Andreas Dilger [Tue, 22 May 2007 20:04:51 +0000 (16:04 -0400)]
mke2fs: Add check to make sure inode_size * num_inodes isn't too big
A quick patch to sanity check the inode ratio vs the inode size. 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. I'm
sure more people will do this now that large inodes are available in
ext4 and documented in e2fsprogs.
Signed-off-by: Andreas Dilger <adilger@clusterfs.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Kalpak Shah [Tue, 8 May 2007 05:07:30 +0000 (01:07 -0400)]
Remove check in e2fsck which requires EA's in inodes to be sorted
This patch removes a code snippet from check_ea_in_inode() in pass1 which checks
if the EA values in the inode are sorted or not. The comments in fs/ext*/xattr.c
state that the EA values in the external EA block are sorted but those in the
inode need not be sorted. I have also attached a test image which has unsorted
EAs in the inodes. The current e2fsck wrongly clears the EAs in the inode.
Eric Sandeen [Tue, 8 May 2007 04:10:54 +0000 (00:10 -0400)]
Fix mke2fs and debugfs to support large (> 16 bit) uid's and gid's
Mke2fs is supposed to set the uid/gid ownership of the root directory when
a non-rooot user creates the filesystem. This wasn't working correctly
if the uid/gid was > 16 bits. In additional, debugfs wasn't displaying
large uid/gid's correctly. This patch fixes these two programs.
Signed-off-by: Eric Sandeen <sandeen@redhat.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Eric Sandeen [Tue, 8 May 2007 03:38:14 +0000 (23:38 -0400)]
Fix mke2fs defaults when /etc/mke2fs.conf doesn't exist
One of our testers filed a bug that said "mkfs.ext3 is much slower
when mke2fs.conf is missing..."
This is because the shipped defaults in mke2fs.conf do not match the
shipped defaults in the mkfs code itself; he wound up making a 1k
block filesystem on a very large block device, for example.
So - How about this patch, to bring them back into line? Which makes
me wonder; having "defaults" in 2 different places is bound to get out
of sync; should we instead generate both code & config file defaults
(and maybe man page defaults) from a common source?
Anyway, here's a patch to bring mke2fs.conf and mke2fs.c into line for
current defaults...
Theodore Ts'o [Mon, 7 May 2007 00:09:42 +0000 (20:09 -0400)]
Debian udeb dependency cleanup
From: Frans Pop <elendil@planet.nl>
Please apply the patch below with your next upload of e2fsprogs.
It removes Provides/Replaces: in the udebs which are no longer needed now
that udeb library dependencies are generated correctly. This has been the
case for a while now, but we've postponed this cleanup until after the
Etch release.
Jim Garlick [Fri, 20 Apr 2007 12:14:27 +0000 (08:14 -0400)]
e2fsck: Count the root directory as found if it has shared blocks in pass1c
Another small bug I think: if the root directory contains shared
blocks, e2fsck pass1c search_dirent_proc() will be looking for
one more containing directory than it will ever find, and thus
loses an opportunity to terminate early.
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.
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.
Jim Garlick [Wed, 11 Apr 2007 02:55:09 +0000 (22:55 -0400)]
e2fsck: pass1c terminates early if hard links
I think this is a small buglet in e2fsck: 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.
Signed-off-by: Jim Garlick <garlick@llnl.gov> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The second part of UUID was copied to a wrong place in the buffer.
Now the UUID shown by blkid is the same as shown by /lib/udev/vol_id
(at least with udev-108), but is not in the same form as used by mdadm
(which prints UUID as 4 32-bit words and uses different endiannes).
Signed-off-by: Sergey Vlasov <vsu@altlinux.ru> Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>
blkid: Fix MD 0.90 superblock detection on little endian systems
MD 0.90 superblock format is host endian - need to check for bith big
endian and little endian magic. Without this change MD components
created on little endian systems were not detected as such, which
could then lead to false positives when detecting filesystems.
Signed-off-by: Sergey Vlasov <vsu@altlinux.ru> Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>
Andreas Dilger [Sat, 7 Apr 2007 19:39:50 +0000 (15:39 -0400)]
Allow the journal size to up to 10,240,000 blocks
Increase the maximum size of the journal to 100 times the previous
maximum, but add a restriction that it can be no more than half the size
of the filesystem.
Signed-off-by: Andreas Dilger <adilger@clusterfs.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Enable e2fsck to use the tdb library to store the dirinfo abstraction
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.
Enable e2fsck to use the tdb library for the inode count abstraction
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.
Jim Garlick [Fri, 6 Apr 2007 12:50:15 +0000 (08:50 -0400)]
Fix memory leak in ext2fs_write_new_inode()
The following patch addresses a memory leak in libext2fs
that occurs when using ext2fs_write_new_inode() on a file system
configured with large inodes.
