while it is being checked). - -2006-03-10 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.h, journal.c (e2fsck_fix_ext3_journal_hint), - problem.c (PR_0_EXTERNAL_JOURNAL_HINT), - problem.h (PR_0_EXTERNAL_JOURNAL_HINT), super.c: Check - to see if the superblock hint for the external journal - needs to be updated, and if so, offer to update it. - (Addresses Debian Bug: #355644) - -2006-01-29 Theodore Ts'o <tytso@mit.edu> - - * unix.c (check_if_skip): When skipping a check due to being on - battery, print an explanatory message to this effect. Add - an e2fsck configuration option defer_check_on_battery - which defaults to true. If the configuration option is - turned off, then e2fsck will not attempt to defer - filesystem checks when the system is running on battery. - (Addresses Debain Bug: #350306) - -2006-01-06 Theodore Ts'o <tytso@mit.edu> - - * profile.c (parse_std_line, dump_profile): Add support for - quoted strings in tag and section names. Changed - top-level section name parsing to ignore leading and - trailing whitespace. - -2006-01-05 Theodore Ts'o <tytso@mit.edu> - - * profile.c (profile_init): If a directory is passed to - profile_init, then read all files which match [0-9a-zA-Z]* - in sort order. This allows handling of /etc/foo/conf.d - style directories. - -2006-01-04 Theodore Ts'o <tytso@mit.edu> - - * profile.c: Eliminate the prf_data_t data structure and fold it - back into prof_file_t. Eschew needless complexity! - - * profile.c: Change syntax accepted by the profile parser. The - value of profile relations must not contain spaces unless - it is quoted; otherwise the parser will throw an error. - Comments are allowed anywhere and can be started with - either a ';' or a '#' character. The only place where - comments will not be intepreted as beginning a comment is - in a quoted string. - - * profile.c (profile_parse_file, profile_set_syntax_err_cb): Add - support for a syntax error callback to allow better error - reporting to the user. - - * e2fsck(PRS): Set a profile syntax error callback so we can - report errors in /etc/e2fsck.conf. - -2006-01-02 Theodore Ts'o <tytso@mit.edu> - - * profile.c, profile_helpers.c, profile_helpers.h: Boil down and - simplify profile.c, and move extra functionality that's - not needed into profile_helpers.c - -2006-01-01 Theodore Ts'o <tytso@mit.edu> - - * profile.c: Fix #include of com_err.h so that it isn't required - that the com_err development environment be installed. - (Addresses Debian Bug: #345519) - -2005-12-31 Theodore Ts'o <tytso@mit.edu> - - * problem.c (fix_problem), problemP.h: Add the ability for the - e2fsck configuration file to override the behaviour of - e2fsck when a particular filesystem problem is - encountered. This allows reconnecting an inode to - lost+found to not stop the boot sequence, if a system - administrator really badly wants this behaviour for some - specialized reason, for example. - - * unix.c (main): If the e2fsck configuration file sets the - allow_cancellation option to be true, then if the - filesystem does not have any known problems, and was known - to be cleanly unmounted, then let e2fsck exit with a - status code of 0 instead of 32 (FSCK_CANCELED) so that the - bootup scripts will continue without stopping the boot. - (Addresses Debian Bug: #150295) - -2005-12-30 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.c (e2fsck_free_context), unix.c (PRS), e2fsck.h: Use - the profile library to read in the /etc/e2fsck.conf file - at startup. - - * profile.c, profile.h, prof_err.et: Add the profile library code - to e2fsck. The profile library was originally written by - Theodore Ts'o in 1995 for use in the MIT Kerberos v5 - library. It has been modified/enhanced/bug-fixed over - time by other members of the MIT Kerberos team. This - version was originally taken from the Kerberos v5 - distribution, version 1.4.2, and radically simplified for - use in e2fsprogs. (Support for locking for multi-threaded - operations, being able to modify and update the - configuration file programmatically, and Mac/Windows - portability have been removed. It has been folded into a - single C source file to make it easier to fold into an - application program.) - - * argv_parse.c, argv_parse.h: Added auxiliary programs used only - for building the profile test program. - -2005-12-18 Theodore Ts'o <tytso@mit.edu> - - * problem.c (PR_0_FUTURE_SB_LAST_MOUNT, - PR_0_FUTURE_SB_LAST_WRITE): Make these problems things - which can be fixd by preen, since Debian's boot sequence - bogusly doesn't set the time correctly until potentially - very late in the bootup process, and this can cause false - positives that will cause users' systems to fail to - booting. (Addresses Debian Bugs #343662 and #343645) - -2005-12-09 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.8.in: Clarify that e2fsck -c does a read-only scan of the - device. - - * unix.c (PRS): Fix typo in error message; thanks to Solar - Designer for pointing out the missing preposition. - (check_mount): Don't let the user run e2fsck -ccn on the - root partition, without warning that he or she might - be doing something Really Stupid. - - * e2fsck.8.in: Clarify description of differences between the -p, - -n, and -y options. - -2005-09-24 Theodore Ts'o <tytso@mit.edu> - - * super.c (check_super_block), problem.c, problem.h: Detect if the - superblock's last mount field or last write field is in - the future, and offer to fix if so. (Addresses Debian Bug - #327580) - -2005-07-25 Theodore Ts'o <tytso@mit.edu> - - * unix.c (main): Fix a use-after-free bug of the e2fsck context - structure at the very end of the e2fsck run. - - * pass1.c (pass1_write_inode): Fix false positive from valgrind; - don't do a needless structure copy via an assignment when - it is a no-op. - -2005-07-04 Theodore Ts'o <tytso@mit.edu> - - * problem.c: Remove period from the Pass 1C header, to be - consistent with the other Pass headers. - - * pass2.c (e2fsck_process_bad_inode): Fixed bug which could cause - e2fsck to core dump if a disconnected inode contained an - extended attribute. This was actually caused by two bugs. - The first bug is that if the inode has been fully fixed - up, the code will attempt to remove the inode from the - inode_bad_map without checking to see if this bitmap is - present. Since it is cleared at the end of pass 2, if - e2fsck_process_bad_inode is called in pass 4 (as it is for - disconnected inodes), this would result in a core dump. - This bug was mostly hidden by a second bug, which caused - e2fsck_process_bad_inode() to consider all inodes without - an extended attribute to be not fixed. (Addresses Debian - Bug: #316736) - -2006-06-30 Theodore Ts'o <tytso@mit.edu> - - * Release of E2fsprogs 1.38 - -2005-06-20 Theodore Ts'o <tytso@mit.edu> - - * unix.c (usage, parse_extended_opts): Clean up messages printed - by e2fsck for grammar and consistency. - -2005-06-19 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.8.in: Make it clear that the -c, -l, and -L options are - never safe to use on a mounted filesystem, and in general, - running e2fsck on mounted filesystems is a bad idea. - -2005-06-16 Theodore Ts'o <tytso@mit.edu> - - * message.c: Add @m (multiply-claimed) and @n (invalid) - expansions. - - * problem.c: Clean up the problem messages displayed by e2fsck, - mostly based on suggestions from Benno Schulenberg. - -2005-06-05 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.8.in: Add a request to the REPORTING BUGS section of the - man page to make sure that e2fsck's messages are printed - in English, and to send a raw e2image filesystem dump if - possible. - -2005-05-05 Theodore Ts'o <tytso@mit.edu> - - * badblocks.c (check_bb_inode_blocks): Clean up warning printf. - Thanks to Benno Schulenberg for the patch. (Addresses - Sourceforge Bug: #1189803) - -2005-04-16 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.8.in: Fix spelling mistakes in man pages. (Addresses - Debian Bugs: #304593) - -2005-04-14 Theodore Ts'o <tytso@mit.edu> - - * pass1b.c (pass1d): Don't offer to clone or delete the resize - inode; it's not allowed! - - * e2fsck.c, e2fsck.h, journal.c, pass1.c, pass1b.c, pass2.c, - pass3.c, pass4.c, super.c, unix.c: Use a centrally stored - current time for "now" which can be overridden using the - E2FSCK_TIME environment variable, for better - reproducibility for regression tests. - -2005-04-06 Theodore Ts'o <tytso@mit.edu> - - * pass1.c (e2fsck_pass1): If the superblock last mount time (not - just the last write time) looks insane, then assume that - we can't do the LOW_DTIME checks. - -2005-03-21 Theodore Ts'o <tytso@mit.edu> - - * Release of E2fsprogs 1.37 - -2005-03-21 Theodore Ts'o <tytso@mit.edu> - - * message.c, pass1.c, problem.c, problem.h, util.c: Integrate code - from Alex Thomas at Clusterfs to check extended attributes - stored in inodes. - - * message.c (expand_inode_expression): Add support for %IS in - problem descriptions. Note that this only works if the - problem context has a pointer to a large inode! - - * problem.h, problem.c (PR_1_EXTRA_ISIZE, PR_1_ATTR_NAME_LEN, - PR_1_ATTR_VALUE_OFFSET, PR_1_ATTR_VALUE_BLOCK, - PR_1_ATTR_VALUE_SIZE, PR_1_ATTR_HASH): Add new problem - codes. - - * util.c (e2fsck_write_inode_full): New function. - - * pass1.c (check_ea_in_inode, check_inode_extra_space): New - function which tests the validity of extended attributes - stored in a large inode. - (e2fsck_pass1): Call ext2fs_get_next_inode_full() instead - of get_next_inode(), and use an allocated inode which is - big enough to store extra portion of large inodes. - -2005-03-20 Theodore Ts'o <tytso@mit.edu> - - * super.c (check_super_block): Add sanity checks for the - superblock's inode_size field. - - * pass3.c (check_root, e2fsck_get_lost_and_found): Call - ext2fs_write_new_inode() instead of ext2fs_write_inode(). - - * pass1.c (check_blocks): Move counting the extended attribute - block earlier so that we don't have to worry about - num_blocks wrapping for files which are too big. - -2006-02-05 Theodore Ts'o <tytso@mit.edu> - - * Release of E2fsprogs 1.36 - -2005-02-04 Theodore Ts'o <tytso@mit.edu> - - * pass2.c (e2fsck_pass2), problem.c (PR_2_HTREE_FCLR), problem.h - (PR_2_HTREE_FCLR): Remove support for - --enable-clear-htree; this was only needed during the - early development of the htree patch. - -2005-01-28 Theodore Ts'o <tytso@mit.edu> - - * pass2.c (e2fsck_process_bad_inode): Fix a corner case involving - big-endian systems, long symlinks and i_file_acl set when - it shouldn't be. Without this bugfix, f_clear_xattr will - fail on big-endian machines. - - * super.c (check_resize_inode): Deal with the case where the - resize inode can't be read; don't try to recreate the - resize inode unless the resize feature is actually - enabled. - -2005-01-27 Theodore Ts'o <tytso@mit.edu> - - * super.c (check_resize_inode): If the resize inode has a zero - i_links_count or isn't a regular file, consider the resize - inode as invalid, and offer to recreate it. - - * pass2.c (e2fsck_process_bad_inode): Offer to clear i_file_acl - before checking to see if an invalid inode should be - removed, since otherwise the fast symlink detection code - can get confused. Also clear the inode's entry in - inode_bad_map if the inode has been completely fixed. - (Addresses Red Hat Bugzilla #146284) - -2005-01-25 Theodore Ts'o <tytso@mit.edu> - - * unix.c (main, check_if_skip): Set the valid flag earlier, and if - it is cleared by the superblock tests, then assume that - the filesystem contains errors and must be checked. - - * super.c (check_super_block): Check the individual block group - inode and block free counts, as well as the filesystem - inode and block free counts. If any of the block/inode - free counts is too large, force a full filesystem check. - (Addresses Debian Bug: #291571) - -2005-01-19 Theodore Ts'o <tytso@mit.edu> - - * unix.c (e2fsck_simple_progress): Use fixed integer math - to check to see if we're at 0 or 100% - (is_on_batt): Clean up gcc -Wall warning - (main): Remove dead code - -2005-01-13 Matthias Andree <matthias.andree@gmx.de> - - * unix.c: Do not call closedir if the handle is zero, to avoid - crashes when the /proc/acpi/ac_adapter directory is - missing, for instance on non-ACPI or non-Linux systems. - -2005-01-09 Theodore Ts'o <tytso@mit.edu> - - * pass1b.c: As part of converting configure.in from using the - "broken by design" 2.13 version of AC_CHECK_TYPE to - AC_CHECK_TYPES, if intptr_t isn't defined, manually - typedef it to long. (Addresses Debian Bug #289133) - -2005-01-06 Theodore Ts'o <tytso@mit.edu> - - * super.c (check_resize_inode): Use ext2fs_{read,write}_ind_block - so that byte swapping is handled on big-endian systems. - (check_super_block): Allow the number of reserved blocks - to be 50% of the number of blocks available, in order to - be consistent to what is allowed by tune2fs. - -2004-12-24 Theodore Ts'o <tytso@mit.edu> - - * pass1.c (e2fsck_pass1): At the end of the pass 1 processing, if - we have been signalled to do so, recreate the resize inode. - - * super.c (check_resize_inode): New function which checks to make - sure the resize inode is valid. It is called by - check_super_block(). If it is invalid, it will signal to - pass1.c that the resize inode needs to recreate. - - * e2fsck.h (E2F_FLAG_RESIZE_INODE): New flag - - * problem.c, problem.h (PR_0_RESIZE_INODE_INVALID, - PR_1_RESIZE_INODE_CREATE): Add new problem codes. - -2004-12-23 Theodore Ts'o <tytso@mit.edu> - - * swapfs.c (swap_inodes): Since swap_inodes bypasses the inode - cache for speed reasons, we must flush it to avoid cache - coherency problems. - -2004-12-16 Theodore Ts'o <tytso@mit.edu> - - * super.c (check_super_block): If the resize_inode feature is not - set, check to make sure that s_reserved_gdt_blocks is - zero, and that the resize inode is clear. - - * problem.h (PR_0_NONZERO_RESERVED_GDT_BLOCKS, - PR_0_CLEAR_RESIZE_INODE): Add new problem codes. - -2004-12-15 Theodore Ts'o <tytso@mit.edu> - - * pass1.c (process_block): Applied resize inode patch. Mark the - DIND block of the resize inode as being in use. - -2004-12-14 Theodore Ts'o <tytso@mit.edu> - - * Makefile.in: Move strip command to install-strip target. - Use Linux-kernel-style makefile output for "make install" - - * Makefile.in (installdirs): Use $(MKINSTALLDIRS) macro. - Remove sync in the "all" target. - -2004-11-30 Theodore Ts'o <tytso@mit.edu> - - * message.c (expand_inode_expression): If the TZ environment - variable is set to GMT, use gmtime() instead of - localtime() or ctime() to force the use of GMT. This is - because the dietlibc doesn't honor the TZ environment - variable. - - * e2fsck.h: Add io_options to e2fsck_struct - - * unix.c: If there is a question mark in the device name, separate - out the options to the IO layer, and pass it on to - ext2fs_open2(). - - * Makefile.in: Use Linux-kernel-style makefile output to make it - easier to see errors/warnings. - -2004-10-04 Richard Mortimer <richm@oldelvet.org.uk> - - * journal.c (e2fsck_check_ext3_journal): Attempt recovery of the - filesystem if the journal contains illegal block numbers. - -2004-07-26 Theodore Ts'o <tytso@mit.edu> - - * pass1.c (process_block): Change the limit of directory size from - 32 MB to 2GB. - -2004-05-11 Theodore Ts'o <tytso@mit.edu> - - * journal.c (sync_blockdev): Flush I/O caches to preserve ordering - constraints required by the journal recovery code. Thanks - to Junfeng Yang from the Stanford Metacompilation group - for pointing this out. - - * jfs_user.h: Define sync_blockdev() as an emulated function, - instead just being a no-op. - -2004-05-04 Theodore Ts'o <tytso@mit.edu> - - * unix.c (check_if_skip): If the checkinterval is zero, then - disregard it when calculating when the next check will - take place. - -2004-04-12 Theodore Ts'o <tytso@mit.edu> - - * unix.c (is_on_batt): Be more flexible about the name of the ACPI - device that corresponds to the AC adapter. (Addresses - Debian bug #242136) - -2004-04-03 Theodore Ts'o <tytso@mit.edu> - - * Makefile.in: Update the modtime even if subst doesn't need to - update the e2fsck man page, to avoid always re-running - subst, especially since there are no dependencies on the - man page. - -2004-02-28 Theodore Ts'o <tytso@mit.edu> - - * Release of E2fsprogs 1.35 - -2004-02-24 Theodore Ts'o <tytso@mit.edu> - - * unix.c (main, PRS), e2fsck.8.in: Add a new -k option which keeps - the existing badblocks list when using the -c option. - (Addresses Debian bug #229103) - -2004-02-23 Theodore Ts'o <tytso@mit.edu> - - * Makefile.in (PROFILED_LIBS, PROFILED_DEPLIBS): Fix reference to - libblkid so that the build will work with --enable-profile - (Addresses Sourceforge bug #811408) - -2004-02-14 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.c (e2fsck_run): Clear the SETJMP_OK flag when returning - so we don't double longjump into an invalid stack frame. - (Thanks to Matthias Andree for providing this fix.) - -2004-01-31 Theodore Ts'o <tytso@mit.edu> - - * problem.c (fix_problem): Don't call print_e2fsck_message if the - message is empty; otherwise, the NLS substitution will - print the .po header, which is Just Wrong. - -2004-01-30 Theodore Ts'o <tytso@mit.edu> - - * pass2.c (deallocate_inode_block): Check to make sure the block - number is invalid before deallocating it, to avoid core - dumping e2fsck. - -2003-12-12 Theodore Ts'o <tytso@mit.edu> - - * pass3.c (check_directory): When reconnecting a directory, we may - need to create a lost+found directory. This may - invalidate our pointer to the directory information, so we - must look it up again after calling - e2fsck_reconnect_file(). (Addresses Debian bug #219640). - -2003-12-10 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.h (E2F_FLAG_RESTARTED): Assign a unique flag