xref: /aosp_15_r20/external/e2fsprogs/doc/RelNotes/v1.40.txt (revision 6a54128f25917bfc36a8a6e9d722c04a0b4641b6)

E2fsprogs 1.40.11 (June 17, 2008)
=================================

Mke2fs, tune2fs, and resize2fs now use floating point to calculate the
percentage of reserved blocks.  (Addresses Debian Bug: #452639)

Updated Spanish and Catalan translations.

Fixed various Debian packaging issues --- see debian/changelog for
details.  (Addresses Debian Bugs: #483962, #483023)

Add detection for ZFS volumes to the libblkid library.

Fixed spelling mistakes, typos, and otherwise clarified man pages.
(Addresses Debian Bug: #486463)

Programmer's notes:
-------------------

Fix marginal C code in probe_lvm2() function to the blkid library more
portable for older compilers.

Fix build problems on MacOS X.  (Addresses Sourceforge Bug: #1972473)

Fix ext2fs_swap{16,32,64} functions so they can be used by external
applications on big-endian machines.  (Addresses Debian Bug: #484879)


E2fsprogs 1.40.10 (May 21, 2008)
================================

When deciding whether or not to revalidate a blkid cache entry, if the
device's mtime is newer than the last time the cached entry was
validated, force a revalidation.

Fix a potential data corruption bug in e2fsck in the journal replay.
The chances of this is happening is extremely remote, especially the
default data=ordered or data=writeback modes.  However, if a block
which has been journalled starts with the first four bytes 0xc03b3998,
when e2fsck replays the journal, those four bytes will be replaced
with zero's.  Fortunately, it is highly, highly unlikely for e2fsck
metadata to begin with those fatal 4 byte sequence, and unless
data=ordered mode is in use, data blocks are never journalled.

Updated German, Dutch, Swedish, and Vietnamese translations.

Programmer's notes:
-------------------

Fixed various Debian packaging issues --- see debian/changelog for
details.

Remove default sizes of types when cross compiling, since autoconf
2.50 can figure this out automatically now.


E2fsprogs 1.40.9 (April 27, 2008)
=================================

SuSE's security team audited uuidd and came up with a few minor
issues.  None of them are serious given that uuidd runs setuid as a
unprivileged user which has no special access other than libuuid
directory, but it's good to get them fixed.

One additional fix in ext2fs_swap_inode_full() needed for resize2fs to
work correctly with in-inode extended attributes.

Updated German, Czech, Dutch, French, Polish, Swedish, and Vietnamese
translations.

Debugfs will avoid using a pager if the standard output is not a tty.

Fix debugfs and tune2fs to correctly handle daylight savings time when
parsing a time string.

Fixed spelling mistakes, typos, and otherwise clarified man pages.

Fix fsck completion bars when multiple filesystems were being checked
in parallel.  (Addresses Debian Bug: #432865, Addresses Launchpad Bug:
#203323, Addresses Sourceforge Bug: #1926023)

Fix fsck so that progress information is sent back correctly when
multiple filesystems are being check and the output of fsck is being
redirected to a file descriptor.  Also, include the device name (w/o
spaces) in the progress information sent back via a file descriptor.
(Addresses Launchpad Bug: #203323, Addresses Sourceforge Bug:
#1926023)

Teach fsck to treat "ext4" and "ext4dev" as ext* filesystems.

If logsave receives a SIGTERM or SIGINT signal, it will now pass that
signal to its child process.

Fix mke2fs's creation of are resize inode when there is a non-standard
s_first_data_block setting.

Fix bug in blkid when run by an unprivileged user; most devices were
not reported correctly.  9Addresses Launchpad Bug: #220275)

Mke2fs will not allow the logically incorrect combination of
resize_inode and meta_bg, which had previously caused mke2fs to create
a corrupt filesystem.

Fix fsck in German locales so that a 'j' means yes.
(Addresses Sourceforge Bug: #1947683)


Programmer's notes:
-------------------

Fixed various Debian packaging issues --- see debian/changelog for
details.

Update valgrind options in test_script to work with valgrind 3.2.3

Update texinfo.tex to a much newer version from the FSF.

Remove bashism for configure script and from the lib/ss Makefile.
Addresses Sourceforge Bug: 1921969

Fix some silently broken tests: m_no_opt, m_meta_bg, and m_raid_opt.

Fix build system so that if texinfo is not installed, it won't print a
(harmless) error message.


E2fsprogs 1.40.8 (March 13, 2008)
=================================

Fixed e2image -I so it works on image files which are larger than 2GB.

Fixed e2fsck's handling of directory inodes with a corrupt size field.  If
the size is larger than the number of blocks found in the inode, don't
try to allocate extra empty blocks at the end of the directory to make
up the difference; there's no point to doing that.  In addition, if
the size is not a multiple of a blocksize, always fix it.