bitfield to - E2F_FLAG_RESTARTED. This fixes a bug where if the user - specifies an alternate superblock, and the journal needs - to be replayed, e2fsck would erroneously assume that - journal had been run already without clearing the - NEEDS_RECOVERY flag, and bomb out with an error. - -2003-12-07 Theodore Ts'o <tytso@mit.edu> - - * badblocks.c, dict.c, ea_refcount.c, ehandler.c, journal.c, - pass1.c, pass1b.c, pass2.c, pass3.c, pass5.c, problem.c, - rehash.c, super.c, swapfs.c, unix.c, util.c, e2fsck.h: Fix - gcc -Wall nitpicks. - - * recovery.c, jfs_user.h: Sync recovery.c with latest 2.5 kernel - version. - -2003-12-02 Theodore Ts'o <tytso@mit.edu> - - * unix.c (main): When testing a disk using e2fsck -c, use the list - of new bad blocks to replace the current list of bad - blocks. This way "e2fsck -c" can be used to recover from - a corrupted bad block inode. - - * badblocks.c (test_disk): Remove the test_disk() file; e2fsck - will just call read_bad_blocks_file() directly with the - file parameter set to NULL. - -2003-09-14 Theodore Ts'o <tytso@mit.edu> - - * unix.c (is_on_batt, check_if_skip): If running on battery, then - give an extra grace period before actually forcing a - check. That way the laptop will be biased to waiting - until we are on AC power before doing the filesystem - check. (Addresses Debian bug #205177) - -2003-09-13 Theodore Ts'o <tytso@mit.edu> - - * unix.c (check_if_skip): If the number of mounts until the next - forced filesystem check is 5 or less, mention this to the - user. (Addresses Debian bug #157194) - - * pass1.c (e2fsck_pass1), problem.h (PR_1_BB_FS_BLOCK), - problem.c (PR_1_BB_FS_BLOCK, PR_1_BBINODE_BAD_METABLOCK_PROMPT): - Fix up the handling of corrupted indirect blocks in the - bad block. We now correctly handle the case where there - is an overlap between a block group descriptor or - a superblock and a bad block indirect block. In the case - where the indirect block is corrupted, we now suggest - "e2fsck -c". - -2003-09-12 Theodore Ts'o <tytso@mit.edu> - - * unix.c (PRS): Check the returned name from blkid_get_devname and - print an error if the requested LABEL/UUID does not exist. - -2003-09-03 Theodore Ts'o <tytso@mit.edu> - - * pass1.c (mark_table_blocks): Use the new function - ext2fs_reserve_super_and_bgd to calculate the blocks to be - reserved. - -2003-08-24 Theodore Ts'o <tytso@mit.edu> - - * util.c (get_backup_sb): Check to make sure the context is passed - in non-NULL before trying to dereference it. Otherwise we - might core dump when called to expand %S in a problem - description. - - * e2fsck.8.in: Adjust description line so that apropos - "ext2" or "ext3" will find the man page. (Addresses - Debian Bug #206845) - -2003-08-21 Theodore Ts'o <tytso@mit.edu> - - * journal.c (e2fsck_move_ext3_journal): When moving the journal or - backing up the journal inode, make sure it gets written to - all the superblocks. - -2003-08-20 Theodore Ts'o <tytso@mit.edu> - - * journal.c (e2fsck_move_ext3_journal): If the superblock's backup - journal information is not set, copy the journal inode - information to the superblock. - (e2fsck_get_journal): If the journal inode is corrupted, - synthesize one from the backup information in the - superblock. - - * problem.c, problem.h (PR_0_BACKUP_JNL): Add new problem code - which is used when we make a backup of the journal inode. - -2003-08-19 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.h, unix.c (PRS, e2fsck_clear_progbar, - e2fsck_simple_progress): Don't print the ^A and ^B - characters which bracket the progress bar when the e2fsck - program is talking directly to a tty, but only when it is - being piped to another program. (Addresses Debian bug - #204137) - * unix.c: Move some initialized variables to the BSS segment to - shrink the size of the e2fsck executable. - -2003-08-01 Philipp Thomas <pthomas@suse.de> - - * Include inttypes.h if present. New macros INT_TO_VOIDPTR and - VOIDPTR_TO_INT. Use them throughout for int<->void * - casts to fix 64-bit warnings. - - * dirinfo.c, dx_dirinfo.c, e2fsck.c, ea_refcount.c, journal.c, - message.c, pass1.c, pass1b.c, pass2.c, pass3.c, pass4.c, - pass5.c, super.c, swapfs.c: ext2fs_getmem(), - ext2fs_free_mem(), and ext2fs_resize_mem() all now take a - 'void *' instead of a 'void **' in order to avoid pointer - aliasing problems with GCC 3.x. - -2003-07-25 Theodore Ts'o <tytso@mit.edu> - - * Release of E2fsprogs 1.34 - -2003-07-06 Theodore Ts'o <tytso@mit.edu> - - * pass2.c: Fix gcc -Wall nitpicks (missing #include <string.h>) - - * dict.c, dict.h: Fix gcc -Wall nitpicks (indented cpp directives) - -2003-06-8 Theodore Ts'o <tytso@mit.edu> - - * problem.c: Fix spelling error. - -2003-05-17 Theodore Ts'o <tytso@mit.edu> - - * unix.c (show_stats), e2fsck.h: Use ngettext (abbreivated with - the macro P_(str1, str2, n)) to simplify the statistics - reporting. - -2003-05-05 Theodore Ts'o <tytso@mit.edu> - - * unix.c (main): If --enable-testio-debug is set, then use the - test_io manager so we can watch what e2fsck is doing. - -2003-05-03 Theodore Ts'o <tytso@mit.edu> - - * unix.c (show_stats): Remove uneeded use of _() around string - under #ifndef NLS - - * problem.c: Mark two strings (PR_1_RELOC_TO and - PR_2_NON_UNIQUE_FILE) as not being printf format strings. - Fix typo (doubled 'is') in PR_2_FILE_ACL_BAD. - - * Makefile.in: Add $(LIBINTL) to the link line so we can support - using the internal gettext library. - - * e2fsck.h: Add #ifndef protection around LOCALEDIR definition. - -2003-04-27 Theodore Ts'o <tytso@mit.edu> - - * message.c (expand_at_expression): Fixed NLS support for message - abbreviations. - -2003-04-21 Theodore Ts'o <tytso@mit.edu> - - * Release of E2fsprogs 1.33 - -2003-04-19 Theodore Ts'o <tytso@mit.edu> - - * rehash.c: Fix lint warnings by including string.h and ctype.h - - * pass2.c (dict_de_cmp): Fix lint warnings by using const pointers. - - * unix.c (e2fsck_simple_progress), e2fsck.h: Fix lint warnings by - using const pointers. Remove unused variables. - -2003-04-17 Theodore Ts'o <tytso@mit.edu> - - * rehash.c (name_cmp): Sort the deleted inodes to the end of the - list, for portability to systems that whose qsort does not - perform a stable sort. - -2003-04-16 Theodore Ts'o <tytso@mit.edu> - - * unix.c: Bracket progress bar output with control-A and control-B - characters. These characters are used as in-band - signalling to allow a logging program to filter out the - progress bar. - (PRS): Use setvbuf instead of setbuf to force stdout and - stderr to be non-buffered when the stdout/stderr are a pipe. - This fixes a cosmetic problem when using e2fsck under - logsave. - -2003-04-12 Theodore Ts'o <tytso@mit.edu> - - * unix.c: Add #ifdef around #include <sys/ioctl.h> - -2003-04-05 Theodore Ts'o <tytso@mit.edu> - - * super.c (check_super_block): Update the global free block and - inode counters from the block group specific counters - quietly. This is needed for an experimental patch which - eliminates locking the entire filesystem when allocating - blocks or inodes; if the filesystem is not unmounted - cleanly, the global counts may not be accurate. - -2003-03-17 Theodore Ts'o <tytso@mit.edu> - - * util.c: Explicitly declare e2fsck_global_ctx as extern for the - benefit of the Apple Darwin port. - -2003-03-15 Theodore Ts'o <tytso@mit.edu> - - * rehash.c (e2fsck_rehash_dir): If user specified the -n option, - skip trying to write out directory. - (e2fsck_rehash_directories): Fix the percentage - calculation in the progress bar. - -2003-03-14 Theodore Ts'o <tytso@mit.edu> - - * problem.c, problem.h (PR_2_REPORT_DUP_DIRENT): Add new problem - code. - - * pass2.c (check_dir_block): Check for duplicate filenames within - a single directory block (because this is the easy case; - we don't currently check for duplicates that span - directory blocks, for now. Eventually for htree - directories we can do this by searching for all directory - blocks that have a hash overflow, and then searching the - adjacent blocks to find all other potential duplicates.) - - * iscan.c, scantest.c, unix.c: And #ifdef protection around - #include of malloc.h - - * rehash.c (duplicate_search_and_fix): Now search for duplicates - filenames, and either prompt to remove a complete - duplicate entry, or to rename a duplicate filename. - (e2fsck_rehash_dir): Use a progress bar to report - progress, and don't print all of the directory inodes as - they are optimized. - - * problem.c, problem.h (PR_2_DUPLICATE_DIRENT, - PR_2_NON_UNIQUE_FILE): New problem codes. - - * unix.c (e2fsck_simple_progress), e2fsck.h: New function which - can be called to provide specialized progress bars that - are not related to the top-level pass-based completion - percentage. - - * pass3.c (e2fsck_adjust_inode_count), e2fsck.h: Export previously - static function. - -2003-03-06 <tytso@mit.edu> - - * e2fsck.8.in: Fix minor nit in the -C option. (Addresses Debian - bug #173612) - -2003-03-01 Theodore Ts'o <tytso@mit.edu> - - * Makefile.in, journal.c, unix.c: Use blkid functions to find the - journal from the UUID, and to interpret the device - specification. - - * e2fsck.c: Free the blkid_cache when releasing the e2fsck context - structure. - - * e2fsck.h: If strnlen is not present, define it as a macro which - calls e2fsck_strlen(). Add prototype for string_copy(). - Add blkid_cache to e2fsck context, and make - filesystem_name, device_name, and journal_name be - non-const variables. - - * pass1.c, pass2.c: Remove static strnlen function - - * util.c (string_copy, e2fsck_strnlen): New functions - -2003-01-29 Theodore Ts'o <tytso@mit.edu> - - * unix.c (usage): Make descripton -c be a bit more explicit - -2003-01-22 Theodore Ts'o <tytso@mit.edu> - - * pass1.c (check_blocks): Use the EXT2_I_SIZE macro. - -2003-01-02 Theodore Ts'o <tytso@mit.edu> - - * unix.c (main): Detect if there is an attempt to run the journal - twice, and abort with an error if this is the case. - (Address IBM Bugzilla bug #1226) - -2002-12-18 Theodore Ts'o <tytso@mit.edu> - - * pass2.c (strnlen): Provide strnlen if libc doesn't. - -2003-11-19 Theodore Ts'o <tytso@mit.edu> - - * unix.c (PRS): Print an error if more than one of the -p/-a, -n - or -y options are specified. - -2002-11-12 Theodore Ts'o <tytso@mit.edu> - - * problem.c: Make HTREE problems PR_PREEN_OK, so that we don't - abort an e2fsck after the filesystem has been mounted - using the 2.4 ext2 codebase. - -2002-11-09 Theodore Ts'o <tytso@mit.edu> - - * Release of E2fsprogs 1.32 - -2002-11-08 Theodore Ts'o <tytso@mit.edu> - - * Release of E2fsprogs 1.31 - -2002-11-08 <tytso@snap.thunk.org> - - * unix.c (main): Fix a bug where e2fsck could erroneously mark a - filesystem as being valid if it was being checked and it - is interrupted with a ^C. (Bug introduced in e2fsprogs - 1.28.) - - * unix.c (PRS), journal.c: Set the default level of journal - debugging to be 0, and allow the user to set the level of - journal debugging by using the E2FSCK_JBD_DEBUG - environment variable. - - * pass1.c (new_table_block), super.c (check_super_block), swapfs.c - (swap_filesys): Clear EXT2_FLAG_MASTER_SB_ONLY to make - sure the backup superblocks and group descriptors are - updated when we make an important change to the - superblocks and/or group descriptors. (i.e., generating a - UUID, allocating filesystem metadata, or byte-swapping the - filesystem.) - -2002-11-07 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.h, unix.c (main, check_mount): Fix e2fsck so that it - returns the appropriate exit code, so that the boot - scripts will be told that they need to automatically - reboot. - -2002-10-31 Theodore Ts'o <tytso@mit.edu> - - * Release of E2fsprogs 1.30 - -2002-10-31 Theodore Ts'o <tytso@mit.edu> - - * unix.c (usage), e2fsck.8.in: Add the -D option to the usage and - command synopsis. (Addresses Debian bug #167108) - - * pass1.c (e2fsck_pass1): Check the superblock write time to see - whether or not we can safely do the LOW_DTIME checks. - (Addresses Sourceforge bug #620980) - -2002-10-30 Theodore Ts'o <tytso@mit.edu> - - * Makefile.in (install): Search all compression extensions when - deleting old man pages. - - * pass1.c (mark_table_blocks): Fix bug in meta_bg support; only - mark blocks legacy group descriptor blocks up to - s_first_meta_bg. - -2002-10-20 Theodore Ts'o <tytso@valinux.com> - - * pass1.c (mark_table_blocks): Add support for the meta_blockgroup - filesystem format. - -2002-10-13 Theodore Ts'o <tytso@mit.edu> - - * pass3.c (e2fsck_get_lost_and_found): Pass in mode 700 to - ext2fs_new_inode(). - - * unix.c (show_stats): Fix format bug if NLS is in use. - - * journal.c, pass1.c, revoke.c, util.c: Fix gcc -Wall nits. - -2002-10-08 Theodore Ts'o <tytso@mit.edu> - - * pass3.c (e2fsck_get_lost_and_found): Create the lost+found - directory with mode 0700. - -2002-10-02 Theodore Y. Ts'o <tytso@mit.edu> - - * pass2.c (parse_int_node, check_dir_block): Add byte-swap - functions around the htree code, so that it works on - big-endian machines. - - * swapfs.c (swap_filesys): For PPC machines, byte-swap the bitmap, - since PPC big-endian filesystems were historically wierd. - This is just for the regression test, since no one - actually uses them anymore... - -2002-09-30 Theodore Ts'o <tytso@mit.edu> - - * problem.c, problem.h (PR_2_HTREE_BAD_LIMIT, - PR_2_HTREE_BAD_COUNT, PR_2_HTREE_HASH_ORDER, - PR_2_HTREE_BAD_DEPTH): Add new problem codes. - - * pass2.c (salvage_directory, check_dir_block): Avoid counting - directory entries twice during a directory salvage. Make - the salvaging algorithm smarter about directory entry size - overruns. - (parse_int_node): Add check to make sure the limit and - count fields in the node headers are sane. Also check - to make sure the hash table in ascending order. - (pass2, htree_depth): For all leaf nodes, determine their - depth in the tree, and make sure it matches with the depth - in the htree root information. - -2002-09-29 Theodore Ts'o <tytso@mit.edu> - - * pass2.c (check_dir_block): Do a more paranoid check when trying - to determine whether or not a directory entry is a - completely empty leaf block or leaf node. Otherwise - e2fsck might get confused into thinking that a valid dxdir - was corrupted. - - * rehash.c (e2fsck_rehash_dir): Apply patch from Christopher Li - which avoids creating an empty directory entry at the end - of the directory block. This screws up earlier versions - of the indexed directory patch. - -2002-09-28 Theodore Ts'o <tytso@mit.edu> - - * rehash.c (write_directory): Clear the index flag if by - reoptimizing the directory, we bring it back into a - non-indexed state. - (e2fsck_rehash_dir): Allow directories that contain two - blocks to be indexed. Otherwise when they grow, they - never will be indexed by the kernel. - - * unix.c (main): Only update the master superblock; there's no - point updating the backup superblocks, and it speeds up - fsck slightly. - - * pass2.c (salvage_directory): New function called by - check_dir_block() which is much more sophisticated about - how it salvages corrupted filesystems. - -2001-09-24 Theodore Tso <tytso@mit.edu> - - * Release of E2fsprogs 1.29 - -2002-09-06 Theodore Ts'o <tytso@mit.edu> - - * rehash.c (copy_dir_entries): Fix bug which caused corrupted - directories if there are 4 or 8 bytes left over in a - directory page when optimizing/reindexing a directory. - -2001-08-31 Theodore Tso <tytso@thunk.org> - - * Release of E2fsprogs 1.28 - -2002-08-31 Theodore Ts'o <tytso@mit.edu> - - * ehandler.c (e2fsck_handle_read_error): If the user asks to - ignore the error, offer to try forcing a write of the - block to the disk; sometimes this will cause the drive to - remap the bad block. - -2002-08-31 Theodore Ts'o <tytso@valinux.com> - - * pass2.c (e2fsck_pass2): If this is a HTREE directory, sort the - dblist so that the first block of all of the directories - is handled first so we can read the hash version - information. - (check_dir_block): Examine the root node for correctness, - and offer to clear it if it is not correct. Also copy the - hash version to the dx_dir structure, so that the proper - hash function can be used for other blocks in the - directory. - - * problem.c, problem.h (PR_2_HTREE_BAD_ROOT): Add new problem code. - -2002-08-21 Theodore Ts'o <tytso@mit.edu> - - * problem.c: Fix PR_1_RELOC_BLOCK_ALLOCATE message to explain that - it is necessary to find contiguous blocks in a particular - block group. - -2002-08-17 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.8.in: Add clarifying text describing how e2fsck works - with ext3 filesytems. (Addresses Debian bug #145044). - - * pass1.c (handle_htree): Add support for EXT2_HASH_TEA - - * e2fsck.h, e2fsck.c (e2fsck_allocate_context): Add new field to - the e2fsck context, ext_attr_ver, which specifies the EA - format version. - - * unix.c (usage, parse_extended_opts, PRS), e2fsck.8.in: Add new - option -E, which allows the users to specify extended - options. Added ea_ver extended option. - - * pass1.c (e2fsck_pass1_check_device_inode): Add ext2_filsys - argument to this function, so we can account for the - presence of extended attribute blocks attached to device - inodes. - (e2fsck_pass1_check_symlink, e2fsck_pass1): Take into - account the fact that symlinks can also have extended - attribute blocks. - (check_ext_attr): Don't be flexible about the EA format - version. Check against the version number in - ctx->ext_attr_ver. - (check_blocks): Check all inodes, so that we account for - extended attribute blocks belonging to special files. - Clean up i_size checks. - - * pass1b.c (pass1b): Check all inodes, so that we account for - extended attribute blocks belonging to special files. - (delete_file_block): Use ext2fs_alloc_block_stats() to - update the filesystem statistics. - (delete_file): Attempt to decrement the extended - attribute refcount, and free the EA block if the count - hits zero. - (clone_file): Fixed bugs in EA handling. Don't call - block_iterate on inodes that don't have a valid i_block[] - array. Reread the base inode since it may have been - changed by ext2fs_block_iterate. When updating inodes as - part of cloning an EA block, write out the correct inode - structure. - - * pass2.c (deallocate_inode_block, deallocate_inode): Use standard - ext2fs_alloc_*_stats functions to update the filesystem - statistics. - (deallocate_inode): Attempt to decrement the extended - attribute refcount, and free the EA block if the count - hits zero. - (e2fsck_process_bad_inode): Add extra argument to calls - to e2fsck_pass1_check_device_inode (). - - * pass3.c (e2fsck_get_lost_and_found): Use standard - ext2fs_alloc_*_stats functions to update the filesystem. - statistics when creating /lost+found. - (adjust_inode_count): Remove debugging code that can never - be triggered. - - * pass4.c (disconnect_inode): Add explanation about why we only - clear inodes that have no data blocks and no EA blocks. - Use ext2fs_inode_alloc_stats2 function to update the - filesystem statistics when clearing a zero-length inode. - - * problem.c, problem.h (PR_1B_ADJ_EA_REFCOUNT, - PR_2_ADJ_EA_REFCOUNT): Add new problem codes. - - * super.c (release_inode_block), (release_orphan_inodes): Use the - standard ext2fs_alloc_*_stats functions to update the - filesystem statistics. - (release_inode_blocks): Attempt to decrement the extended - attribute refcount, and free the EA block if the count - hits zero. - -2002-08-01 Theodore Ts'o <tytso@mit.edu> - - * dict.c, dict.h: New file from kazlib 1.20 which implements a - red-black tree - - * pass1b.c: Massive changes to take advantage of dict.c. This - removes several O(n**2) algorithms from the rare case - where there are a large number of blocks claimed by - multiple inodes. - -2002-07-24 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.8.in, e2fsck.h, pass3.c (e2fsck_pass3), unix.c, rehash.c: - Add new option -D, which requests e2fsck to optimize all - directories. Rehash.c will also compress and sort - non-indexed directories. - - * problem.c, problem.h: Rename PR_3A_REHASH_* to PR_3A_OPTIMIZE_* - - * unix.c (PRS): Make sure the signal_cancel is registered without - SA_RESTART. - - * rehash.c (e2fsck_rehash_dir, copy_dir_entries, calculate_tree): - Restructured code in e2fsck_rehash_dir into two new - subroutines to make the code more understandable/maintainable. - (set_root_node): Fixed bug which caused the root node to - be invalid on non-FILETYPE filesystems due to an - unitialized variable. - (calculate_tree): Fix bug where pointers which might get - invalidated if realloc() moves outdir->buf around. - (e2fsck_rehash_dir): Fix memory leak bug. - - * pass3.c (e2fsck_get_lost_and_found), e2fsck.h, e2fsck.c: - Exported as a public interface a function for finding the - lost and found directory. Cache the location of the lost - and found directory in the e2fsck context structure. - - * util.c (ask_yn, read_a_char): Note when the user has typed ^C, - and abort processing by longjmp'ing to ctx->abort_loc. - -2002-07-23 Theodore Ts'o <tytso@mit.edu> - - * pass1.c (e2fsck_pass1): If e2fsck is run with -n, don't create - the dirs_to_hash list, since we don't want to reindex - directories if the filesystem is opened read-only. - -2002-07-21 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.8.in: Document new exit code FSCK_CANCELED - - * unix.c (PRS, signal_cancel): Capture SIGINT and SIGTERM signals - and set a flag in the e2fsck context indicating that - cancellation has been requested, so that e2fsck will exit - only at safe points. - (main): Change the exit handling so that if a cancellation - is requested, return FSCK_CANCELED (a new exit code 32). - e2fsck can now return more than one exit code as part of a - bitmask (as had been documented in the man page). - - * pass2.c (e2fsck_pass2, check_dir_block), pass3.c (e2fsck_pass3), - pass4.c (e2fsck_pass4): Check to see if a cancellation was - requested, and abort processing if necessary. - -2002-07-19 Theodore Ts'o <tytso@mit.edu> - - * rehash.c, Makefile.in: New file which rewrites directories using - the htree format. - - * problem.c (fix_problem), problemP.h (PR_PREEN_NOHDR): Add option - which suppresses the header printed when in preen mode. - - * pass3.c (e2fsck_pass3): If there are entries on the dirs_to_hash - list, call e2fsck_rehash_directories to reindex those - directories. - (e2fsck_expand_directory): Generalize the old - expand_dirctory() function so it can expand a directory to - a guaranteed minimum size. - - * e2fsck.h (struct e2fsck_struct): Add the dirs_to_hash list. Add - new function prototypes for rehash.c and for - e2fsck_expand_directory(). - - * e2fsck.c (e2fsck_reset_context): Free the dirs_to_hash list. - - * pass1.c (e2fsck_pass1): Initialize the dirs_to_hash list if the - htree feature is present in the filesystem. - (check_blocks): If a non-htree directory has more than 2 - blocks, put it on the dirs_to_hash list. - - * pass2.c (clear_htree): Add corrupt htree directories to the - dirs_to_hash list. - - * problem.h, problem.c (PR_3A_PASS_HEADER, PR_3A_REHASH_ITER, - PR_3A_REHASH_DIR_ERR, PR_3A_REHASH_DIR_HEADER, - PR_3A_REHASH_DIR, PR_3A_REHASH_DIR_END): Add new problem codes - - * pass2.c (parse_int_node), problem.c (PR_2_HTREE_BADBLK): Fix - problem display. - -2002-07-15 Theodore Ts'o <tytso@mit.edu> - - * pass2.c (e2fsck_pass2): Use dx_dir->numblocks instead of - dx_dir->ino to indicate that a bad inode was cleared. - -2002-06-26 Theodore Ts'o <tytso@mit.edu> - - * pass1.c (check_blocks): Move htree handling to handle_htree(). - Factor out calls to ext2fs_write_inode so that it happens - if dirty_inode is non-zero. - (handle_htree): Add checks for invalid htree root, invalid - hash version, invalid hash version, and htree depth too deep. - - * problem.h, problem.c (PR_1_HTREE_NODIR, PR_1_HTREE_BADROOT, - PR_1_HTREE_HASHV, PR_1_HTREE_INCOMPAT, PR_1_HTREE_DEPTH): - Add new problem codes. - - * pass2.c (parse_int_node): Fix silly typo. - (check_dir_block): Change to use new ext2fs_dirhash() - function prototype. - (pass2): Fixed two minor bugs discovered by the test case: - Don't stop checking dxdir's after clearing a bad inode. - If there is a bad max hash, make sure the bad_dir flag - is set to make sure we try to clear inode. - -2002-06-25 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.c (e2fsck_reset_context): Free the dx_dirinfo structure. - - * message.c: Add new abbrevations @h and @p, "HTREE directory - inode" and "problem in". - - * pass1.c (check_blocks): If the inode has the INDEX_FL flag, - register the block into the indexed directory data - structures. Or if the filesystem doesn't have the - DIR_INDEX flag, offer to clear the INDEX_FL. - - * pass2.c (e2fsck_pass2, parse_int_node): Add support check htree - directories (we don't check all possible corruptions yet). - - * problem.h, problem.h (PR_1_HTREE_SET, PR_2_HTREE_NOTREF, - PR_2_HTREE_DUPREF, PR_2_HTREE_MIN_HASH, PR_2_HTREE_MAX_HASH, - PR_2_HTREE_CLEAR, PR_2_HTREE_FCLR, PR_2_HTREE_BADBLK): Add - new problem codes. - - * unix.c (main): If ENABLE_HTREE is not defined, complain if the - filesystem has the dir_index feature. - - * Makefile.in, e2fsck.h, dx_dirinfo.c: New file (and group of - functions) which keeps track of blocks in HTREE directory - blocks. - -2002-05-22 Andreas Dilger <adilger@clusterfs.com> - - * super.c (check_superblock): Check that the number of inodes and - blocks in a group is less than 2^16, so that the free inode - and block counts for a group fit into the group descriptor - table fields. Any more than that would need a COMPAT flag. - -2002-05-22 Theodore Ts'o <tytso@mit.edu> - - * pass1.c (check_ext_attr): Update to support the V2 Bestbits EA - format. We automatically determine from the first EA - block we visit whether we are using the V1 or V2 format. - -2002-05-21 Theodore Ts'o <tytso@mit.edu> - - * pass1.c (process_block): If an inode has too many blocks or - is too big, then offer to truncate the inode. - (check_blocks): Don't bother checking the size to see if - it's too big, since that's just a symptom, not the disease - (which we're now appropriately checking in process_block). - - * problem.c, problem.h: Add new problem codes PR_1_INODE_TOOBIG, - PR_1_TOOBIG_DIR, PR_1_TOOBIG_REG, PR_1_TOOBIG_SYMLINK, and - add the latch code PR_LATCH_TOOBIG. - -2002-05-20 Theodore Ts'o <tytso@mit.edu> - - * e2fsck.h, pass1.c (e2fsck_pass1_check_symlink), pass2.c - (e2fsck_process_bad_inode): Add an extra - argument so a scratch buffer can be passed into the - function. - - * pass1.c (e2fsck_pass1_check_device_inode, - e2fsck_pass1_check_symlink): If the EXT2_INDEX_FL flag is - set, then the device/FIFO is invalid. - (check_immutable): Don't check for EXT2_INDEX_FL; we take - care of that elsewhere now. - (e2fsck_pass1): Check to see if the symlink is valid - before we offer to clear any immutable bits. This matches - the order in which we check other special files. - - * pass2.c (e2fsck_pass2): Allocate a bigger scratch buffer so we - can pass part of it to e2fsck_process_bad_inode. - - * pass4.c (e2fsck_pass4): If we need to call - e2fsck_process_bad_inode(), allocate the necessary scratch - buffer. - - * problem.h, problem.c: Change PR_2_SYMLINK_SIZE to be - PR_2_INVALID_SYMLINK, and change the text - accordingly. Revert the text change for PR_1_SET_IMMUTABLE - since we no longer check for EXT2_INDEX_FL in - check_immutable(). - -2002-05-18 Andreas Dilger <adilger@clusterfs.com> - - * pass1.c (e2fsck_pass1_check_symlink): Remove checks of - immutable/append-only checks, since this is handled by - check_immutable. For non-fast symlinks, read in the - data block and check the length to make sure it - matches with the inode size. - (check_immutable): Generalize to also check for the - EXT2_INDEX_FL flag. - (check_blocks): For non-regular files, signal a bad inode - size if i_size_high is non-zero. - - * pass2.c: Only count large files if they are regular files - (not just if they aren't directories). - - * problem.c, problem.h: Change comment for PR_2_SYMLINK_SIZE - to indicate that it can be triggered by any symlink, - not fast symlink. Change text for PR_1_SET_IMMUTABLE - to indicate that it the EXT2_INDEX_FL was set. - -2002-05-17 Theodore Ts'o <tytso@mit.edu> - - * pass1.c (e2fsck_pass1): When calculating max_sizes, omit capping - this value by the calculated value of max_sect_sizes. - The inode may be a sparse inode, and the limit of - max_sect_sizes assumed a non-sparse inode and a limit - based on a 32-bit i_blocks field measured in 512 sectors. - - * super.c (check_super_block): Fix check for s_log_block_size so - that takes into account fact that there is an offset of - EXT2_MIN_BLOCK_LOG_SIZE. - - * pass1.c: Allocate max_sizes array to be big enough to support - the larger block sizes. - -2002-05-16 Andreas Dilger <adilger@clusterfs.com> - - * unix.c (main), util.c (get_backup_sb): Use EXT2_MAX_BLOCK_SIZE - instead of using a hardcoded value of 4096. - - * super.c (check_super_block): Change checks for s_log_block_size - s_log_frag_size so they support the larger block sizes. - - * pass1.c (e2fsck_pass1): Calculate max_sizes array for block - sizes up to EXT2_MAX_BLOCK_SIZE (to support block sizes - greater than 4k). - -2002-03-08 Theodore Tso <tytso@mit.edu> - - * Release of E2fsprogs 1.27 - -2002-03-07 Theodore Tso <tytso@mit.edu> - - * journal.c, pass5.c, revoke.c: Fix gcc -Wall complaints - -2002-03-07 Theodore Tso <tytso@mit.edu> - - * unix.c (main): Don't close the filesystem before calling - fatal_error(); this just causes a needless warning about a - bad I/O manager magic number. - - * journal.c (e2fsck_check_ext3_journal): Offer to clear the inode - in case the journal is too small, or the inode number in - the superblock doesn't appear to be random. - -2002-03-06 Theodore Tso <tytso@mit.edu> - - * problem.h, problem.c: PR_5_BLOCK_RANGE_UNUSED, - PR_5_BLOCK_RANGE_USED, PR_5_INODE_RANGE_UNUSED, - PR_5_INODE_RANGE_USED: Add new problem codes. - - * pass5.c (check_block_bitmaps, check_inode_bitmaps): Collapse - reporting of adjacent blocks and inodes into a range. - -2002-03-04 Philipp Thomas <pthomas@suse.de> - - * unix.c (main): Newer libintl needs LC_CTYPE to be set in - addition to LC_MESSAGES. - -2002-02-24 Theodore Tso <tytso@mit.edu> - - * Makefile.in (install): Install hard links for man pages for - fsck.ext2 and fsck.ext3. Remove any compressed man - pages before installing the man pages. - -2002-02-22 Theodore Tso <tytso@mit.edu> - - * journal.c: Improve code maintainability and reduce code size by - moving common code paths in e2fsck_journal_init_dev() and - e2fsck_journal_init_inode() into e2fsck_get_journal(). - Also fixed a memory leak in recover_ext3_journal(). - - * super.c (release_orphan_inodes): Fix memory leak. - -2002-02-03 Theodore Tso <tytso@thunk.org> - - * Release of E2fsprogs 1.26 - -2001-12-24 Theodore Tso <tytso@mit.edu> - - * unix.c (PRS): Don't allow the -c and -l/-L options to be - specified at the same time. - - * e2fsck.h (E2F_OPT_WRITECHECK), unix.c (PRS), - badblocks.c (read_bad_blocks_file): If two -c options are - specified, then perform a non-destructive read/write scan - of the disk. - - * e2fsck.8.in: Document the double -c option; also encourage users - to use -c instead of the -l/-L options since it's too hard - for users to get things like the blocksize parameter correct. - -2001-12-23 Theodore Tso <tytso@mit.edu> - - * util.c (get_backup_sb): This function now searches for the - backup superblock by iterating over possible blocksizes - instead of defaulting a guess of 8193 if the superblock - isn't available. - - * message.c (expand_percent_expression), unix.c (main): Pass in - new parameters to get_backup_sb. Also, in unix.c, use the - blocksize paramter in the e2fsck context structure instead - of using a static variable, since get_backup_sb wants to - be able to set the blocksize paramter. - - * e2fsck.h: Update function prototype for get_backup_sb; also add - the blocksize parameter to the e2fsck context structure. - - * Makefile.in, jfs_user.h: Move linux/jbd.h to - ext2fs/kernel-jbd.h, to avoid using the system header - file version of hbd.h when using diet glibc (since it - forcibly adds /usr/include to the beginning of the - include search path.) - -2001-12-21 Theodore Tso <tytso@mit.edu> - - * problem.c (fix_problem): Use ctx->filesystem_name if the - ctx->device_name is NULL. - - * journal.c (e2fsck_check_ext3_journal): Give the opportunity to - run the journal even if recovery flag is cleared. If - we're using a backup superblock, run the journal by - default. - - * e2fsck.h (E2F_OPT_FORCE), unix.c (PRS, check_if_skip): Use a - bitfield in the e2fsck context flags word to indicate - whether or not a check should be forced. This allows the - journal code to set the option if necessary to force a - filesystem check. - - * problem.h, problem.c: Remove PR_0_JOURNAL_RESET_JOURNAL, and add - PR_0_JOURNAL_RUN and PR_0_JOURNAL_RUN_DEFAULT. Update - problem decription texts. - -2001-12-16 Theodore Tso <tytso@mit.edu> - - * e2fsck.h (ext2fs_get_refcount_size), unix.c (check_mount, PRS), - pass1.c (adjust_extattr_refcount), - ea_refcount.c (ea_refcount_create): Fix gcc -Wall nits. - - * recovery.c, revoke.c: Update to versions from 2.4.17-pre8. - - * journal.c, jfs_user.h: Update support code for new version of - recover.c and revoke.c. Fix support for filesystems with - external journals. - -2001-11-30 Gabriel Paubert <paubert@iram.es> - - * journal.c (e2fsck_journal_load): Fix an endianness bug. - -2001-11-26 Theodore Tso <tytso@mit.edu> - - * super.c (check_super_block): Make sure that if the inode table - or allocation bitmap is zero, that it is marked as - invalid, so that in pass #1, a new bitmap/inode table gets - properly allocated. (Addresses Debian Bug #116975) - -2001-11-24 Theodore Tso <tytso@mit.edu> - - * e2fsck.8.in: Fix minor typo in man page and clarify device - specification. - -2001-10-07 Theodore Tso <tytso@mit.edu> - - * journal.c (clear_v2_journal_fields, e2fsck_journal_load): If the - V2 fields are set on a V1 journal superblock, or an - internal V2 journal has s_nr_users is non-zero, clear the - entire journal superblock beyond the V1 superblock. This - fixes botched V1->V2 updates. - - * problem.c, problem.h (PR_0_CLEAR_V2_JOURNAL): Add new problem code. - -2001-09-20 Theodore Tso <tytso@mit.edu> - - * e2fsck.h, journal.c (e2fsck_move_ext3_journal): Add new function - which will automatically relocate the ext3 journal from a - visible file to an invisible journal file if the - filesystem has been opened read/write. - - * super.c (check_super_block): Add call to e2fsck_move_ext3_journal - - * problem.c, problem.h (PR_0_MOVE_JOURNAL, PR_0_ERR_MOVE_JOURNAL): - Add new problem codes. - -2001-09-20 Theodore Tso <tytso@thunk.org> - - * Release of E2fsprogs 1.25 - -2001-09-19 Theodore Tso <tytso@mit.edu> - - * unix.c (main): If a superblock is specified explicitly by the - user, don't try to automatically fall back to an - alternate superblock. - -2001-09-02 Theodore Tso <tytso@thunk.org> - - * Release of E2fsprogs 1.24a - -2001-08-30 Theodore Tso <tytso@thunk.org> - - * Release of E2fsprogs 1.24 - -2001-08-30 Theodore Tso <tytso@valinux.com> - - * pass1.c (e2fsck_pass1): For EXT2_RESIZE_INO, allow its i_mode to - either be zero or a regular file (for compatibility with - Andreas's on-line resizing programs). - -2001-08-27 Theodore Tso <tytso@valinux.com> - - * unix.c (main): Remove EXT2FS_VERSION from the version display, - since it only confuses people. - - * pass1.c (strnlen): Provide strnlen if libc