E2fsck handled a pass 2 "should never happen error" by not giving
enough information and then core dumping.  Unfortunately, it was all
too easy to trigger the "should never happen" situation if a
directory's inode size was not correct.  This has been fixed, but
e2fsck has also been taught how to handle this situation more
gracefully, by simply removing the inode hash tree information, so
that it can be rebuilt again after e2fsck's pass 3.  (Addresses
Launchpad Bug: #129395)

Resize2fs had a bug resizing large inodes with extended attributes
that was fixed in 1.40.6; unfortunately, it turned out it wasn't fixed
completely on big-endian systems such as PowerPC.  The bug should be
completely fixed now.  Yay for regression test suites.  (Addresses Red
Hat Bugzilla: #434893)

Updated German, Czech, Dutch, Polish, Swedish, and Vietnamese
translations.  Many thanks to Philipp Thomas from Novell for stepping
up to become the new German translation maintainer!  (Addresses Debian
Bugs: #302512, #370247, #401092, #412882).

When e2fsck is clearing a corrupt inode's HTREE directory information,
make it clear that it is just clearing the HTREE information, not the
entire inode.

Fixed spelling mistakes, typos, and otherwise clarified man pages.

Programmer's notes
------------------

Add new functions, ext2fs_dblist_get_last() and
ext2fs_dblist_drop_last(), which allows the caller to examine the last
directory block entry added to the list, and to drop if it necessary.

Fixed a portability problem in libblkid with DJGPP.

Fix an obvious typo in an "internal error" message in e2fsck.  Thanks
to Philipp Thomas for pointing this out.

If the info files are not built, change "make install" so it doesn't
fail with an error code.


E2fsprogs 1.40.7 (February 28, 2008)
====================================

Remove support for clearing the SPARSE_SUPER feature from tune2fs, and
depreciate the -s option, since it can result in filesystems which
e2fsck can't fix easily.  There are very good reasons for wanting to
disable sparse_super; users who wants to turn off sparse_super can use
debugfs.  (Addresses Sourceforge Bug: #1840286)

Add missing options to mke2fs's usage message.  (Addresses Sourceforge
Bug: #1751393)

Fix bug in resize2fs when large (greater than 128 byte) inodes are
moved when a filesystem is shrunk; it was only moving the first 128
bytes, so extended attributes were not getting moved.  (Addresses Red
Hat Bugzilla: #434893)

E2fsck now prints an explicit message when the bad block inode is
updated, to avoid confusion about why the filesystem was modified.
(Addresses Sourceforge Bug: #756460)

Allow mke2fs and tune2fs manipulate the large_file feature.
Previously we just let the kernel and e2fsck do this automatically,
but e2fsck will no longer automatically clear the large_file feature.
It still isn't really necessary to worry about this feature flag
explicitly, but some users seem to care.  (Addresses Red Hat Bugzilla:
#258381)

Suppress message about an old-style fstab if the fstab file is empty.
(Addresses Debian Bug: #468176)

Fix (really minor) bug in debugfs's find_free_block so it avoids
reporting a free block more than once if there are too few free blocks
in the filesystem.  (Addresses Sourceforge Bug: #1096315)

Change e2fsck to no longer clear the LARGE_FILES feature flag
automatically, when there are no more > 2GB files in the filesystem.
It's been almost a decade since there have been kernels that don't
support this flag, and e2fsck clears it quietly without telling the
user why the filesystem has been changed.

Fix bug which could cause libblkid to seg fault if a device mapper
volume disappears while it is being probed.  (Addresses RedHat
Bugzilla: #433857)

Enhance e2fsck's reporting of unsupported filesystem feature flags.
(Addresses Sourceforge Feature Request: #1175808)

Fix option syntax in dumpe2fs for explicit superblock and blocksize
parameters.  What was currently documented in the man page has been
broken for some time, due to getopt() implementation changes.  The
option syntax has been changed to one which is can be more portable
supported and which is consistent with the format for extended options
in mke2fs and tune2fs.  (Addresses Sourceforge Bug: #1830994)

Add support to tune2fs to clear the resize_inode feature.  This
requires an fsck afterwards.  (Addresses Red Hat Bugzilla: #167816)

Teach blkid to detect LVM2 physical volumes.  (Addresses Red Hat
Bugzilla: #409321)

Add support for setting RAID stride and stripe-width via mke2fs and
tune2fs.  Teach dumpe2fs to print the RAID parameters.

Add support for setting new superblock fields to debugfs's
set_super_value.

Add support for printing "mostly-printable" extended attributes in
Debugfs.

Add support for the -M option to fsck, which causes it to ignore
mounted filesystem.

Fix uuidd so that it creates the pid file with the correct pid number.
(Addresses Sourceforge Bug: #1893244)

Fix various gcc -Wall warnings.

Update Czech, Dutch, Polish, Swedish, and Vietnamese translations

Fixed spelling mistakes, typos, and otherwise clarified man pages.
(Addresses Sourceforge Patch: #1399325)


Programmer's notes:
-------------------

Fixed various Debian packaging issues --- see debian/changelog for
details.  (Addresses Debian Bug: #466929)

Add new flag EXT2_FLAG_NONFREE_ON_ERROR ext2fs_open2() which returns a
partially filled-in filesystem object on an error, so that e2fsck can
print more intelligent error messages.

Add a new function e2p_edit_feature2() which allows the caller to
specify which feature flags are OK to set or clear, and which returns
more specific information about feature flags which are not allowed to
be set/cleared.

Set the C locale in the test_script driver since it uses [A-Za-z].
(Addresses Sourceforge Bug: #1890526)

Use fcntl locking instead of lockf in libuuid since Cygwin doesn't
support lockf().

Change configure.in to avoid using the 'dc' command unless it is
absolutely needed. (i.e., when using parsing a WIP-style version
number)  (Addresses Sourceforge Bug: #1893024)

Add portability checks to support compilation under DJGPP.

Update to the latest samba tdb code before the LGPLv3 change, which
fixes a realloc() leak on failure.

Fix memory leak in ext2fs_alloc_block().

Fix makefile dependency issues for various install targets.
(Addresses-Sourceforge-Patches: #1903484, #1903466, #1903456)

Improve descriptions for the r_move_itable and r_resize_inode tests.


E2fsprogs 1.40.6 (February 9, 2008)
===================================

Add support for returning labels for UDF filesystems in the blkid
library.

Fix bug in the blkid library where cached filesystems was not being
flushed when opening USB devices returned the error ENOMEDIUM.
(Addresses Debian Bug: #463787)

Added logic to the blkid library to automatically choose whether a
filesystem should be mounted as ext4 or ext4dev, as appropriate.

Allow tune2fs to set and clear the test_fs flag on ext4 filesystems.

Fix a bug in e2fsck which caused it to core dump if e2fsprogs had been
configured with --enable-jbd-debug.

Document the BLKID_FILE environment variable in the libblkid man page

Programmer's Notes:
-------------------

Update e2fsprogs translation template and Vietnamese and Czech translations

Fixed various Debian packaging issues --- see debian/changelog for
details.  (Addresses Debian Bugs: #436058)

Don't try to create $DESTDIR/etc/init.d as part of make install as we
don't install the init.d script (and it's not the recommended way to
start uuidd anyway).  (Addresses Sourceforge Bug: #1885085)

Use thread local storage to fix a theoretical race condition if two
threads try to format an unknown error code at the same time in the
com_err library.


E2fsprogs 1.40.5 (January 27, 2008)
===================================

Fix a potential overflow big in e2image if the device name is too
long.

Mke2fs will now create new filesystems with 256 byte inodes and the
ext_attr feature flag by default.  This allows for much better future
compatibility with ext4 and speeds up extended attributes even on ext3
filesystems.

Teach e2fsck to ignore certain "safe" filesystem features which are
set automatically by the kernel.  Having the kernel set features
automagically behind the user's back is a bad idea, and we should try
to break the kernel of this habit, especially for the newer ext4
feature flags.  But for now, we will try to avoid needless full checks
which can annoy users who are doing fresh installs.

Add support in tune2fs and mke2fs for making a filesystem as being "ok
to be used with test kernel code".  This will be needed for using test
filesystems with the latest ext4 kernel code.

Change e2fsck -fD so that it sorts non-htree directories by inode
numbers instead of by name, since that optimizes performances much
more significantly.  (Addresses-Sourceforge-Feature-Request: #532439)

If e2image fills the disk, fix it so it exits right away instead of
spewing large numbers of error messages.
(Addresses-Sourceforge-Feature-Request: #606508)

If ftruncate64() is not available for resize2fs, let it use ftrucate()
instead, but have it check to see if the size would get truncated, and
skip calling ftruncate in that case.

Add support for detecting HFS+ filesystems in the blkid library.

Add support in the blkid library for ext4/ext4dev filesystems.

Fix a bug in blkid where it could die on a floating point exception
when presented with a corrupt reiserfs image.

Fix blkid's handling of ntfs UUID's so that leading zeros are printed
such that UUID string is a fixed length.

Add sample python bindings for the uuid library to the contrib
directory.  (Addresses-Sourceforge-Patches: #778817)

Fix debugfs's 'lsdel' command so it uses ext2fs_block_iterate2 so it
will work with large files.  (Addresses Sourceforge Feature Request:
#1257500 and Sourceforge Support Request: #1253511)

Allow the debugfs 'undel' command to undelete an inode without linking
it to a specific destination directory, since this might require
allocating blocks that could overwrite some yet-to-be-recovered
deleted files.  (Addresses-Sourceforge-Feature-Request: #967141)

Update Swedish translation from the Translation Project.

Programmer's Notes:
-------------------

Fix configure handling of --sbindir (which should rarely be used, but
someone did complain, so let's fix it).  (Addresses Sourceforge Bug:
#498381)

Updated e2fsprogs.spec file to include a new uuidd package

Use pkg-config to determine where to find the devmapper library so we
can find out where it is located on different distributions.