doesn't. - -2001-08-15 Theodore Tso <tytso@valinux.com> - - * Release of E2fsprogs 1.23 - -2001-08-13 Theodore Tso <tytso@valinux.com> - - * super.c (release_orphan_inodes): If the filesystem contains - errors, don't run the orphan * list, since the orphan list - can't be trusted. - - * pass1.c (check_size): Fix logic in check_size; the previous code - only offered to clear the inode size fields if both size - and i_size_high were zero. - (e2fsck_pass1_check_device_inode): If i_blocks is - non-zero, then assume that the device/socket/fifo inode - is bogus. - -2001-08-09 Theodore Tso <tytso@valinux.com> - - * pass1.c, pass2.c, problem.c, problem.h: Fix bug introduced by - Andreas's symlink code; check_blocks() was unconditionally - testing inode_bad_map without checking to see if it - existed first. Fixed problem a different way; we now no - longer check inode_bad_map at all, since the file might - not get deleted in pass 2 anyway. We move the large file - feature reconciliation code to to e2fsck_pass2(), and in - deallocate_inode() in pass2.c, we decrement the large - files counter if we're about to delete a large file. - - * unix.c (show_stats): Print the number of large files in verbose - mode. - -2001-08-07 Theodore Tso <tytso@valinux.com> - - * journal.c (recover_ext3_journal): If s_errno is set in the - journal superblock, set the EXT2_ERROR_FS flag in the - filesystem superblock after the journal is run. - -2001-08-04 Andreas Dilger <root@lynx.adilger.int> - - * message.c: Change comments for %D and %d expansion in e2fsck - problem codes. It was not consistent which was for dirent - expansion, and which was for directory number expansion. - - * problem.c (PR_2_FINAL_RECLEN, PR_2_BAD_FILETYPE): Fix problem - codes which got caught by the confusion between %D and %d. - -2001-08-04 Theodore Tso <tytso@valinux.com> - - * problem.c (PR_2_SYMLINK_SIZE): Change description to make it - more clear (and remove %s expansion). Also add missing - periods to the end of a number of problem descriptions. - - * pass2.c (e2fsck_process_bad_inode): Remove unneeded problem - context string set now that the problem description for - PR_2_SYMLINK_SIZE has been changed. - - * pass1.c (e2fsck_pass1_check_symlink): Consolidate some checks, - and check the validity of the symlink block here (so that - we detect this case here instead of later). Also use - sizeof(inode->i_block) instead EXT2_LINK_DIR. - -2001-07-30 Theodore Tso <tytso@valinux.com> - - * unix.c (check_mount): Remove the code which tested for the root - filesystem being mounted read-only, and depend on the - results flags from ext2fs_check_if_mounted. - -2001-07-29 Theodore Tso <tytso@valinux.com> - - * unix.c (check_if_skip): Free the e2fsck context structure on a - normal clean filesystem exit, to make it easier to find - real memory leaks. - (PRS): Only update the path to include /sbin at the - beginning if the -c option is given, again to make it - easier to find memory leaks. - (main): Move the final print_resource_track call after the - filesystem and the context are freed. - - * journal.c (e2fsck_journal_init_dev): Avoid memory leak if we - need to search for the journal device. - (e2fsck_journal_release): Free the journal IO channel when - we release the journal handle, to avoid a memory leak. - - * e2fsck.c (e2fsck_reset_context): Fix bug; only close the io - channel if it is *different* from the filesystem io - channel. - -2001-07-27 Theodore Tso <tytso@valinux.com> - - * problem.c (PR_1_SET_IMMUTABLE): Clarify problem message. - - * pass1.c (e2fsck_pass1): Check for symlinks that have the - immutable flag set (and offer to clear them). - -2001-07-26 Theodore Tso <tytso@valinux.com> - - * pass1.c (e2fsck_pass1): Free ctx->block_ea_map at the end of - pass 1. - -2001-07-25 Theodore Tso <tytso@valinux.com> - - * pass1.c (check_ext_attr): Skip zero-length EA entries. - - * problem.c: PR_1_EA_ALLOC_COLLISION shouldn't abort, but should - prompt to clear the EA block. - -2001-07-22 Theodore Tso <tytso@valinux.com> - - * journal.c (ll_rw_block): Use ctx->journal_io instead of the - filesystem's io_channel. - (e2fsck_journal_init_dev): New function which supports - initialization of the external journal. - (e2fsck_get_journal): Remove code which flagged an error - if the superblock reported the use of an external journal. - (ext3_journal_via_mount): Remove unsued, #ifdefed out function. - - * problem.c, problem.h: Removed error codes no longer used - (PR_0_JOURNAL_UNSUPP_DEV, PR_0_JOURNAL_BAD_DEV, - PR_0_JOURNAL_UNSUPP_UUID) and replace them with new error - codes related with failures in loading the external - journal (PR_0_JOURNAL_UNSUPP_MULTIFS, - PR_0_CANT_FIND_JOURNAL, PR_0_EXT_JOURNAL_BAD_SUPER). - Also changed the text assocated with PR_0_JOURNAL_BAD_UUID - to reflect the case where the external journal isn't - correct for this filesystem. - - * unix.c (PRS), e2fsck.8.in: Add new option -j which allows - the user to specify the pathname to find the external journal. - - * e2fsck.c (e2fsck_reset_context): Close journal_io if it isn't - the same as the filesystem io_channel. - - * e2fsck.h: Add new fields (journal_io and journal_name) in the - context structure to support external journals. - -2001-07-20 Theodore Tso <tytso@valinux.com> - - * unix.c (main): Add an explicit warning when the filesystem is - left not completely fixed when e2fsck exits. (Addresses - Debian bug #104502.) - -2001-07-19 Theodore Tso <tytso@valinux.com> - - * pass1.c (check_ext_attr): The entire EA block is now checked to - make sure that parts of the EA block aren't being used for - multiple purposes. - - * Makefile.in e2fsck.h, region.c: New file which is used to detect - collisions in extended attribute block. - - * problem.h, problem.c (PR_1_EA_MULTI_BLOCK, PR_1_EA_ALLOC_REGION, - PR_1_EA_ALLOC_COLLISION, PR_1_EA_BAD_NAME, - PR_1_EA_BAD_VALUE): Add new problem codes. - -2001-07-10 Theodore Tso <tytso@valinux.com> - - * journal.c (e2fsck_run_ext3_journal): Only call ext3_flush() if - the superblock is dirty. - -2001-07-07 Theodore Tso <tytso@valinux.com> - - * pass1b.c (pass1b, process_pass1b_block): Change the num_bad - field calculation so that it only counts EA block entries - as a single multiply claimed block (since once we clone - the EA blocks for one inode, we fix the problem for all of - the other inodes). Also, I moved the num_bad calculation - from process_pass1b_block to the end of pass1b. This - fixes a *significant* performance bug in pass1b which hit - people who had to had a lot of multiply claimed blocks. - (Can you say O(n**3) boys and girls? I knew you could... - Fortunately, this case didn't happen that much in actual - practice.) - - * pass1.c (e2fsck_pass1): Defer inodes which have an extended - attribute block for later processing to avoid extra seeks - across the disk. - (process_inode_cmp): If there is no indirect block, sort - by the extended attribute (i_file_acl) block. - - * pass1b.c (clone_file_block): Fix bugs when cloning extended - attribute blocks. Moved free of block_buf to after the - code which clones the extattr block, and fixed logic for - changing pointers to the extended attribute field in the - inodes which were affected. - (decrement_badcount): New function which is used whenever - we need to decrement the number of files which claim a - particular bad block. Fixed bug where delete_file wasn't - checking check_if_fs_block() before clearing the entry in - block_dup_map. This could cause a block which was claimed - by multiple files as well as the filesystem metadata to - not be completely fixed. - - * pass1.c (adjust_extattr_refcount): Add new function which - adjusts the reference counts of extended attribute blocks - if needed, both up and down. - (e2fsck_pass1): If the refcount or refcount_extra - structure are present, call adjust_extattr_refcount(), - and free it afterwards. - - * problem.h, problem.c (PR_1_EXTATTR_READ_ABORT, - PR_1_EXTATTR_REFCOUNT, PR_1_EXTATTR_WRITE): Add new - problem codes. - -2001-07-02 Theodore Tso <tytso@valinux.com> - - * pass1.c (e2fsck_pass1, check_ext_attr, check_blocks): Add - initial support for scanning extended attribute blocks. - - * e2fsck.c (e2fsck_reset_context): free extended attribute - refcount structure. - - * e2fsck.h: Add new fields for accounting for the extended - attribute blocks. - - * Makefile.in, ea_refcount.c, e2fsck.h: Add new file which - implements a reference count abstraction. - - * problem.c, problem.h: Add new problem codes PR_1_ALLOCATE_REFCOUNT, - PR_1_READ_EA_BLOCK, PR_1_BAD_EA_BLOCK, PR_2_FILE_ACL_BAD - - * message.c: Add "@a" as an abbreviation for "extended attribute" - - * pass1b.c (clone_file): Add code which handles cloning an - extended attribute block. - - * pass1b.c (delete_file): - * pass2.c (deallocate_inode): If the inode contains an extended - attribute block in the file_acl field, clear it and - deallocate the block. - -2001-06-28 Theodore Tso <tytso@valinux.com> - - * pass2.c (e2fsck_process_bad_inode): Deal with inodes which are - marked bad because of an invalid file_acl (either because - EXT2_FEATURE_COMPAT_EXT_ATTR is not set, or because the - block number in file_acl is bad. - -2001-06-29 Theodore Tso <tytso@valinux.com> - - * unix.c (show_stats): Use long long to calculate the percentage - of the number of blocks in use in the filesystem. - -2001-06-23 Theodore Tso <tytso@valinux.com> - - * Release of E2fsprogs 1.22 - -2001-06-22 Theodore Tso <tytso@valinux.com> - - * flushb.c: Use platform independent method of defining the - BLKFLSBUF ioctl. Also include sys/mount.h since on newer - platforms BLKFLSBUF is defined there. - -2001-06-19 Theodore Tso <tytso@valinux.com> - - * super.c (release_inode_blocks): Don't try to release the blocks - if the orphaned inode is a device file, symlink, or some - other kind of special file that doesn't have a block list. - -2001-06-15 Theodore Tso <tytso@valinux.com> - - * Release of E2fsprogs 1.21 - -2001-06-13 Theodore Tso <tytso@valinux.com> - - * unix.c (check_if_skip): Adapted patch from Andreas Dilger which - prints the number of mounts or days elapsed since the last - check when e2fsck is forced to check an otherwise clean - filesystem. - -2001-06-12 Theodore Tso <tytso@valinux.com> - - * badblocks.c: Stop using the compatibility badblocks function, - and use the ext2fs_badblocks_* functions instead. - -2001-06-11 Theodore Tso <tytso@valinux.com> - - * unix.c (PRS): Fix bug introduced in 1.20 which broke the -F - flag. - - * Makefile.in: Add message.c and swapfs.c to the list of source - files to build the make depend. - - * swapfs.c, unix.c: Only support the -s and -S options to e2fsck - if ENABLE_SWAPFS is defined. - -2001-06-08 Theodore Tso <tytso@valinux.com> - - * recover.c, revoke.c: Synchronize with ext3 0.7a - -2001-06-02 Theodore Tso <tytso@valinux.com> - - * recovery.c (scan_revoke_records): Fix bug in recovery code; - missing byte order conversion. - - * pass1.c (mark_inode_bad): Replace alloc_bad_map with a function - which sets the bit in the bad inode bitmap. - (e2fsck_pass1): Check for fast symlinks with an invalid - size, and set the bad inode map in that case. - (check_size): Check i_size_high for special files and - prompt to clear it along with i_size if non-zero. - - * pass2.c (e2fsck_process_bad_inode): Check for fast symlinks with - an invalid size and prompt the user if the inode should be - cleared. - - * problem.h, problem.c (PR_2_SYMLINK_SIZE): Added new problem code. - -2001-06-01 Theodore Tso <tytso@valinux.com> - - * problem.c, problem.h: Change PR_0_JOURNAL_UNSUPP_INCOMPAT and - PR_0_JOURNAL_UNSUPP_ROCOMPAT so they aren't fatal errors, - but prompt to see if the user should abort. Removed the - PR_0_JOURNAL_RESET_COMPAT problem code. - - * journal.c (e2fsck_journal_load): If there are incompatible - journal flags, just return an error code. - (e2fsck_check_ext3_journal): If e2fsck_journal_load - returns an error code indicating that there are - incompatible journal flag, check to see if we should - abort, and then offer to clear the journal. - (Addresses Debian bug #98527.) - - * Makefile.in: Move include/asm/types.h.in to - lib/ext2fs/ext2_types.h.in. - -2001-06-01 Theodore Tso <tytso@valinux.com> - - * pass1.c (pass1_get_blocks, pass1_read_inode, pass1_write_inode, - pass1_check_directory): Add a safety check to make sure - ctx->stashed_inode is non-zero. - - * pass1b.c (pass1b): Use e2fsck_use_inode_shortcuts() to disable - the inode shortcut processing, instead of manually - clearing only half of the function pointers that needed to - be NULL'ed out. This caused nasty bugs if the last inode - in the filesystem needed dup block processing. - - * pass1b.c (clone_file_block): When cloning a directory's metadata - block, don't try to update the directory block list - database, since indirect blocks aren't stored in the - database and the resulting error will abort the file clone - operation. - -2001-05-25 Theodore Tso <tytso@valinux.com> - - * Release of E2fsprogs 1.20 - -2001-05-25 Theodore Tso <tytso@valinux.com> - - * journal.c (e2fsck_journal_reset_super): Remove extraneous line - -2001-05-23 Theodore Tso <tytso@valinux.com> - - * journal.c (e2fsck_journal_reset_super): Fix bug; the reset - journal wasn't getting written out to disk since the dirty - bit wasn't being set on the buffer. - (e2fsck_journal_load): Don't print an error message if the - journal version number is wrong; just return a error code - reflecting this fact. If the block type in the journal - superblcok is obviously not a version number, report the - journal is corrupted. - (e2fsck_check_ext3_journal): On an unsupported journal - version, prompt to abort by default, but then offer a - chance to clear the journal as corrupt. - - * problem.c, problem.h (PR_0_JOURNAL_UNSUPP_VERSION): Added new - problem code. - -2001-05-14 Theodore Tso <tytso@valinux.com> - - * pass1.c: Treat inodes with a low dtime (that were from a - corrupted orphan list) specially. - - * problem.c, problem.h: Add new problem codes PR_1_LOW_DTIME and - PR_1_ORPHAN_LIST_REFUGEES, and a new latch group, - PR_LATCH_LOW_DTIME. - - * problemP.h: Expand the size of the problem flag to be an int - instead of a short. Expand space in the flag word which - is reserved for problem latch flags from 3 bits to 8 bits. - - * e2fsck.h, scantest.c: Change location of ext2_fs.h to be - ext2fs/ext2_fs.h - - * super.c (check_super_block): Be more strict on checking - s_r_blocks_count superblock field. - -2001-05-13 Theodore Tso <tytso@valinux.com> - - * problem.c, problem.h (PR_0_JOURNAL_UNSUPP_ROCOMPAT, - PR_0_JOURNAL_UNSUPP_INCOMPAT, PR_0_JOURNAL_RESET_COMPAT): - New problem codes. - - * journal.c (e2fsck_journal_load): Use a problem code to - report unsupported feature flags. There is code to - clear unsupported flags, but since this is dangerous, - it's not allowed in the problem code table. - -2001-05-11 Andreas Dilger <adilger@turbolinux.com> - - * journal.c (e2fsck_journal_reset_super): initialize the journal - sequence number to a random value to avoid recovering - bad transactions from a corrupt journal. - -2001-05-13 Theodore Tso <tytso@valinux.com> - - * journal.c: Code cleanup; initialize journal_enable_debug using - an initializer. - -2001-05-12 Theodore Tso <tytso@valinux.com> - - * unix.c (PRS): Skip validation of -C's file descriptor if it is - zero, since that is special case. - -2001-05-09 Theodore Tso <tytso@valinux.com> - - * super.c (release_orphan_inodes): Add gettext quoting around - "Truncating" and "Clearing" for i18n. - -2001-05-05 Theodore Tso <tytso@valinux.com> - - * util.c (fatal_error): Use the correct magic number when checking - the magic number for the io_channel data structure. Also - remove extraneous call to io_channel_flush() that was left - over from an editing session. - - * pass2.c (check_dir_block): Ignore EXT2_ET_DIR_CORRUPTED errors - from ext2fs_read_dir_block(). - -2001-05-01 Theodore Tso <tytso@valinux.com> - - * unix.c (PRS): Validate the completion information file - descriptor so that the completion bar won't get - accidentally written onto the disk. - - * e2fsck.8.in: Add explanation that you can use mke2fs -n -b - blocksize to printout alternate superblock locations. - -2001-04-26 Theodore Tso <tytso@valinux.com> - - * unix.c (check_if_skip): If the max_mount_count is zero, treat it - as having no count set. - -2001-04-16 Theodore Tso <tytso@valinux.com> - - * super.c (check_super_block): Fix bad calculation of - inodes_per_block, and tighten min/max checks to be a bit - more paranoid. Thanks to Andreas Dilger for pointing out - this bug. - -2001-03-29 Theodore Tso <tytso@valinux.com> - - * journal.c (mark_buffer_uptodate): Add emulation of kernel - function mark_buffer_uptodate. - - * recovery.c, revoke.c: Synchronize with 0.6b ext3 files. - -2001-02-12 Theodore Tso <tytso@valinux.com> - - * journal.c (e2fsck_run_ext3_journal): Force a flush of the - filesystem and io_channel before replaying the journal. - -2001-01-17 Theodore Ts'o <tytso@valinux.com> - - * pass1.c (check_blocks): Fix large file checking code so that - files > 2GB are considered large files, and restrict - directories from being > 2GB. - -2001-01-12 Theodore Ts'o <tytso@valinux.com> - - * journal.c (e2fsck_journal_fix_unsupported_super): Remove unused - function. Add FIXME notes to e2fsck_get_journal(), from - Andreas Dilger. - - * pass1.c (e2fsck_pass1): Cap the maximum legal size of a file by - the limit caused by the fact that i_blocks is in 512 byte - units, and that the Linux buffer cache also fundamentally - assumes 512 byte sectors. - Make sure that the journal inode is a regular file, and - when clearing an unused journal inode, make sure the - icount db is updated. - - * problem.c, problem.h (PR_1_JOURNAL_BAD_MODE): Add new