Fix Makefile race so that "make -j3 distclean" works correctly

Fix portability problems on non-Linux/non-Hurd/non-Masix systems,
especially on MacOS X systems.  (Addresses Sourceforge Bugs: #1861633,
#1819034, #1863819)

Fixed spelling mistakes, typos, and otherwise clarified man pages.

Fixed various Debian packaging issues --- see debian/changelog for
details.  (Addresses Debian Bugs: #459403, #459475, #459614)

Remove the --enable-dynamic-static configure option, and build e2fsck
dynamically by default.  If the user wants e2fsck.static, he/she will
need to build it via "make -C e2fsck e2fsck.static"

Fix various build warnings due to missing prototypes.
(Addresses Sourceforge Patch: #1861663, #1861659)


E2fsprogs 1.40.4 (December 31, 2007)
====================================

Improve time-based UUID generation.  A new daemon uuidd, is started
automatically by libuuid if necessary.  This daemon is setuid to allow
updates to /var/lib/libuuid, so the clock sequence number can be
stored and so if the clock is set backwards, it can be detected.
(Addresses Sourceforge Bug: #1529672, Addresses Red Hat Bugzilla:
#233471)

Filter out the NEEDS_RECOVERY feature flag when writing out the backup
superblocks.  This avoids e2fsck from concluding that a full
filesystem check is required before backing up the superblock due to
changes in the feature flags.  (Addresses Debian Bug: #454926)

Fix fsck to only treat the '#' character as a comment at the beginning
of the line in /etc/fstab.  Otherwise fstabs for the fuse filesystem
will cause fsck to issue an bogus warning message.
(Addresses Gentoo bug: #195405, Addresses Sourceforge bug: #1826147)

Format control characters and characters with the high eighth bit set
when printing the contents of the blkid cache, to prevent filesystems
with garbage labels from sending escape sequences to the user's screen
that might, for example place it in graphics mode.  (Addresses Ubuntu
Bug: #78087)

Fix sign-extension problem on 64-bit systems in in the com_err
library.  (Addresses Sourceforge Bug: #1809658)

Avoid division by zero error when probing an invalid FAT filesystem in
the blkid library.  (Addresses Sourceforge Bug: #1831627)

Update Dutch, Polish, and Vietnamese translations from the Translation
Project.  Remove the Rwandan translation upon advice of the
Translation Project.

Programmer's Notes:
-------------------

Fix the libss "make check" regression test so that it works if the
current directory is not in the user's path or if the libss shared
library is not installed.  (Addresses Sourceforge Bug: #1848974)

Fixed spelling mistakes, typos, and otherwise clarified man pages.
(Addresses Debian Bugs: #444883, #441872)

Fixed various Debian packaging issues --- see debian/changelog for
details.  (Addresses Debian Bugs: #437720, #451172, #458017)

Fix build failure on non-Linux/non-Hurd/non-Masix systems.
(Addresses Sourceforge Bug: #1859778)

Fix Hurd portability issues.  (Addresses Debian Bug: #437720)


E2fsprogs 1.40.3 (December 5, 2007)
===================================

Fix a potential security vulnerability where an untrusted filesystem
can be corrupted in such a way that a program using libext2fs will
allocate a buffer which is far too small.  This can lead to either a
crash or potentially a heap-based buffer overflow crash.  No known
exploits exist, but main concern is where an untrusted user who
possesses privileged access in a guest Xen environment could corrupt a
filesystem which is then accessed by the pygrub program, running as
root in the dom0 host environment, thus allowing the untrusted user to
gain privileged access in the host OS.  Thanks to the McAfee AVERT
Research group for reporting this issue.  (Addresses CVE-2007-5497.)

Fix heuristics in blkid which could cause a disk without partitions to
be incorrectly skipped when a loopback device is present.  (Addresses
Red Hat Bugzilla #400321.)

Fix e2image so that in raw mode it does not create an image file which
is one byte too large.

Change mke2fs's usage message so it recommends the preferred -E option
instead of the deprecated -R option.

Enhance the blkid library so it will recognize squashfs filesystems.
(Addresses Red Hat Bugzilla #305151.)

Enhance e2fsck so it will force the backup superblocks to be backed up
if the filesystem is consistent and key constants have been changed
(i.e., by an on-line resize) or by e2fsck in the course of its
operations.

Enhance blkid's detection of FAT filesystems; so that USB disks with
only a single bootable partition will not get missed.

E2fsck will no longer mark a filesystem as invalid if it has time
errors (i.e., if superblock mount time or last write time is in the
future) and the user refuses to fix the problem.

The Ubuntu init scripts don't properly set the system time correctly
from hardware clock if the hardware clock is configured to tick local
time instead of GMT time.  Work around this as best as we can by
providing an option, buggy_init_scripts, in /etc/e2fsck.conf which can
be set on Ubuntu systems.  (Addresses Debian Bug #441093, and Ubuntu
Bug #131201.)