problem code. - - * problem.c: For PR_1_RESERVED_BAD_MODE, print a description of - the reserved inode. In PR_0_JOURNAL_HAS_JOURNAL, prompt - to clear the journal, rather than deleting it (which is - more accurate). (From Andreas Dilger.) - - * journal.c: Make sure all functions which return an error code - use the errcode_t return type, and not "int" - (e2fsck_journal_release): Add new parameter, "drop", - which is used when we just want to deallocate the journal - without trying to write out any changes. - (mark_buffer_clean): New function - (e2fsck_check_ext3_journal): If we fail loading the - journal, make sure we free all memory associated with it. - (recover_ext3_journal): If we fail to load the journal or - initialize the revoke data structures, make sure we free all - memory associated with the journal. - - * message.c (special_inode_name): Add more special inode names - (From Andreas Dilger) - - * util.c (fatal_error): Check to make sure the io_manager data - structure is sane before trying to flush the io_channel. - (From Andreas Dilger) - - * mtrace.h, super.c, pass2.c: Minor whitespace cleanups, from - Andreas Dilger. - - * journal.c (e2fsck_journal_fix_bad_inode): Set pctx->ino to the - bad journal number. - - * problem.c (PR_0_JOURNAL_BAD_INODE): Use pctx->ino instead of - pctx->num when printing the bad journal inode number. - -2001-01-11 <tytso@snap.thunk.org> - - * pass1.c (process_block): Remove dead code which was never - getting executed. - - * iscan.c, unix.c, e2fsck.h: Don't use NOARGS, and use - ext2fs_sync_device() instead of using BLKFLSBUF. - - * flushb.c: Don't use NOARGS anymore; everything is STDC these days. - - * dirinfo.c, e2fsck.h, emptydir.c, iscan.c, jfs_user.h, journal.c, - message.c, pass1.c, pass1b.c, pass2.c, pass3.c, pass4.c, - pass5.c, problem.h, scantest.c, super.c, swapfs.c: Change - ino_t to ext2_ino_t. - -2001-01-09 <tytso@snap.thunk.org> - - * problem.c: Fix another broken @F vs @f problem. - -2001-01-06 <tytso@snap.thunk.org> - - * journal.c, pass1.c, pass1b.c, pass3.c, recovery.c, revoke.c, - super.c, unix.c, util.c: Fix random gcc -Wall complaints. - - * jfs_user.h: Use more sophisticated inline handling to allow - building with --enable-gcc-wall - -2001-01-03 <tytso@snap.thunk.org> - - * pass1.c (e2fsck_pass1): Moved journal inode handling out to its - own block; if the journal inode is not in use, and it - contains data, offer to clear it. - - * problem.h, problem.c (PR1_JOURNAL_INODE_NOT_CLEAR): Add new - problem code. - - * problem.c: Modified problem table to use a new abbreviations. - - * message.c: Add @j abbreviation for journal, and @v abbreviation - for device. - - * jfs_user.h: Moved contents of jfs_e2fsck.h into jfs_user.h. - - * journal.c (e2fsck_check_ext3_journal): Force a fsck if we remove - the journal stored on a reserved inode. Also force a fsck - if the journal appears to exist while !NEEDS_RECOVERY, and - we forcibly reset the journal. - -2001-01-01 <tytso@snap.thunk.org> - - * journal.c, pass1.c, super.c, unix.c: Replace use of struct - ext2fs_sb with struct ext2_super_block. - - * pass1.c (check_blocks): Remove use of EXT2_HAS_*_FEATURE macros. - -2000-12-31 <tytso@snap.thunk.org> - - * jfs_compat.h: Remove uneeded header file. - -2000-12-30 <tytso@snap.thunk.org> - - * malloc.h, mtrace.c: Renamed malloc.h to mtrace.h to avoid - conflicts with the system header file. - - * problem.h: Fixed numbering of pass1 error messages; an extra 0 - had slipped into some of the numbers. (Pointed out by - Andreas Dilger) - - * journal.c (e2fsck_journal_fix_corrupt_super): Clean up - unnecessary automatic variable. Add explanatory comment - about the kernel emulation routines. (Suggested by - Andreas Dilger) - -2000-12-13 Theodore Ts'o <tytso@valinux.com> - - * journal.c (e2fsck_check_ext3_journal): Check to make sure the - journal fields are consistent if any of the superblock - fields are set. (Backs out erroneous change made by sct, - pointed out by Andreas.) - - * unix.c (main): Clarify coments (no code changes) - - * super.c (release_orphan_inodes): Fix spelling typo in error message. - - * pass1.c (e2fsck_pass1): Offer to update the filesystem revision - level if we need to set large files flag. Patch from - Andreas Dilger. - - * super.c (check_super_block): If we have any of the compatibility - flags set, we need to have a revision 1 filesystem. Most - kernels will not check the flags on a rev 0 filesystem - and we may have corruption issues because of the - incompatible changes to the filesystem. Patch from Andreas - Dilger. - - * problem.c, problem.h (PR_0_FS_REV_LEVEL, PR_1_FS_REV_LEVEL): Add - new problem codes. - -2000-12-09 <tytso@snap.thunk.org> - - * flushb.c: Fix flushb so that it does something other than waste - disk space when built on systems with modern header files - and add a non-subtle Copyright Licensing restriction so - Yann will remove it from the Debian Distribution. (Now - violates the Debian Free Software Guidelines, on purpose.) - - * journal.c (e2fsck_check_ext3_journal): If JFS_DEBUG is defined - at the top level, set the JFS debuging level to 2. - - * jfs_e2fsck.h, jfs_user.h: Replaces jfs_compat.h. The jfs.h file - has been moved to the include/linux directory. - - * journal.c, revoke.c, recovery.c: Updated files from Stephen to - support the V2 superblock and revoke processing. The - journal.c and revoke.c files are copies from the ext3 - kernel source. - - * Makefile.in: Added revoke.c to the list of source/object files. - -2000-11-16 Theodore Ts'o <tytso@valinux.com> - - * pass1b.c: Change routines to use PR_1B_BLOCK_ITERATE when - reporting problems rather than using com_err directly. - - * problem.c, problem.h (PR_1B_BLOCK_ITERATE): Add new problem code. - - * message.c (expand_percent_expression): Add safety check. If - ctx->str is NULL, print "NULL" instead of dereferencing - the null pointer. - - * pass1b.c, pass2.c, pass3.c: Change calls to ext2fs_block_iterate - to ext2fs_block_iterate2, to support 64-bit filesizes and - to speed things up slightly by avoiding the use of the - ext2fs_block_iterate's compatibility shim layer. - -2000-10-30 <tytso@snap.thunk.org> - - * util.c (get_backup_sb): Calculate backup superblock correctly - when the blocksize is > 1k. - -2000-10-26 <tytso@snap.thunk.org> - - * jfs.h, jfs_compat.h, journal.c: Updated to include the - definition of the new journal superblock format; still - only supports V1 superblocks for now. (From sct) - -2000-10-24 <tytso@snap.thunk.org> - - * super.c (release_inode_block, release_inode_blocks, - release_orphan_inodes): Add code to deal with truncating - inodes which are still in use (but which are on the orphan - list because they need truncation). - - * problem.c, problem.h: Rename PR_0_CLEAR_ORPHAN_INODE to - PR_0_ORPHAN_CLEAR_INODE, and remove - PR_0_ORPHAN_INODE_INUSE. - - * journal.c (e2fsck_run_ext3_journal): Add i18n support, and print - a message when the journal is being recovered. - - * pass1.c (e2fsck_pass1): Don't check the i_mode field for the - journal inode, if it is in use. - -2000-09-12 <tytso@valinux.com> - - * extend.c: - * flushb.c: Add include of nls-enable.h which is necessary so that - they can compile correctly. - -2000-08-22 <tytso@valinux.com> - - * unix.c (main): If we're doing a read-only check, skip the - journal playback, but don't abort the e2fsck run. - - * super.c (release_orphan_inodes): Fix typo; should do bounds - checking on next_ino instead of ino. - - * jfs_compat.h (J_ASSERT): - * journal.c (e2fsck_run_ext3_journal): - * pass3.c (adjust_inode_count): Use fatal_error() instead of exit(). - - * unix.c: Use fatal_error() instead of exit() whenever possible. - Also fix the fsck exit codes so that we use FSCK_USAGE - when it is appropriate. Rename global_signal_ctx to - e2fsck_global_ctx and let it be exported globally. - - * util.c (fatal_error): Try to flush the I/O manager before - forcing an exit. - -2000-08-20 <tytso@valinux.com> - - * journal.c (e2fsck_journal_load): Fix **nasty** bug which caused - e2fsck_check_ext3_journal to smash the journal because - journal->j_transaction_sequence wasn't getting - initialized. - - * journal.c: (recover_ext3_journal, e2fsck_run_ext3_journal): Move - call to e2fsck_clear_recover from recover_ext3_journal to - after the filesystem has been closed and reopened. - Otherwise, the superblock in the filesystem handle will - probably be stale, and will overwrite the newer version of - the superblock written by the log recovery. - - * message.c (expand_inode_expression): Add support for %Iu and %Ig - - * problem.h (PR_0_CLEAR_ORPHAN_INODE): Add new problem code. - - * super.c (release_orphan_inodes, release_inode_block, - release_inode_blocks): Update the block group descriptor - counts when freeing the orphan inode. Use - PR_0_CLEAR_ORPHAN_INODE to report when we clear an orphan. - - * journal.c (e2fsck_run_ext3_journal): Fix a bug where we - attempted to reopen the filesystem using the device name - instead of the filesystem name. - -2000-08-18 <tytso@valinux.com> - - * Makefile.in: Update the make dependencies - - * problem.c, problem.h: Add the problem codes: - PR_0_ORPHAN_ILLEGAL_BLOCK_NUM, - PR_0_ORPHAN_ALREADY_CLEARED_BLOCK, - PR_0_ORPHAN_ILLEGAL_HEAD_INODE, - PR_0_ORPHAN_ILLEGAL_INODE, PR_0_ORPHAN_INODE_INUSE - - * super.c (release_inode_blocks, release_orphan_inodes, - check_super_block): Add support for clearing orphaned - inodes from the unmounted filesystem. - - * journal.c (e2fsck_recover_ext3_journal): Remove the last orphan - check; this is now handled in check_super_block --- - non-journaled filesystems can use the orphan list in the - future. Also, move the the re-opening of the filesystem - to e2fsck_run_ext3_journal(). - -2000-07-12 Andreas Dilger <adilger@turbolinux.com> - - * journal.c: implement loading of ext3 journal for recovery code - - * problem.c (fix_problem): return answer from PR_AFTER_CODE to caller. - Add journal problems. - - * recovery.c (journal_recover): user-space ext3 journal recovery code - - * unix.c (main) : check journal and do recovery in separate steps - -2000-08-07 <tytso@snap.thunk.org> - - * unix.c (calc_percent): Make sure that we don't take a floating - exception if the max value is zero. (should normally - never happen, but...) - -2000-07-13 <tytso@valinux.com> - - * Release of E2fsprogs 1.19 - -2000-07-06 Theodore Ts'o <tytso@valinux.com> - - * unix.c (check_if_skip): Modify algorithm for checking - s_max_mnt_count to match with the kernel. (If - s_max_mnt_count is negative, ignore the mnt_count check.) - - * unix.c (e2fsck_update_progress): Adjust the width of the - progress bar dynamically, based on the filesystem name - that we need to display. - - * unix.c (main): If the ext3 needs_recovery flag is set, call - e2fsck_run_ext3_journal() and then restart the e2fsck run. - - * journal.c (e2fsck_run_ext3_journal): New file which contains - logic to recover the ext3 journal. This version relies on - the kernel being able to mount the filesystem in order to - run the journal. - -2000-07-05 Theodore Ts'o <tytso@valinux.com> - - * unix.c (e2fsck_update_progress): Only save and check the last - percentage after multiplying it by 10, nor 1000, since we - only need to save values to a tenth of a percent (and the - percentage is already from 0 .. 100%, not 0 .. 1). - Also, re-arrange the logic so that we do the time - check only after doing the percentage check, and we - only advance the spinner if we're about to display it. - -2000-07-04 Theodore Ts'o <tytso@valinux.com> - - * pass1.c (e2fsck_pass1): Check to see if the ext3 s_last_orphan - field is set; if so, clear it, so that ext3 mounting code - won't get confused by the fact that we've cleared out the - orphaned inodes. - -2000-06-10 Theodore Ts'o <tytso@valinux.com> - - * pass5.c (check_block_bitmaps, check_inode_bitmaps): Add error - checking for a "should never happen case". - - * problem.c, problem.h (PR_5_COPY_IBITMAP_ERROR, - PR_5_COPY_BBITMAP_ERROR): Add new error codes. - -2000-05-27 Theodore Ts'o <tytso@valinux.com> - - * pass1.c (pass1, check_size): Apply patch from Chris Wedgewood - (cw@foof.org) which checks to see if special devices have - a non-zero size, and deals with it. - - * problem.c, problem.h (PR1_SET_NONZSIZE): Add new problem code. - -2000-05-18 Theodore Ts'o <tytso@valinux.com> - - * Makefile.in (install): Create a symbolic link for fsck.ext3 as - well. - -2000-05-08 Theodore Ts'o <tytso@valinux.com> - - * problem.c, problem.h (PR_0_HURD_CLEAR_FILETYPE): Add new problem - code. - - * super.c (check_super_block): If the OS type in the superblock is - the Hurd, check to see if the filetype feature is set, and - offer to clear it if so. This needs to be done since the - Hurd doesn't properly support the filetype feature. - (And since the hurd allows the transmogrification of files - to special files and vice versa --- for no good reason - that I can understand --- it can't support the filetype - feature for the forseeable future, either.) - -2000-04-03 Theodore Ts'o <tytso@valinux.com> - - * unix.c: For platforms that don't define optarg.h, manually - define optarg and optind. - -2000-03-20 Theodore Ts'o <tytso@valinux.com> - - * pass1.c (check_immutable, e2fsck_pass1_check_device_inode): - Check for the append-only as well as the immutable flag. - - * problem.c (PR_1_SET_IMMUTABLE): Adjust message to include - append-only flag. Fix comment for compression flag. - -2000-02-12 <tytso@snap.thunk.org> - - * unix.c (e2fsck_update_progress): Limit the number of updates to - the progress bars to 8 times a second. This allows a 9600 - baud console link to keep up. - -2000-02-11 <tytso@snap.thunk.org> - - * unix.c (main): If compression is enabled on the filesystem, - print a warning message (for now). - - * message.c: Add new compression shortcut: @c == compress - - * problem.c, problem.h (PR_1_COMPR_SET): Add new error code. - - * pass1.c (check_blocks): If the inode has EXT2_COMPRBLK_FL flag - set, check to see if the filesystem supports compression. - If it does pass this information down to process_block() - so it can treat the compressed block flag words - correctly. If not, offer to clear the flag, since it - shouldn't be set. - (process_block): If an inode has the compressed inode flag - set, allow EXT2FS_COMPRESSED_BLKADDR. - - * pass1b.c (process_pass1b_block, delete_file_block, - clone_file_block): - * pass2.c (deallocate_inode_block): Use HOLE_BLKADDR to check to - see if the block can be skipped. - -2000-02-08 <tytso@snap.thunk.org> - - * util.c: Make resource tracking message more concise. - - * e2fsck.h: - * pass1.c (mark_table_blocks, e2fsck_pass1): Remove - ctx->block_illegal_map, since it's not needed by pass1, - and pass1b has been modified to calculate it manually if - needed. This reduces the memory footprint needed by e2fsck. - - * pass1b.c (check_if_fs_block): New static function which returns - whether or not a block overlaps with filesystem metadata. - This replaces consulting the block_illegal_map bitmap. - - * Makefile.in: Call sync after finishing building all in this - directory. - - * unix.c (PRS): sync the filesystem before trying to use - BLKFLSBUF, to minimize the chance of causing dirty blocks - to get dropped. - - * e2fsck.h: Manually define BLKFLSBUF if not defined, and we're on - a Linux/i386 system. - - * pass3.c (check_directory): Only do the loop detection algorithm - if we've searched over 2048 parent directories and haven't - found the end yet. This means that in the common case, we - don't allocate or clear the inode_loop_detection bitmap, - which for large systems, merely clearing the bitmap for - each directory was turning out to be quite expensive. - Thanks to Jani Jaakkola (jjaakkol@cs.helsinki.fi) for - identifying this problem. - -2000-02-06 Theodore Ts'o <tytso@valinux.com> - - * badblocks.c, e2fsck.h, ehandler.c, emptydir.c, extend.c, - flushb.c, iscan.c, message.c, pass1.c, pass1b.c, pass3.c - pass4.c, pass5.c, problem.c, scantest.c, swapfs.c, - unix.c, util.c: Add Internationalization support as - suggested by Marco d'Itri <md@linux.it>. - -2000-02-02 Theodore Ts'o <tytso@valinux.com> - - * e2fsck.h, flushb.c, scantest.c: Remove uneeded include of - linux/fs.h - -2000-01-18 Theodore Ts'o <tytso@valinux.com> - - * Makefile.in: Since LIBUUID can sometimes include "-lsocket" - we need a separate DEPLIBUUID that can be used in - Makefile's dependency rules. - -1999-11-23 <tytso@valinux.com> - - * e2fsck.8.in: Update language about where to find a backup - superblock. - -1999-11-19 <tytso@valinux.com> - - * pass1.c (process_inodes): Add shortcut handling; if - process_inodes_count is zero, return right away, to avoid - calling qsort with a non-positive count. - - * message.c (safe_print): Fix to properly display ^A, ^B, etc. and - to print Delete as ^? - - * Makefile.in (distclean): Remove TAGS and Makefile.in.old from - the source directory. - -1999-11-10 <tytso@valinux.com> - - * Release of E2fsprogs 1.18 - -1999-11-10 <tytso@valinux.com> - - * problem.c (fix_problem): Support a new flag, PR_PREEN_NO which - means the answer is assumed to be no in preen mode. This - is now used in the PR_1_SET_IMMUTABLE code, so that in - preen mode we ignore these inodes and just print a warning - message. - -1999-11-09 <tytso@valinux.com> - - * pass1.c (e2fsck_pass1): If the filesystem does not support - imagic inodes, if an inode has the imagic flag set, offer - to clear the imagic flag. If a valid device/fifo/socket - has the immutable flag set, call the new helper function - check_immutable() to offerto clear the immutable flag. - - * pass2.c (check_filetype): Use the new ext2_file_type() helper - function instead of calculating the file_type information - manually. - - * pass3.c (e2fsck_reconnect_file): When adding a link to - lost+found, calculate the filetype information so that - ext2fs_link() can use the information if applicable. - (get_lost_and_found): Create the /lost+found directory - with the correct filetype information if applicable. - - * util.c (ext2_file_type), e2fsck.h: New function which returns - the directory entry file type information given the - inode's mode bits. - - * problem.c, problem.h: Added new problem codes PR_1_SET_IMAGIC - and PR_1_SET_IMMUTABLE. - -1999-11-07 <tytso@valinux.com> - - * pass4.c (e2fsck_pass4): Clear inode_imagic_map after freeing it, - to prevent it from getting freed twice. - -1999-11-06 <tytso@valinux.com> - - * unix.c (main): Close the filesystem before freeing the context, - so that in the event of a free()-related segmentation - violation, the filesystem is properly closed and written - out. - -1999-10-27 <tytso@valinux.com> - - * e2fsck.c (e2fsck_reset_context): When freeing - ctx->inode_reg_map, we weren't zero'ing - ctx->inode_reg_map, which could cause a segfault later on - in the e2fsck run. - -1999-10-26 <tytso@valinux.com> - - * problem.h (PR_2_SPLIT_DOT): Fix excess zero in problem code (now - matches the standard convention). - -1999-10-26 <tytso@valinux.com> - - * Release of E2fsprogs 1.17 - -1999-10-26 <tytso@valinux.com> - - * message.c (safe_print): Make safe_print take an char instead of - an unsigned char to fix gcc warnings. - -1999-10-25 <tytso@valinux.com> - - * util.c: For NT portability, don't redefine getchar(), since - stdio defines that. Instead we introduce a new - abstract macro read_a_char() which is #defined to the - right function as necessary. - - * problem.c, problem.h (PR_2_NULL_NAME): Add new problem code. - - * pass2.c (check_dir_block): Require that the length of the - directory entry be at least 12 bytes. Check to see if the - filename is zero-length, and flag that as an error. - -1999-10-22 <tytso@valinux.com> - - * Release of E2fsprogs 1.16 - -1999-10-22 <tytso@valinux.com> - - * pass2.c (check_filetype): If the filetype filesystem feature is - not set, and a directory entry has a dirent feature, offer - to clear it (since 2.0 kernels will do complain will - interpret it as a very large name length field). - - * problem.c (PR_2_CLEAR_FILETYPE): Add new problem code. - -1999-10-21 <tytso@valinux.com> - - * e2fsck.8.in: Update man page to use a more standard format (bold - option flags and italicized variables), as suggested by - Andreas Dilger (adilger@enel.ucalgary.ca) - - * pass4.c (e2fsck_pass4): If an inode is set in the - inode_imagic_map bitmap, don't check to see if it is - disconnected from the inode tree (because it almost - certainly will be). Free inode_imagic_map at the end of - pass 4. - - * pass2.c (check_dir_block, check_filetype): If the FILETYPE - feature is set, check the directory entry's filetype - information field, and fix/set it if necessary. - (e2fsck_pass2): Free the inode_reg_map bitmap at the end - of pass 2. - - * pass1.c (e2fsck_pass1, alloc_imagic_map): Allocate and fill in - information for inode_reg_map and inode_imagic_map, which - indicates which inodes are regular files and AFS inodes, - respectively. - Since only the master superblock is written during a - restart, force that superblock to be used after a restart; - otherwise changes to the block group descriptors end up - getting ignored. - - * problem.c, problemP.h: If e2fsck is run -n, make def_yn variable - be 0 for "no". Add support for a new flag, PR_NO_NOMSG, - which supresses the problem message if e2fsck is run with - the -n option. - - * problem.c, problem.h (PR_2_SET_FILETYPE, PR_2_BAD_FILETYPE): Add - new problem codes. - - * message.c (expand_dirent_expression): Add support for %dt which - prints the dirent type information. - - * e2fsck.c (e2fsck_reset_context): Free new bitmaps (inode_reg_map - and inode_imagic_map). - - * e2fsck.h (e2fsck_t): Add new inode_reg_map and inode_magic_map - to the context structure. - -1999-09-24 <tytso@valinux.com> - - * unix.c (PRS), util.c (ask_yn): Add #ifdef's to make - e2fsprogs easier to port to non-Unix platforms. - -1999-09-07 <tytso@valinux.com> - - * pass3.c (adjust_inode_count): Fix bug where we didn't keep the - internal and external inode counts in sync when we - decremented an inode whose link count was already zero. - Now we skip incrementing or decrementing both link counts - if we would cause an overflow condition. - (expand_dir, expand_dir_proc): Change where we update the - inode block count and size files so that the block count - field is updated correctly when we create an indirect block. - -1999-08-05 <tytso@valinux.com> - - * super.c (check_super_block): Check to see whether the - inodes_per_group value in the superblock is insanely too - high. - -1999-08-02 <tytso@valinux.com> - - * pass1b.c (clone_file_block): Don't clear the dup_map flag if - the block also shares data with the fs metadata when - the count drops to 1, since the block should still be - cloned, as fs metadata isn't included in the count. - -1999-07-18 Theodore Ts'o <tytso@valinux.com> - - * Release of E2fsprogs 1.15 - -1999-07-19 <tytso@valinux.com> - - * unix.c (usage): Add minimalist emergency help to the usage - message. - -1999-07-18 <tytso@valinux.com> - - * unix.c: Add support for calculating a progress bar if the -C0 - option is given. The function e2fsck_clear_progbar() - clears the progress bar and must be called before any - message is issued. SIGUSR1 will enable the progress bar, - and SIGUSR2 will disable the progress bar. This is used - by fsck to handle parallel filesystem checks. Also, set - the device_name from the filesystem label if it is - available. - - * e2fsck.h: Add new flags E2F_FLAG_PROG_BAR and - E2F_FLAG_PROG_SUPRESS. Add new field in the e2fsck - structure which contains the last tenth of a percent - printed for the user. - - * message.c (print_e2fsck_message): Add call to - e2fsck_clear_progbar(). - - * pass1.c (e2fsck_pass1): - * pass2.c (e2fsck_pass2): - * pass3.c (e2fsck_pass3): - * pass4.c (e2fsck_pass4): - * pass5.c (e2fsck_pass5): Add call to e2fsck_clear_progbar when - printing the resource tracking information. - - * pass5.c (check_block_bitmaps, check_inode_bitmaps): If there is - an error in the bitmaps, suppress printing the progress - bar using the suppression flag for the remainder of the - check, in order to clean up the display. - -1999-06-30 <tytso@valinux.com> - - * unix.c (check_mount): Clean up the abort message displayed when - the filesystem is mounted and either stdout or stdin isn't - a tty. - -1999-06-25 <tytso@valinux.com> - - * e2fsck.h: - * pass1.c (pass1_get_blocks, pass1_read_inode, pass1_write_inode, - pass1_check_directory, e2fsck_use_inode_shortcuts): Make - pass1_* be private static functions, and create new - function e2fsck_use_inode_shortcuts which sets and clears - the inode shortcut functions in the fs structure. - - * e2fsck.h: - * pass2.c (e2fsck_process_bad_inode): Make process_bad_inode() an - exported function. - - * pass4.c (e2fsck_pass4): Call e2fsck_process_bad_inode to check - if a disconnected inode has any problems before - connecting it to /lost+found. Bug and suggested fix by - Pavel Machek <pavel@bug.ucw.cz> - -1999-06-21 <tytso@valinux.com> - - * unix.c (main): Add missing space in the disk write-protected - message. - -1999-05-22 <tytso@rsts-11.mit.edu> - - * problem.c, problem.h (PR_0_INODE_COUNT_WRONG): Add new problem - code. - - * super.c (check_super_block): Add check to make sure the total - number of inodes is sane, since this can be calculated - from the number of groups times the number of inodes per - group. Offer to correct it if it is incorrect. - -1999-03-19 <tytso@rsts-11.mit.edu> - - * pass5.c (check_block_end): Fix fencepost condition where when - clearing the block padding we were missing the last position - in the bitmap. - -1999-05-17 <tytso@rsts-11.mit.edu> - - * unix.c (reserve_stdio_fds): Add safety check in case - reserve_stdio_fds couldn't open /dev/null. - -1999-03-14 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * util.c (print_resource_track): Use mallinfo if present to get - more accurate malloc statistics. - - * pass3.c (get_lost_and_found): Check to see if lost+found is a - plain file; if so, offer to unlink it. - - * problem.c, problem.h (PR_3_LPF_NOTDIR): Add new problem code. - -1999-03-09 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * problem.c: Fix problem message for PR_1_BAD_GROUP_DESCRIPTORS so - that the block group number is printed. Add new prompt, - PROMPT_UNLINK. - -1999-01-09 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Release of E2fsprogs 1.14 - -1999-01-09 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * message.c (safe_print): New function which prints strings, - converting non-printable characters using the '^' and - M-notation. This function is now used to print directory - name entries and pathnames. - -Mon Jan 4 02:28:59 1999 Theodore Y. Ts'o <tytso@mit.edu> - - * unix.c (main): Reset the context before calling ext2fs_close(), - to avoid referencing already freed memory. - -1998-12-15 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Release of E2fsprogs 1.13 - -1998-12-03 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Makefile.in: Updated dependencies. - -1998-11-27 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass3.c (get_lost_and_found): If the filesystem is not opened - read-only, then force /lost+found to be created if it is - not present. - - * problem.c: Allow PR_3_NO_LF_DIR to be handled during a preen - operation. - -1998-10-28 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * unix.c (main): Move ext2fs_close() after e2fsck_free_context() - since e2fsck_free_context may reference data in ctx->fs. - - * e2fsck.c (e2fsck_reset_context): Make sure ctx->fs is non-NULL - before checking ctx->fs->dblist. - -1998-10-09 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (e2fsck_pass1): Use the device check subroutine on - FIFO's and Socket's, so that we catch bogus immutable inodes. - - * pass2.c (process_bad_inode): Process bad socket and fifo's. - - * problem.h, problem.c: Define new problem codes PR_2_BAD_FIFO and - PR_2_BAD_SOCKET. - -1998-09-02 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * problem.c: Add PR_3_NO_DIRINFO error code. - - * super.c (check_super_value): Rename min and max to min_val and - max_val to avoid possible cpp macro conflicts. - - * pass4.c (e2fsck_pass4): Rename max to maxgroup, to avoid - possible cpp macro conflicts. - - * pass3.c (e2fsck_pass3): Rename max to maxdirs, to avoid possible - cpp macro conflicts. - (check_directory): Fix logic to avoid possible core dump - in the case of ext2fs_get_dir_info returning NULL. (By - the time we get here, it should never happen, but...). - Also simply/streamline the control flow of the function. - -1998-08-17 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * unix.c (check_if_skip): Move the "not cleanly mounted" check - ahead of the maximal mount and too long since checked tests. - (reserve_stdio_fds): Make sure 0,1,2 file descriptors are - open, so that we don't open the filesystem using the same - file descriptor as stdout or stderr. - -1998-08-01 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass2.c (e2fsck_pass2): Fix the progress accounting so that we - get to 100%. - - * pass3.c (e2fsck_pass3): Change progress accounting to be - consistent with the other e2fsck passes. - - * e2fsck.c (e2fsck_run): At the end of each pass, call the - progress function with the pass number set to zero. - - * unix.c (e2fsck_update_progress): If the pass number is zero, - ignore the call, since that indicates that we just want to - deallocate any progress structures. - -1998-07-09 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Release of E2fsprogs 1.12 - -1998-07-09 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * unix.c (main): Fix typo in checking the incompat feature set; it - should be checked against EXT2_LIB_FEATURE_INCOMPAT_SUPP. - -1998-07-07 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * badblocks.c (test_disk): Don't clear the existing bad blocks - list when using e2fsck -c, since it may cause blocks with - marginal errors to be dropped from the bad blocks list. - -Mon Jul 6 10:32:11 1998 Theodre Ts'o <tytso@lurch.mit.edu> - - * pass1.c (e2fsck_pass1): Use ext2fs_sb structure for - compatibility with older kernels. - -1998-06-25 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * unix.c (e2fsck_update_progress): Remove unused variables. - -1998-06-10 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c, problem.h: Change blkcnt_t to be e2_blkcnt_t to avoid - collision with LFS API. - - * pass1.c (e2fsck_pass1): Fix bug so that ext2_max_sizes is - properly initialized. - (e2fsck_pass1): Don't allow the the boot loader inode to - be a directory (clear the inode mode field if so). - -1998-05-07 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * unix.c (PRS): Added new option -C, which causes e2fsck to print - progress updates so that callers can keep track of the - completion progress of e2fsck. Designed for use by - progress, except for -C 0, which prints a spinning report - which may be useful for some users. - - * pass5.c (e2fsck_pass5): Use a finer-grained progress reporting - scheme (useful for larger filesystems). - - * e2fsck.h: Add progress_fd and progress_pos, for use by the Unix - progress reporting functions. - -1998-04-28 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (process_inode_cmp): Use EXT2_QSORT_TYPE to define the - appropriate return type for comparison functions for qsort. - - * e2fsck.h: Add #ifdef protection for unistd.h - - * super.c: Remove header files already included by e2fsck.h - -1998-04-26 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * dirinfo.c (e2fsck_add_dir_info): Update function to pass the old - size of the memory to be resized to ext2fs_resize_mem(). - -1998-03-30 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Makefile.in: Change to use new installation directory variables - convention. Fix uninstall rules to take $(DESTDIR) into - account. Remove cat8dir from the installdirs target, - since modern man package don't necessarily put the cat - directory in /usr/man/cat?. - -1998-03-29 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * super.c, e2fsck.h: Always declare e2fsck_get_device_size() as an - extern in e2fsck.h to prevent -Wall warnings. - - * pass4.c (e2fsck_pass4): Remove unused variable 'j'. - -1998-03-28 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * super.c (check_super_block): Fix broken superblock sanity check - when calculating blocks_per_group if s_log_frag_size != - s_log_block_size. Since we don't support fragments, this - isn't a bug deal, but it's good to get it fixed. - -1998-03-23 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * unix.c: Fix bug in check of feature set, to make sure we can - really fix this filesystem. - - * problem.h: Make blkcount type to be of type blkcnt_t. Make the - num field be a 64 bit type. Add the problem code - PR_1_FEATURE_LARGE_FILES - - * problem.c: Add table entry for the problem code - PR_1_FEATURE_LARGE_FILES. - - * pass1.c (e2fsck_pass1): A non-zero i_dir_acl field is only - a problem for directory inodes. (Since it is also - i_size_high now.) If there are no large_files, then - clear the LARGE_FLAG feature flag. If there are - large_files, but the LARGE_FLAG feature flag is not set, - complain and offer to fix it. - (check_blocks): Add support to deal with non-directory - inodes that have i_size_high set (i.e., large_files). - Don't give an error if a directory has preallocated - blocks, to support the DIR_PREALLOC feature. - (process_block, process_bad_block): The blockcnt variable - is a type of blkcnt_t, for conversion to the new - block_iterate2. - - * pass2.c (process_bad_inode): A non-zero i_dir_acl field is only - a problem for directory inodes. (Since it is also - i_size_high now.) - - * message.c (expand_inode_expression): Print a 64-bits of the - inode size for non-directory inodes. (Directory inodes - can only use a 32-bit directory acl size, since - i_size_high is shared with i_dir_acl.) Add sanity check - so that trying to print out the directory acl on a - non-directory inode will print zero. - (expand_percent_expression): %B and %N, which print - pctx->blkcount and pctx->num, can now be 64 bit - variables. Print them using the "%lld" format if - EXT2_NO_64_TYPE is not defined. - - * e2fsck.h: Add the large_flagsfield to the e2fsck context. - - * e2fsck.c (e2fsck_reset_context): Clear the large_flags - field. - -Sun Mar 8 23:08:08 1998 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass3.c (fix_dotdot_proc): - * pass2.c (check_dot, check_dotdot, check_name, check_dir_block): - * message.c (expand_dirent_expression): Mask off the high eight - bits of the directory entry's name_len field, so that it - can be used for other purposes. - -Fri Feb 27 00:01:39 1998 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.c (e2fsck_run): Since E2F_FLAG_SIGNAL_MASK doesn't - include EXT2_FLAG_RESTART anymore, we need to adjust this - routine so that it *does* return in the case of it seeing - EXT2_FLAG_RESTART. - - * pass1.c (e2fsck_pass1): ext2_get_next_inode() may call the group - done callback function, which may set context abort - flags. So we need to test the context abort flags after - we call ext2_get_next_inode(). - (process_inodes): If we abort due out of process_inodes, - do a clean exit by breaking out of the for loop instead of - just returning. - - * e2fsck.h (E2F_FLAG_SIGNAL_MASK): EXT2_FLAG_RESTART shouldn't be - considered a SIGNAL mask (i.e., requiring an immediate - abort of processing to restart). FLAG_RESTART just means - that we want to restart once