Fix fsck to ignore /etc/fstab entries for bind mounts.  (Addresses Red
Hat Bugzilla #151533.)

Fix e2fsck so that if the superblock is corrupt, but still looks
vaguely like an ext2/3/4 superblock, that it automatically tries to
fall back to the backup superblock, instead of failing with a hard
error.

Make the e2fsprogs program more robust so that they will not crash
when opening a corrupt filesystem where s_inode_size is zero.

Change e2fsck so it uses sscanf() instead of atoi() so it non-numeric
arguments are detected as such and the parse error is reported to the
user.   (Addresses Debian Bug #435381.)

Change e2fsck so it will not complain if a file has blocks reallocated
up to the next multiple of a system's page size.

Fix bug in ext2fs_check_desc() which will cause e2fsck to complain
about (valid) filesystems where the inode table extends to the last
block of the block group.  (Addresses Red Hat Bugzilla #214765.)

Fix a bug in ext2fs_initialize() which causes mke2fs to fail while
allocating inode tables for some relatively rare odd disk sizes.
(Addresses Red Hat Bugzilla #241767.)

Add Catalan translation and update Dutch and Swedish translations
from the Translation Project.

Fix big-endian byte-swapping bug in ext2fs_swap_inode_full().  We
still had an issue when trying to figure out whether we need to
byte-swap fast symlinks that contained extended attributes.

Fixed spelling mistakes, typos, and otherwise clarified man pages.
(Addresses SourceForge Bug #1821333.)


Programmer's Notes:
-------------------

Fix mke2fs tests to avoid needing any significant ^M (CR) characters

Add "make check" to the RPM spec file

Fix "make install" and 'make uninstall" in misc/Makefile.in so that it
works correctly when the prefix is not the root directory.

Fix the resize2fs tests, r_move_itable and r_resize_inode, so they
clean up after themselves by deleting the test.img temporary file
after completing the test.

Fixed a corner case bug ext2fs_unlink() when trying to delete the
first directory entry in a directory block and the last directory
entry in the previous directory block is not in use.  Fortunately
ext2fs_unlink() is only used by debugfs and e2fsck, and in e2fsck in
places where it is extremely unlikely to run into this corner case.

Fix missing dependency which would cause parallel builds to fail.
(Addresses Sourceforge Bug #1842331.)

Fix a build error on newer gcc caused by lib/ext2fs/ismounted.c
calling open(O_CREATE) with a missing mode parameter.

Fix the test_ss.c in lib/ss so it can be used as an example
application program for the library as well as a regression test
suite.

Fix ext2fs_dblist_dir_iterate() so that error codes and abort codes
are properly passed back up through the call stack.

E2fsprogs 1.40.2 (July 12, 2007)
================================

A recent change to e2fsck_add_dir_info() to use tdb files to check
filesystems with a very large number of filesystems had a typo which
caused us to resize the wrong data structure.  This would cause a
array overrun leading to malloc pointer corruptions and segfaults.
Since we normally can very accurately predict how big the the dirinfo
array needs to be, this bug only got triggered on very badly corrupted
filesystems.

Fix a bug in e2fsck which caused it to incorrectly salvage
directories when the last entry's rec_len is bogusly too big.  This
resulted in a nonsense filesystem corruption to be reported, and
required a second run of e2fsck to fully fix up the directory.

Update tune2fs man page to include more discussion of reserved blocks
(Addresses Launchpad bug #47817)

Update Turkish, Polish, Dutch, and Vietnamese PO files from the
Translation Project


E2fsprogs 1.40.1 (July 7, 2007)
===============================

Fix bug which could cause libblkid to loop forever.  When revalidating
a partition where there is obsolete information in /etc/blkid.tab, we
end up freeing a the type tag without clearing dev->bid_type, causing
blkid_verify() to loop forever.  (Addresses Debian Bug: #432052)

The Turkish translation has a bug in it where it has the translation
of "E@e '%Dn' in %p (%i)" to "E@E".  This causes @E to be expanded at
@E, recursively, forever, until the stack fills up and e2fsck core
dumps.  We fix this by making e2fsck stop @-expansions after a
recursive depth of 10, which is far more than we need.
(Addresses Sourceforge Bug: #1646081)

Compile the default mke2fs.conf into mke2fs program.  People are
getting surprised by mke2fs creating filesystems with different
defaults than earlier versions of mke2fs if mke2fs.conf is not
present.  So we now create a built in version of mke2fs.conf file
which is used by mke2fs if the /etc/mke2fs.conf is not present.
(Addresses SourceforgeBug: #1745818)

Improve the config/parse_types.sh helper script.  Fix a potential
security problem if e2fsprogs is built as root (as Gentoo does!).  In
addition fix the script and how it is called from the configure script
so that it does the right thing when cross-compiling.  (Fixes Gentoo
bug: #146903)

Update Vietnamese, French, and Dutch PO files from the Translation
Project.  Also created a new e2fsprogs.pot file for translator.

Fix bogus strip permission errors when building under Debian.  When
building the e2fsprogs dpkg's, the dh_strip command emits a large
number of error messages caused by the permissions not being right.
So run dh_fixperms before running dh_strip.

Programmer's Notes:
-------------------

Add new function: profile_set_default().  This function sets the value
of the pseudo file "<default>".  If the file "<default>" had
previously been passed to profile_init(), then def_string parameter
will be parsed and used as the profile information for the "<default>"
file.

Fix mk_cmds's error reporting so that it is unambiguous that it is the
mk_cmds script which is generating the error.  (Obviates Gentoo patch:
e2fsprogs-1.32-mk_cmds-cosmetic.patch)

Fix the test suite to use LC_ALL instead of LANG.  LC_ALL is the "high
priority" environment variable that overrides all others, where as
LANG is the lowest priority environment variable.  If LC_ALL is set, it
doesn't matter whether LANG, LANGUAGE, LC_COLLATE, LC_MESSAGES, and
the all the rest are set.  This will assure that the locale when
running the test suites is the "C" locale.  (Obviates Gentoo patch:
e2fsprogs-1.38-tests-locale.patch)


E2fsprogs 1.40 (June 29, 2007)
==============================

Fix divide by zero error in blkid's NTFS probing logic.

Add new blkid -g option which causes the blkid cache to be garbage
collected.

Fix a bug in libblkid which could cause the internal field bid_type to
become corrupted.  Fortunately bid_type isn't used much, and bid_label
and bid_uuid is only used by debugging code, so the impact of this bug
was very minor.

Mke2fs will now store the RAID stride value when a filesystem is
created with a requested RAID stride, and then use it automatically in
resize2fs.

Mke2fs has a sanity check added to make sure (inode_size * num_inodes)
isn't too big.  In some cases Lustre users have tried specifying an
inode size of 4096 bytes, while keeping an inode ratio of one inode
per 4096 bytes.

Improve sanity check in e2fsck's algorithm for finding a backup
superblock, so that it won't accidentally find a superblock that was
located in the journal, and then later reject it as being not a valid
backup superblock.

Fix e2fsck get_size logic so that it will work with the Linux floppy
driver.  The Linux floppy driver is a bit different from the other
block device drivers, in that if the device has been opened with
O_EXCL, it disallows another open(), even if the second open() does
not have the O_EXCL flag.  (Addresses Debian Bug: #410569)

Fix error checking of badblock's last-block and start-block arguments.
(Addresses Debian Bug: #416477)

Fix e2fsck so that it doesn't overwrite the backup superblocks when
recovering a journal until the master superblock has been confirmed as
being sane.

Change the blkid library to be much more paranoid about concluding
that a partition contains an NTFS filesystem, and fetch the UUID and
LABEL information from NTFS filesystems.  (Addresses Launchpad Bug:
#110138)

Factor out the code which sets the default journal size and move it
into libext2fs.

Enhance e2fsck so it will recreate the ext3 journal if the original
journal inode was cleared to due it being corrupt after finishing the
filesystem check.

Fix e2fsck so that it updates the journal inode if it is corrupted and
the backup journal information from the superblock was successfully
used to recover the filesystem.

Fix e2fsck so that it checks all of the blocks in the journal inode
for validity.  The original code only checked the direct blocks to
make sure the journal inode was sane.  Unfortunately, if some or all
of the indirect or doubly indirect blocks were corrupted, this would
not be caught.

Add support in blkid to detect LUKS encrypted partitions.

Add extra sanity checks for extended attributes in the case where the
size is zero but the offset is very large.

Fix byte-swapping issues for large inodes in ext2fs_read_inode_full()
and ext2fs_get_next_inode_full().

Clarify the copyright licenses used by the various libraries in
the top-level COPYING  file (Red Hat Bugzilla: 166058)

Make mke2fs's defaults when /etc/mke2fs.conf doesn't exist more sane.

Fix mke2fs and debugfs to support large (> 16 bit) uid's and gid's.

Remove check in e2fsck which requires EA's in inodes to be sorted;
they don't need to be sorted, and e2fsck was previously wrongly
clearing unsorted EA's stored in the inode structure.

Allow mke2fs or tune2fs to create a substantially larger journal (up
to 10,240,000 blocks).

Fix MD superblock detection, and make sure the correct UUID is
reported from the MD superblock.

Fix a signed vs. unsigned bug in debugfs.

Enhance debugfs's date parser so that it accepts integer values.

Fix e2fsck's pass1c accounting so it doesn't terminate too early if a
file with multiply claimed blocks is hard linked. or not at all if the
root directory contains shared blocks

Enhance debugfs so it can modify the block group descriptors using the
command set_block_group_descriptor.