pass 1 is complete. - -Tue Feb 24 15:19:40 1998 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Change the progress function to return an integer; if returns 1, - then the progress function is expected to have set the - e2fsck context flag signalling a user abort, and the - caller should also initiate a user abort. - -Tue Feb 17 19:03:44 1998 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass5.c (check_block_bitmaps, check_inode_bitmaps): Don't call - end_problem_latch() unless there was actually a problem - latched. Simplifies semantics of the latch processing. - -Mon Feb 16 17:31:44 1998 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.h: Add new field, priv_data to the e2fsck context - structure. It should be used by callers of the e2fsck - functions only, and not by anything in e2fsck itself. - -Mon Feb 7 17:31:04 1998 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * super.c: Instead of call ext2fs_get_device_size(), define and call - e2fsck_get_device_size(). (This function may be stubbed - out in special versions of e2fsck.) - - * pass3.c, pass4.c: Remove extra calls to the progress function - that weren't needed. - - * mke2fs.8.in: Update man page to note that the format of the bad - block file is the same as the one generated by badblocks. - -Sun Feb 1 07:57:14 1998 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * dirinfo.c, e2fsck.c: Don't include com_err.h; it isn't needed. - - * e2fsck.h: Include <time.h> since it is needed - - * super.c: If EXT2_SKIP_UUID is defined, then skip the UUID - processing. - -Tue Jan 20 15:37:01 1998 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * unix.c (main): In the case where the filesystem revision is too - high, print the message about the superblock possibly - being corrupt. - - * e2fsck.8.in: Add expanded comments about how the -b option - works. - -Sat Jan 17 13:02:16 1998 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.h: If EXT2_FLAT_INCLUDES is defined, then assume all of - the ext2-specific header files are in a flat directory. - - * dirinfo.c, ehandler.c, pass1.c, pass1b.c, pass2.c, pass5.c, - super.c, swapfs.c, unix.c: Explicitly cast all assignments - from void * to be compatible with C++. - -Tue Jan 6 11:30:24 1998 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * unix.c (sync_disk): Remove sync_disk and calls to that function, - since ext2fs_close() now takes care of this. - -Mon Dec 29 14:45:42 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c, pass1b.c, pass2.c, pass3.c, swapfs, badblocks.c, - ehandler.c, unix.c: Change use of private to be priv_data, - to avoid C++ reserved name clash. - -Fri Nov 28 09:30:26 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * dirinfo.c (e2fsck_add_dir_info): Use ext2fs_get_num_dirs instead - of e2fsck_get_num_dirs, which has been removed. - -Tue Nov 25 15:54:35 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Makefile.in (PROGS): Remove @EXTRA_PROGS@, since we don't want - to compile and install flushb. - -Mon Nov 24 06:48:00 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (e2fsck_pass1_check_device_inode): For now, only check - to make sure the extra i_blocks in a device are cleared if - the immutable flag is set. - - * util.c (print_resource_track): Fix typo which caused the - time/resource tracking to print "Pass 1 :" instead of - "Pass 1: ". - -Thu Nov 20 16:02:23 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass3.c (expand_directory): Fix bug which could cause core dump - when expanding the directory, and the bitmaps hadn't yet - been read in. Also, only use ext2fs_write_dir_block when - writing a directory block, not when writing out a fresh - indirect block. - -Wed Nov 19 16:15:44 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (process_bad_block): Fix bug where first_block wasn't - getting incremented, which caused user to get a - "Programming error?" warning if there was a bad block in a - non-primary superblock/group_descriptor. Also fixed - another bug so that ext2fs_bg_has_super() is consulted, to - avoid problems when using a filesystem with the - sparse_groups option set and there are bad blocks at the - beginning of a group which doesn't have a superblock. - -Thu Nov 6 16:10:20 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c, pass2.c, pass3.c, pass4.c, pass5.c: Add calls to the - progress indicator function. - - * pass1.c (scan_callback): Add call to the progress feedback - function (if it exists). - -Tue Nov 4 09:45:36 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * super.c (check_super_block): Skip the device size check if the - get_device_size returns EXT2_EXT_UNIMPLEMENTED. - - * iscan.c (main): Don't use fatal_error() anymore. - - * pass1b.c, swapfs.c, badblocks.c: Set E2F_FLAG_ABORT instead of - calling fatal_error(0). - - * problem.c, pass3.c (PR_3_ROOT_NOT_DIR_ABORT, - PR_3_NO_ROOT_INODE_ABORT): New problem codes. - - * problem.c, pass2.c (PR_2_SPLIT_DOT): New problem code. - - * problem.c, pass1.c (PR_1_SUPPRESS_MESSAGES): New problem code. - - * problemP.h: New file which separates out the private fix_problem - data structures. - - * util.c, dirinfo.c, pass1.c, pass1b.c, pass2.c, pass5.c, super.c, - swapfs.c util.c: allocate_memory() now takes a e2fsck - context as its first argument, and rename it to be - e2fsck_allocate_memory(). - -Mon Nov 3 14:35:29 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * unix.c (main): Add a special case check for the error code EROFS - and display an appropriate error message for this case. - - * [lots of files]: Change ext2fs_read_inode, ext2fs_write_inode - to take the e2fsck context as their first argument. - Change dir_info.c routines to take an e2fsck_context, - renamed them to start with e2fsck_ to avoid namespace - issues, and changed them to store the directory - information inside the e2fsck context. - Added e2fsck_run() which calls all of the e2fsck passes in - the correct order, and which handles the return of abort - codes. Added abort processing, both via setjmp/longjmp - and via flags in the e2fsck context. Use a flag in the - e2fsck context instead of the restart_e2fsck global - variable. Change uses of free and malloc to - ext2fs_free_mem and ext2fs_get_mem. - -Fri Oct 31 01:12:43 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c, pass3.c: Rename new error codes to _ET_ in them for - consistency. - -Sat Oct 25 00:10:58 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass3.c (get_lost_and_found): Check error return of - EXT2_FILE_NOT_FOUND instead of ENOTDIR - - * pass1.c (pass1_check_directory): Return EXT2_NO_DIRECTORY - instead of ENOTDIR - -Fri Oct 24 00:12:39 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * unix.c (PRS): Make the variable which getopt returns into be - an int, so that it won't lose on platforms where char is - unsigned. - - * pass1b.c (clone_file): Fix bug in error reporting in the case - where cs.errcode is non-zero. - -Sun Oct 19 21:12:11 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass*.c, super.c, unix.c, util.c, e2fsck.h: Place #ifdef - RESOURCE_TRACK around code which uses init_resource_track - and print_resource_track. (Not all systems have timeval) - - * super.c: Remove excess #includes which are not necessary. - - * e2fsck.h: Add #ifdef's for HAVE_SYS_TYPES_H and HAVE_SYS_TIME_H - -Fri Oct 3 13:40:03 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass*.c, super.c: Massive changes to avoid using printf and - com_err routines. All diagnostic messages are now routed - through the fix_problem interface. - -Sat Sep 6 17:13:28 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass2.c (check_dir_block): Check for duplicate '.' and '..' - entries. - - * problem.c, problem.h: Add new problem codes PR_2_DUP_DOT and - PR_2_DUP_DOT_DOT. - -Tue Sep 2 09:04:51 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * problem.c: Added new problem codes for some of the - superblock corruption checks, and for the pass header - messages. ("Pass 1: xxxxx") - - * util.c (print_resource_track): Now takes a description - argument. - -Mon Aug 25 10:23:13 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * super.c, unix.c, e2fsck.c: New files to separate out the - operating-specific operations out from e2fsck.c. - e2fsck.c now contains the global e2fsck context management - routines, and super.c contains the "pass 0" initial - validation of the superblock and global block group - descriptors. - - * pass1.c, pass2.c, pass3.c, pass4.c, pass5.c, util.c: Eliminate - (nearly) all global variables and moved them to the e2fsck - context structure. - - * problem.c, problem.h: Added new problem codes PR_0_SB_CORRUPT, - PR_0_FS_SIZE_WRONG, PR_0_NO_FRAGMENTS, - PR_0_BLOCKS_PER_GROUP, PR_0_FIRST_DATA_BLOCK - -Thu Aug 14 10:55:21 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * message.c: Add compression for the word "Illegal" - - * problem.c: Added entries for PR_2_BAD_CHAR_DEV and - PR_2_BAD_BLOCK_DEV - -Wed Aug 13 09:55:57 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (pass1, check_device_inode), pass2.c - (process_bad_inode): Use a more stringent test for a valid - device. - -Sun Aug 10 18:58:02 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.c (check_mount): Add stronger warning message about the - perils of running e2fsck on a mounted filesystem. - -Tue Jun 17 01:33:20 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Release of E2fsprogs 1.11 - -Thu Jun 12 00:25:31 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (mark_table_blocks): Mark the superblock and group - descriptor blocks first, so that any conflicts between - these and the bitmap or inode table blocks is noticed. - - * problem.c: Fix message printed out when a block or inode bitmap - conflicts with other fs data, has the correct group number - in it. - -Tue Jun 10 12:07:37 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass2.c (pass2): Check the error return from ext2fs_dblist_iterate. - -Thu May 8 22:45:27 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.8.in: Fix minor typos and grammer oops found by Bill - Hawes (whawes@star.net). - - * badblocks.c (read_bad_blocks_file): Pass the blocksize to the - bad blocks command so that all of the filesystem gets - tested in the case where the blocksize 2048 or 4096. - -Thu Apr 24 12:16:42 1997 Theodre Ts'o <tytso@localhost.mit.edu> - - * Release of E2fsprogs version 1.10 - -Mon Apr 21 22:43:08 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1b.c (pass1b): While scanning for inodes, simply skip inodes - where ext2fs_get_next_inode returns the - EXT2_ET_BAD_BLOCK_IN_INODE_TABLE error. - -Thu Apr 17 12:23:38 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Release of E2fsprogs version 1.09 - -Fri Apr 11 18:56:26 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Release of E2fsprogs version 1.08 - -Thu Apr 10 13:51:16 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1b.c (clone_file_block): If we clone a directory, we need to - update the dblist entry so that we check (and correct) the - right directory block. - -Sun Apr 6 09:13:12 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (process_block): Don't clear blocks references to - filesystem metadata; let pass 1B handle this case. - - * problem.c, problem.h: Add new problem, PR_1B_SHARE_METADATA. - - * pass1b.c (pass1d): Deal with a block which is shared with - filesystem metadata. - - * e2fsck.h: Make block_illegal_map be a global variable - -Sat Apr 5 11:51:58 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.c, pass1.c (mark_table_blocks): Support the sparse_super - feature. - (get_backup_sb): New function which attempts to obtain the - correct backup superblock (if possible). - -Fri Apr 4 10:46:26 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.c (main): Check the version of the library, and warn if - the library is out of date; this happens generally due to - users who manually install e2fsprogs. - - * pass1.c (pass1_get_blocks): If the passed in inode number for - get_blocks isn't what we're expecting pass back - EXT2_ET_CALLBACK_NOT_HANDLED. - -Wed Mar 12 13:32:05 1997 Theodore Y. Ts'o <tytso@mit.edu> - - * Release of E2fsprogs version 1.07 - -Tue Mar 11 10:31:47 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * icount.c: New file which implements an inode count abstraction. - This significantly reduces amount of memory needed to - check really large filesystems. - - * pass1.c, pass2.c, pass3.c, pass4.c: Modified to use the icount - abstraction. - -Fri Mar 7 08:28:55 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * dirinfo.c (dir_info_iter): New function which allows iteration - over the directories in the dir_info map. - - * pass3.c (pass3, check_directory): Speed up pass 3 by iterating - over all directories in the dir_info structure, instead of - iterating all inodes looking for directories, and then - looking up the directories using get_dir_info(). - -Sat Feb 1 11:33:43 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (pass1, process_block): - * pass2.c (pass2): Use the ext2_dblist abstraction instead of - manual code to manage the directory block list information. - - * pass1.c (check_blocks), pass1b.c (pass1b), pass2.c - (deallocate_inode): Call the ext2 library routine - ext2_inode_has_valid_blocks() instead of - inode_has_valid_blocks(). - - * swapfs.c (swap_inodes): Add check so that we don't try to call - swap_inode_blocks unless the inode has valid blocks. - (Otherwise a long fast symlink might cause - swap_inode_blocks to erroneously get called.) - -Wed Jan 22 14:42:53 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * problem.c, problem.h: Added entries for PR_2_DOT_NULL_TERM and - PR_2_DOT_DOT_NULL_TERM. - - * pass2.c (check_dot, check_dot_dot): Make sure the new . and - .. entries are null-terminated, since the 2.0 kernel - requires this (for no good reason). - -Mon Jan 20 20:05:11 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (pass1): Set the EXT2_SF_SKIP_MISSING_ITABLE flag so - that we can recover from missing inode tables. - - * dirinfo.c (get_dir_info): If there are no directories in the - dir_info abstraction, don't core dump (because dir_info is - NULL). - - * e2fsck.c (main): Don't try using the backup superblocks if there - aren't any. - (check_super_block): If there are illegal inode table or - bitmaps, set the filesystem as being in error. - -Wed Jan 15 11:32:01 1997 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass2.c (check_dir_block): Added check to make sure that rec_len - is a a multiple of 4 (so that the directory entries are - 4-byte aligned). - -Sat Dec 28 12:16:32 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Makefile.in (uninstall): Uninstall all programs in the PROGS - line. - (PROGS): Don't build and install the extend program by - default. - - -Sat Dec 7 16:41:02 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (process_inodes): Make sure the stashed_ino variable is - saved and restored as well. - (pass1): For fast sym links, skip the check_blocks - processing step altogether. - -Mon Dec 2 09:28:24 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * problem.c, message.c: New files, to completely refurbish how - filesystem problems are reported to the user. The - diagnostic messages are now encoded out in an easily - customizable, extensible format. The messages printed out - in preen mode are all on one line, and contain the device - name. - -Fri Nov 29 20:26:08 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * swapfs.c (swap_inodes): When swapping a filesystem, ignore - deleted files. - - * pass1.c (pass1): Ignore missing inode table errors during the - scan, and just skip past those inodes. - - * pass3.c (check_root): Remove root_ino argument, and assume that - the root inode must be EXT2_ROOT_INO. Move responsibility - of setting the parent of the root inode in the directory - inode structure to pass2(). - - * pass2.c (check_dir_block): Don't allow links to the root - directory. - - * dirinfo.c (add_dir_info): Remove last argument to add_dir_info, - since the inode is no longer used. - -Tue Oct 15 00:06:49 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.c (main): If the superblock magic number is wrong, or the - block group fails a sanity check, then automatically - restart trying to use the backup superblocks. - - * pass1.c (mark_table_blocks): Make the inode tables ahead of - everything else; in the case where a bitmap block overlays - the inode table, the inode table should take precedence. - - * pass2.c (maybe_clear_entry): Make the deleted/unused error - message fit on one line, since the error can happen during - a preen pass. (We eventually need to revamp the whole - e2fsck error reporting and prompting system, but that's a - job for another day.) - -Mon Oct 14 22:29:49 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.c (main): Read in the list badblocks into fs->badblocks - for the benefit of the inode scan functions. - - * pass1.c (pass1): If ext2fs_get_next_inode() returns an error - indicating that an inode is in a bad block, mark that - inode as being used, as well as in the inode "bb" map. - - * pass2.c (maybe_clear_entry): New function which generalizes the - error cases when a directory entry needs to be cleared. - (check_dir_block): If an inode is in the "bb" map, offer - to clear the directory entry, since the inode is in a bad - block. - - * pass4.c (pass4): If an inode is marked as used, but is is marked - in the "bb" map, don't process it as a disconnected inode. - -Tue Oct 8 02:02:03 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Release of E2fsprogs version 1.06 - -Mon Oct 7 00:45:30 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.c (main): Print out the version number of the shared - library when using the -V option. - - * swapfs.c (swap_filesys): Change