Improve e2fsck's reporting of I/O errors so it's clearer what it was
trying to do when an error happens

Fix a bug in in how e2fsprogs byte swaps inodes containing fast
symlinks that have extended attributes.  (Addresses Red Hat Bugzilla:
#232663 and LTC Bugzilla: #27634)

Fix potential file descriptor leak in ext2fs_get_device_size() in an
error case.

Add libreadline.so.5 support to libss.

Improve badblocks -n/-w exclusive usage message.

Fix dump_unused segfault in debugfs when a filesystem is not open

Fix memory leak in blkid library.  (Addresses Debian Bug: #413661)

Allow the debugfs lcd command to work w/o a filesystem being open.
(Addresses LTC Bugzilla #27513)

Fix e2fsck to clear i_size for special devices with a bogus i_blocks
field on the first pass.

Fix e2fsck to set the file type of the '..' entry when connecting
a directory to lost+found.  (Addresses Lustre Bug: #11645)

Enhance e2fsck to recover directories whose modes field were corrupted
to look like special files.  This is probably only useful in
artificial test cases, but it will be useful if we ever do the "inodes
in directory" idea for ext4.

Allow debugfs to dump (and rdump) > 2GB files.  (Addresses Debian Bug:
#412614)

Fix resize2fs parsing of size parameter (in sector units).  This was
actually a bug in libe2p's parse_num_blocks() function.  When handling
the 's' suffix, it was ignoring the blocksize information passed in
from the caller and always interpreting the number in terms of a 1k
blocksize.  (Addresses Debian Bug: #408298)

There was a floating point precision error which could cause e2fsck to
loop forever on really big filesystems with a large inode count.
(Addresses Debian Bug: #411838)

Fix memory leak in ext2fs_write_new_inode()

Add support for using a scratch files directory to reduce e2fsck's
memory utilization on really big filesystems.  This uses the TDB
library.  See the [scratch_files] section of the e2fsck.conf man page
for more details.

Fixed type-punning bug which caused dumpe2fs to crash on the Arm
platform (Addresses Debian Bug: #397044)

Add explanatory message to badblocks that -n and -w are mutually exclusive
(Addresses Debian Bug: #371869)

Allow debugfs and dumpe2fs to support fs features under development.

Add support for the new flag EXT2_FLAG_SOFTSUPP_FEATURES flag to
ext2fs_open() , which allows application to open filesystems with features
which are currently only partially supported by e2fsprogs.

Allow unix_io to support offsets greater than 2G (Addresses
SourceForge Bug: #1547922)

Fixed overflow and signed/unsigned problems caused by the number of
blocks or inodes exceeding 2**31 or being close to 2**32-1.

Add support for unsigned directory hash calculations with hints in the
superblock to fix cross-architectural portability for htree
directories with filenames where the high 8th bit is set.  (Addresses
Debian: #389772)

Fix resize2fs so that it gives user-intelligible error messages if the
filesystem or the kernel does not support on-line resizing.
(Addresses Debian Bug: #380548)

Require mke2fs -F -F for really dangerous operations, since -F is
needed for less dangerous operations such as creating filesystems
images in regular files, or creating filesystems on whole block
devices.  These relatively innocuous usages should NOT be confused
with running mke2fs on an apparently-mounted or in-use filesystem.

Allow the default inode size to be specified into the mke2fs.conf
file.

Make the smallest default journal size is big enough so that on-line
resizing should always work.

Fix silly spelling error in e2fsck.  (Addresses SourceForge bug:
#1531372)

Fix debugfs coredump when lsdel is run without an open filesystem
(Addresses Debian Bug: #378335)

Fix debugfs display bug us that bytes that have the high bit set are
displayed as "ec" instead of "ffffffec".

Add support in lsattr so it will display the EXT4_EXTENTS_FL flag.

Device mapper scanning wasn't working in the blkid library because the
pathnames had an extra "/dev" when they were being probed.

Add GFS/GFS2 support to the blkid library.

Fix blkid support of empty FAT filesystem labels.

Avoid recursing forever (or for a long time) when the blkid library
searches for a device and there are symlinks to directories in /dev.

Avoid unaligned halfword access in blkid when accessing FAT
superblocks, as this will cause Sparc/Solaris systems to throw a
SIGBUS error.

The latest devmapper libraries requires pthreads, add -lpthreads to
the static link libraries for e2fsck.static if devmapper is enabled.
(Addresses Debian bug: #388718)

Improve the (non-installed, for experts only) findsuper program by
printing the uuid and label from the superblocks, as well as the
starting and ending offsets of the filesystem given the information in
the superblock.  Omit by default printing superblocks that are likely
found in located in an ext3 journal unless an explicit -j option is
given.

Updated Spanish, French and Dutch translations and added Catalan
translation.  (Addresses Debian bug: #411562)

Use FreeBSD's DIOCGMEDIASIZE and DIOCGDINFO ioctls if available when
determining a partition's size, since binary searching to determine
the device doesn't work on FreeBSD.

Documentation about UUID's is available in enough places, and it's
awkward to deal with debian-legal's insanities.  So I'm caving in the
"more-lunatic-than-RMS" wing of Debian by removing RFC-4122 so we
don't have do the dfsg tarball.  Also remove the rule that only tried
to install RFC-4122 on Ubuntu, since Ubuntu seems to want to fetch
e2fsprogs exclusively from Debian.  (Addresses Debian Bug: #407107)

Fix the info-dir line so that the menu name does not contain a .info
prefix.  First of all, it's ugly, secondly, it causes the install-info
command to fail to remove the com_err info file from the
/usr/share/info/dir file when the comerr-dev package is removed and
purged.  (Addresses Debian Bug: #401711)

Fixed spelling mistakes, typos, and otherwise clarified man pages.
(Addresses Debian Bug: #369761, #373004, #379695)

Fixed various Debian packaging issues --- see debian/changelog for
details.  (Addresses Debian Bugs #389554, #390664, #413208, #419605,
#408352, #415560, #399155)


Programmer's notes:
-------------------

E2fsck now supports the %It expansion when printing a problem report.
It will print the type of the inode in the problem context.

Fix misc/Makefile.in so that it builds even if e2fsck hasn't been built yet
(Addresses Sourceforge Bug: #1565561)

Remove unused variables and other lint/gcc -Wall cleanups

Add check to ext2fs_get_device_size() so it will return EFBIG for for
filesystems contained in regular files where the filesystem image size
is returned by stat64().

Set local environment variables to C so mk_cmds and compile_et always
work.  (Addresses SourceForge Bug: #1532177)

Added the 64-bit byte swapping function ext2fs_swab64().

Added two new helper functions to prevent 2**31/2**32-1 overflow
problems: ext2fs_div_ceil() and e2p_percent().

Create new ext2fs library inline functions ext2fs_group_first_block()
and ext2fs_group_last_block() in order to calculate the starting and
ending blocks in a block group.

Create the generated files read-only to remind developers not to edit them.

Add support for autoconf 2.60 (with backwards compatibility for older
versions of autoconf).

Added an "make rpm" target to top-level Makefile

Added various FreeBSD portability fixes.

Exclude mercurial files from the RPM build tree to speed up copy/build.

Use root_sysconfdir to define the locations of mke2fs.conf and
e2fsck.conf instead of using a hard-coded /etc pathname.

Prevent e2fsck.h and ext2_ext_attr.h from getting included multiple times.

Fixed "make clean" in blkid's Makefile.in file from removing tst_*.c files.

If diff -u is supported, use it to report test failures.

Updates/improvements to RPM spec file

Add on-disk format definitions for the following new features:
EXT4_FEATURE_RO_COMPAT_HUGE_FILE, EXT4_FEATURE_RO_COMPAT_GDT_CSUM,
EXT4_FEATURE_RO_COMPAT_DIR_NLINK, EXT4_FEATURE_INCOMPAT_64BIT,
EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE

Add a new make target "checked-failed" in the tests directory which
reruns any failed tests

Update draft-leach-uuids-guids-01.txt with rfc4122.txt

Fix miscellaneous bugs reported by Coverity: Dead code, potential nul
pointer dereferences, memory leaks, etc.  None were security-critical
problems.

Fix up usage and decrement error messages in the test_icount program

Add debugging code to the com_err library; if the environment variable
COMERR_DEBUG is set to 1, print out debugging messages as error tables
are added and removed from the com_err library.  If the
COMERR_DEBUG_FILE environment variable is set (and the process is not
setuid) the debugging messages may be redirected to a file.

Change all of the e2fsprogs programs to use the newer add_error_table()
and remove_error_table() interfaces instead of the much older
initialize_*_error_table() function.

Add TDB support into the ext2fs library.  This allows us to have a
guaranteed library we can count on always being present so we can
store data in an on-disk database.

Add support for using TDB to store the icount data, so we don't run out
of memory when checking really large filesystems.

Change the regression test suite so that it skips empty test directories.

Define the l_i_iversion field in ext2_inode.  The l_i_version field is
now defined from the old l_i_reserved1 field in the ext2 inode.  This
field will be used to store high 32 bits of the 64-bit inode version
number.

Add Makefile production rule for e2fsprogs.spec in case it gets
deleted.

Add new function profile_get_uint() to allow for a clean way to fetch
unsigned integers from the context.

Add test to make sure the ext2 superblock structure is 1024 bytes.

Fix typo in name of f_dup4 regression test

Add new function blkid_gc_cache() which performs a garbage collection
pass on the /etc/blkid.tab file.

The ext2fs_open() function now sets EXT2_FLAG_MASTER_SB_ONLY.  In
general, only e2fsck (after the filesystem is clean), tune2fs, and
resize2fs should change the backup superblocks by default.  Most
callers of ext2fs_open() should not be touching any superblock fields
which require the backups to be touched.

Add new function to libext2fs, ext2fs_default_journal_size(), which
returns the default journal size.