EXT2_SWAP to EXT2_FLAG_SWAP for - consistency's sake. - - * e2fsck.c (main): By setting EXT2_FLAG_MASTER_SB_ONLY, only write - out the backup superblocks when we know we have a valid - filesystem. - -Tue Oct 1 22:00:29 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * util.c (preenhalt): Make it explicit on preenhalt that running - e2fsck manually means without the -a or -p flag. - -Fri Sep 27 14:41:08 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (pass1): Add support for the EXT2_BOOT_LOADER inode. - (Linux/FT actually sets the mode bits, so we have to - handle it specially.) - - * e2fsck.c (check_if_skip): Print a message if the filesystem is - just dirty, so that the user knows that it's about to be - checked (since this will take a while). - -Mon Sep 16 17:00:01 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass4.c: If a disconnected inode is zero-length, prompt to - delete it instead of connecting it to lost+found. - -Thu Sep 12 15:23:07 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Release of E2fsprogs version 1.05 - -Fri Aug 30 20:24:30 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass4.c (pass4): If the user refuses to connect an unattached - inode to lost+found, don't try to set i_links_count. This - is bad, since if the user says yes, the inode will be - marked as unused, which is not necessarily the right - thing, especially since the rest of the cleanup doesn't - happen here. - - * pass2.c (deallocate_inode): Set inode_link_info[ino] when - dellocating an inode. (Not strictly necessary, but...) - - * pass4.c (pass4): Add "bonehead" explanation to the "programming - error" message. - -Tue Aug 27 11:26:32 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.c (PRS,main): Added new options -s and -S. -s will - byte-swap the filesystem so that it is normalized. -S - will byte-swap the filesystem regardless of its current - byte-order. - - * swapfs.c: New file, which will byte-swap a filesystem. - -Tue Aug 20 09:41:37 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (pass1): Change handling on files with non-zero dtime - and non-zero i_link_count; before we treated them as - deleted file per botched ext2 0.3c kernel behavior. We - now clear dtime instead. - -Mon Aug 19 23:33:57 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.c (main): If e2fsck sets the clean bit, even if - nothing else is changed, make sure FSCK_NONDESTRUCT is - set (since after all having the filesystem set to - invalid is an error. :-) - -Fri Aug 9 10:25:13 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * pass1.c (process_block): Make sure that meta data doesn't get - accidentally set in the dir_blocks array (which could - happen in some error condtions). - - * pass1.c (pass1): - * pass2.c (process_bad_inode): Check for fragments in a - OS-independent fashion. - -Thu Aug 8 15:20:54 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.c (check_if_skip): Close the filesystem when skipping the - cleanup for the filesystem. - -Mon Jul 22 22:03:28 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.c: Improve corrupt_msg, so that it's less confusing. - -Thu May 16 11:12:30 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * Release of E2fsprogs version 1.04 - -Wed May 15 21:41:29 1996 Theodore Ts'o <tytso@rsts-11.mit.edu> - - * e2fsck.c (relocate_hint): Issue a hint that the user may wish to - try "e2fsck -b 8193" before allowing e2fsck to relocate - inode table blocks. - - * Makefile.in (e2fsck): Build e2fsck statically or dynamically - depending on the option to configure. Added targets for - e2fsck.static and e2fsck.shared for people who want to - build a static or shared variant manually. - -Wed Mar 27 00:33:40 1996 <tytso@rsts-11.mit.edu> - - * Release of E2fsprogs version 1.03 - -Tue Mar 26 12:03:42 1996 <tytso@rsts-11.mit.edu> - - * e2fsck.c (show_stats): Don't use floating point to display - percentage of non-contiguous files, as different libc - handle result truncation differently, and this causes the - test suite to bomb out depending on which libc you are - using. - - * util.c (allocate_memory): Fix error message to omit extraneous - %%s. - -Tue Mar 5 03:50:40 1996 <tytso@rsts-11.mit.edu> - - * pass4.c (pass4): - * pass2.c (check_dir_block): - * pass1.c (pass1): Add support for dynamic first inode revision. - -Wed Feb 14 16:27:30 1996 <tytso@rsts-11.mit.edu> - - * pass3.c (check_root): Fix spelling typo - -Mon Feb 5 22:30:30 1996 <tytso@rsts-11.mit.edu> - - * e2fsck.c (check_super_block): If the superblock fails certain - internal consistency checks, exit with a fatal error after - printing the "superblock is corrupt message". - -Wed Jan 31 11:06:08 1996 <tytso@rsts-11.mit.edu> - - * Release of E2fsprogs version 1.02 - -Wed Dec 15 21:24:26 1996 <tytso@rsts-11.mit.edu> - - * pass1.c (process_block): Check to see if a file is "fragmented". - i.e., non-contiguous. Note that any file which is larger - than the block group is guaranteed to be non-contiguous. - We may want to use a different hueristic for deciding - whether or not a file is "fragmented". - - * e2fsck.c (show_stats): Print statistics of how many - non-contiguous files are on the system. - -Fri Dec 15 19:19:47 1995 <tytso@rsts-11.mit.edu> - - * badblocks.c (read_bad_blocks_file, test_disk): Fold - functionality of test_disk() (which runs badblocks) into - read_bad_blocks_file(); test_disk() now calls - read_bad_blocks_file() with a NULL bad_blocks_file - argument. - -Mon Nov 20 18:30:10 1995 <tytso@rsts-11.mit.edu> - - * e2fsck.c (check_mount): Use #if defined(__linux__) instead of - #if defined(linux). The latter won't work if we're - compiling -ansi. - -Mon Oct 30 20:31:17 1995 <tytso@rsts-11.mit.edu> - - * e2fsck.c (check_mount): For Linux systems, the check to see if - the root is mounted read-only has to be done for all - filesystems, not just for the root filesystem, due to the - way that some /etc/rc scripts are set up. - -Thu Oct 26 12:05:30 1995 <tytso@rsts-11.mit.edu> - - * Makefile.in (install): Strip programs when they are installed. - (e2fsck): Build e2fsck statically. - -Wed Oct 25 21:18:16 1995 <tytso@rsts-11.mit.edu> - - * util.c (preenhalt): Preenhalt now takes an argument, which is an - ext2fs_filsys; this allows it to set the EXT2_ERROR_FS - flag in the superblock in cases where preenhalt is called. - All calls to preenhalt() were changed to either - preenhalt(fs) or preenhalt(NULL) in a few cases where the - fs pointer was not available. (Most notable, for block - read/write errors.) - -Mon Sep 4 21:41:03 1995 Remy Card <card@bbj> - - * ehandler.c: - util.c: Include <sys/time.h> before <sys/resource.h>. BSD needs it. - -Mon Sep 4 10:14:49 1995 <tytso@rsts-11.mit.edu> - - * e2fsck.c (show_stats): Show statistics about how many inodes - have indirect, doubly indirect, and triply indirect - blocks. Allow up to 8 digits for statistics, instead of - merely 6, so things look pretty for large filesystems. - - * pass1.c (pass1): Keep statistics about indirect, doubly - indirect, and triply indirect blocks. - - * pass1.c (unwind_pass1): Clear the above statistics when unwinding - pass 1. - -Fri Aug 18 15:17:10 1995 Theodore Y. Ts'o <tytso@dcl> - - * util.c, ehandler.c: Move #include of <sys/resource.h> after - #include of "e2fsck.h", since sys/resource.h may depend on - sys/time.h, which is #included in e2fsck.h. - -Thu Aug 17 22:33:37 1995 <tytso@rsts-11.mit.edu> - - * e2fsck.c (check_mount): Use the new ext2fs_check_if_mounted() - function to determine if the device is mounted. - - * e2fsck.c (main): Add better error messages if ext2fs_open() - fails. - -Wed Aug 16 16:25:02 1995 <tytso@rsts-11.mit.edu> - - * pass1.c (check_blocks): If we're clearing a directory, clear - pb.is_dir so we don't do the following check of making - sure the directory size matches; this is pointless, since - we've already cleared the inode. - -Fri Aug 11 09:08:54 1995 Theodore Y. Ts'o <tytso@lurch.mit.edu> - - * pass1.c (bad_primary_block): New function, called by - process_bad_block, which explains the facts of life to the - user when a block in the primary superblock or primary - group descriptors is bad. - - * pass2.c (check_dot): Handle the case where the first directory - entry is used, but not ".". - - * pass2.c (check_dotdot): Handle the case where the second directory - entry is used, but is not "..". - -Thu Aug 10 10:05:10 1995 Theodore Y. Ts'o <tytso@lurch.mit.edu> - - * e2fsck.c (check_super_block): Get the size of the physical - device and if it is smaller than the reported size of the - filesystem, report an error. - -Sat Aug 12 03:39:18 1995 Remy Card <card@bbj> - - * e2fsck.c (check_if_skip): Print the number of allocated files and - blocks on clean filesystems. - -Fri Aug 11 14:15:36 1995 Remy Card <card@bbj> - - * e2fsck.8: Updated date and version number. - -Thu Aug 10 14:26:01 1995 Remy Card <card@bbj> - - * pass1.c (check_blocks): Check that directory size matches *exactly* - the count of allocated blocks. - -Wed Aug 9 21:21:24 1995 Theodore Y. Ts'o <tytso@dcl> - - * pass1b.c (pass1d): Free the shared[] array when we're done with - it to avoid a memory leak. - - * pass1.c (unwind_pass1): Use ext2fs_free_block_bitmap to free the - block_dup_map. - - * pass2.c (process_bad_inode): When clearing the inode, make sure - the pathname is freed, to prevent a memory leak. - - * pass5.c (check_inode_bitmaps): Free free_array and dir_array - when we're finished with them. - (check_block_bitmaps): Free free_array when we're finished - with them. - - * Makefile.in (e2fsck, flushb): Use $(LD) instead of $(CC) when - linking the executable. - - * pass2.c (process_bad_inode): Even on OS's that don't support the - fragment fields, make sure the Linux equivalent fields are - set to zero. If an OS wants to reuse these fields, which - is probably a bad idea (although we may get desperate in - the future) this code will have to be changed. - - * pass1.c (dir_block_cmp): If the block numbers are equal, compare - on the inode field, and then blockcnt field. This is just - to keep the ordering of dir_blocks the same on all - platforms when there are more than on missing directory - blocks, which are indicated directories with holes, which - are indicated with the block number being set to zero. - -Sun Aug 6 15:40:58 1995 Theodore Y. Ts'o <tytso@lurch.mit.edu> - - * pass1.c (check_blocks, process_block): check_blocks() modified - to call the ext2fs_block_iterate() with BLOCK_FLAG_HOLE if - the inode is a directory. process_block() now checks to - see if a directory has a "hole", or missing block. If so, - this fact is recorded in the directory block list so that - the problem can be resolved in pass #2. - - * pass2.c (allocate_dir_block): Added allocate_dir_block() to - allocate new blocks for directories with "holes". Called - out of check_dir_block if a block in the directory block - list is zero. - - * pass3.c (get_lost_and_found): Move location of free(block) to - prevent possible memory leak. - -Sat Aug 5 12:42:22 1995 Theodore Y. Ts'o <tytso@lurch.mit.edu> - - * pass2.c (check_dir_block): Use a automatic, fixed-saize array - instead of alloca() --- alloca is not portable! Check to - make sure the filename is not longer than EXT2_NAME_LEN, - and offer to fix it by truncating it, since it should - never happen. - - * e2fsck.c (PRS): Use malloc() instead of alloca() --- alloca() is - not portable!! In any case putenv() in some systems must - take a static character array or malloc()'ed memory; - passing memory allocated using alloca() to putenv() is not - advisable. - - * pass2.c (check_dot, check_dotdot): Use malloc() instead of - alloca() --- alloca() is not portable!!! - -Tue Jul 18 20:04:02 1995 <tytso@rsx-11.mit.edu> - - * pass1b.c (pass1c): - * pass3.c (check_root, get_lost_and_found): - * pass2.c (check_dir_block): Use ext2fs_{read,write}_dir_block - to read/write the directory block. - -Mon Jul 17 04:00:56 1995 <tytso@rsx-11.mit.edu> - - * util.c (ask_yn): Apply patch supplied by Peter A. Zaitcev to - make sure VMIN and VTIME are set correct. - -Fri Jul 14 19:26:29 1995 <tytso@rsx-11.mit.edu> - - * pass1.c (mark_block_used): Change to be an inline function. - Assume that the block validity checks are already done, - and use the fast variant of the bitmap functions. - -Thu Jul 13 08:10:55 1995 <tytso@rsx-11.mit.edu> - - * pass5.c (check_block_bitmaps, check_inode_bitmaps): Check the - bounds of the bitmaps in advance, and then use the fast - variant of e2fs_test_{block,inode}_bitmap. - - * pass1.c (mark_block_used): Use ext2_fast_mark_block_bitmap since - the bounds checking has already been done earlier. - -Wed Jul 12 02:22:46 1995 <tytso@rsx-11.mit.edu> - - * pass1.c (pass1): Allocate and free the block_illegal_map, which - is used for shortcut processing in process_block. - (mark_table_blocks): Initialize block_illegal_map with the - filesystem blocks. - (describe_illegal_block): New helper function that - describes why a block is illegal. - (process_block): Use block_illegal_map as a shortcut - to determine whether a block is bad. Use - describe_illegal_block to print out why the block is illegal. - -Mon Jun 12 19:11:06 1995 Theodore Y. Ts'o (tytso@dcl) - - * flushb.c: Don't include <linux/fs.h> if it doesn't exist. - - * scantest.c: Don't include <linux/fs.h>, <getopt.h>, or - <mntent.h> if they don't exist. (Mostly so that "make - depend" works.) - - * pass1.c, pass1b.c, pass3.c, badblocks.c: Include <errno.h> (if - it exists). - - * e2fsck.c, scantest.c: Don't include <getopt.h> if it doesn't - exist. - -Mon Jun 12 08:37:49 1995 Theodore Y. Ts'o <tytso@lurch.mit.edu> - - * pass2.c (process_bad_inode, check_for_zero_long, - check_for_zero_char): Change long to u32, and char to u8. - -Sun Jun 11 15:05:57 1995 Theodore Y. Ts'o <tytso@lurch.mit.edu> - - * util.c (inode_has_valid_blocks): - * pass2.c (process_bad_inode): - * pass1.c (pass1, check_blocks, pass1_check_directory): Use - LINUX_S_IS* instead of S_IS*. - - * e2fsck.h: Don't #include <sys/stat.h> - - * flushb.c (main): Add #ifdef BLKFLSBUF around ioctl. (Although - this program is pretty much useless if BLKFLSBUF isn't - supported.) - - * e2fsck.c, badblocks.c: Add #include <errno.h>, since errno is - used. - -Thu Jun 8 12:31:19 1995 Miles Bader <miles@churchy.gnu.ai.mit.edu> - - * pass2.c (check_dot, check_dotdot, check_dir_block): Use alloca - to allocate space for file names instead of using fixed size buffers. - (process_bad_inode): Only check inode frag fields if - HAVE_EXT2_FRAGS is defined (by configure). - * pass1.c (pass1): Only check the inode frag fields if - HAVE_EXT2_FRAGS is defined (by configure). - - * e2fsck.c (check_mount): Only check for a mounted filesystem if - HAVE_MNTENT_H is defined (by configure). - (PRS): Use alloca to allocate the new path string, instead of - having a fixed size buffer (which was the wrong size anyway). - (PRS): Only support the -F (flush) option if the BLKFLSBUF ioctl - is defined. - - * e2fsck.h: Only include <linux/fs.h> if HAVE_LINUX_FS_H is - defined (by configure). - - * Makefile.in: Rewritten to conform to GNU coding standards and - support separate compilation directories. - -Thu Apr 6 15:04:36 1995 Remy Card <card@bbj.ibp.fr> - - * pass1.c (pass1): Test the mode in reserved inodes (must be zero). - -Sat Mar 11 13:12:16 1995 Theodore Y. Ts'o <tytso@localhost> - - * pass1.c (unwind_pass1): Clear the file type statistics counter - when pass 1 needs to be restarted from scratch. - - * pass1.c (handle_fs_bad_blocks): Fix bug where bitmap blocks were - being reallocated to blocks in the next block group, - instead of the current block grup. - - * pass1.c (pass1, check_blocks): Set inode_link_info[ino] whenever - inode.i_links_count is set. - -Tue Feb 14 01:38:04 1995 Theodore Y. Ts'o (tytso@rt-11) - - * pass1.c (process_block): Add checks for if the block is - trepassing on a superblock or group descriptor table. - -Sat Dec 31 00:52:11 1994 <tytso@rsx-11.mit.edu> - - * main.c (corrupt_msg): Extend the message which is printed out - when the superblock is corrupt, to include the suggestion - of using the -b option to specify an alternate superblock. - -Thu Nov 24 09:29:58 1994 Theodore Y. Ts'o (tytso@rt-11) - - * badblocks.c (read_bad_blocks_file): If we are adding or - replacing bad blocks in the bad blocks inode, sanity check - the bad block inode first, and clear out any illegal blocks. - - * pass2.c (check_name): Don't bomb out if the attempt to get the - pathname of the containing directory returns an error; the - directory may be too badly damaged to expect that - ext2fs_get_pathname will always succeed. Use "???" if the - pathname can't be obtained (it's only for a printf to the - user anyway). - - The name of the containing directory and the bad filename - were incorrectly interchanged in the user message. Fixed. - - * pass2.c (check_name, check_dir_block): Use a common static - string for the unknown pathname. - -Mon Nov 7 22:30:54 1994 Remy Card <card@bbj> - - * Fixed lots of printf formats to make sure that block and inode - numbers are printed as unsigned integers. - -Mon Oct 24 14:10:46 1994 (tytso@rsx-11) - - * pass5.c (check_block_end): Fix calculation of how the last block - in the block bitmap should be calculated. - -Wed Sep 7 10:01:13 1994 (tytso@rsx-11) - - * pass1b.c (pass1_dupblocks): Fix declaration of dup_inode_map to - be an ext2fs_inode_bitmap, and free it properly. - - * e2fsck.h - * e2fsck.c (main): Folded in Remy Card's changes to add a revision - level to the superblock. - -Wed Aug 17 22:00:20 1994 Remy Card (card@bbj) - - * e2fsck.c (usage): Fixed bogus usage message. - -Wed Aug 17 11:21:45 1994 Theodore Y. Ts'o (tytso@rt-11) - - * pass1.c (process_bad_block): Fixed bug so that blocks in the - backup superblocks and group descriptors are handled gracefully. - |