Signed-off-by: Jim Garlick <garlick@llnl.gov> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Rework e2fsck's dirinfo abstraction to be more friendly for databases
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.
Fix potential infinite loop in e2fsck on really big filesystems
Prevent floating point precision errors on really big filesystems from
causing the search interpolation algorithm in the icount abstraction
from looping forever.
resize2fs (libe2p): 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.
Theodore Ts'o [Sat, 31 Mar 2007 23:18:24 +0000 (19:18 -0400)]
Fix e2fsck to set the filetype of '..' when connecting a dir, to lost+found
If there is an orphaned inode whose '..' entry is pointing at a special
file, the filetype of the '..' entry will set to the type of the special
file. When the orphaned directory is reconnected to /lost+found, the
filetype of the '..' field is not reset to EXT2_FT_DIR, so a second
e2fsck is required to repair the filesystem.
We address this situation by setting the filetype of '..' when we
reconnect the inode to /lost+found.
Theodore Ts'o [Sat, 31 Mar 2007 22:56:09 +0000 (18:56 -0400)]
Fix edge case when i_size doesn't get fixed until second e2fsck run
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 in
i_blocks[4..15], so there's no point doing the check for an illegal
i_blocks value.
Brian Behlendorf [Wed, 28 Mar 2007 13:48:07 +0000 (09:48 -0400)]
[COVERITY] Fix memory leak in fsck on error paths
The memory allocated by inst is not reclaimed. There also was a
call to exit that coverity did not catch the resource leak. This
might not really be a big issue since the memory will be freed when
fsck exits, but it should be done anyway imho.
Coverity ID: 32: Resource Leak
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Brian Behlendorf [Wed, 21 Mar 2007 21:34:47 +0000 (17:34 -0400)]
[COVERITY] Fix memory leak in libss (ss_execute_line)
Fix a memory leak by freeing the argv[] array if ss_parse_line returns 0
for argc 0 (which will happen if the user his return and sends an empty
line to the application).
Potentially need to free argv before early return since it was allocated
memory. Need to be careful since it may be possible for ss_parse() to have
freed the memory allocated to it if it detects an unbalanced set of quotes
passed to it.
Coverity ID: 21: Resource Leak
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Brian Behlendorf [Wed, 21 Mar 2007 19:09:15 +0000 (15:09 -0400)]
[COVERITY] Avoid static buffer overruns in debugfs
Add an extra byte to EXT2_NAME_LEN in the static allocation for the
required trailing null. This allows filenames up to the maximum
length of EXT2_NAME_LEN withover an overrun.
Brian Behlendorf [Mon, 19 Mar 2007 12:58:47 +0000 (08:58 -0400)]
[COVERITY] Add missing NULL check to e2fsck_get_dir_info()
It is possible that e2fsck_get_dir_info() returns a NULL pointer.
We do not want to blow up when dereferencing p. It seems to be
more sane/safe to call fix_problem(ctx, PR_3_NO_DIRINFO, pctx)
if p is NULL at this point since we do not have any DIRINFO
for pctx->ino.
Also fix another (already existing) error check for
e2fsck_get_dir_info() later in the function so that it reports the
correct inode number if the dirinfo information is not found for
p->parent.
(Both of these are "should-never-happen" internal e2fsck errors that
would indicate a programming bug of some kind.)
Coverity ID: 10: Null Returns
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Brian Behlendorf [Mon, 19 Mar 2007 12:52:10 +0000 (08:52 -0400)]
[COVERITY] Check for NULL return from dict_lookup() in e2fsck
The dict_lookup() function can potentially return a NULL dnode_t. It is
not checked in two places in the clone_file() function. Looks to be
safe to continue if n is NULL, so just print a warning message and
continue.
Coverity ID: 9: Null Returns
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Brian Behlendorf [Mon, 19 Mar 2007 12:39:32 +0000 (08:39 -0400)]
[COVERITY] Fix bad error checking for NULL parameter in ss library
Looks like flawed reasoning. Here if info_dir is NULL then you are
guaranteed to blow up since you will dereference it. It seems like the
correct thing to do here (what the code author meant to do) was to set
*code_ptr = SS_ET_NO_INFO_DIR if info_dir was NULL or if *info_dir was
an empty string (aka *info_dir == '\0').
Coverity ID: 8: Forward Null
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Brian Behlendorf [Mon, 19 Mar 2007 12:25:38 +0000 (08:25 -0400)]
[COVERITY] Fix dead code bug in mke2fs
If the fs_type is not specified and we are creating a journal device, to
use a fs_type of "journal"; this used to be the behavior before we added
support for the /etc/mke2fs.conf file, so let's fix it to restore the
old behavior.
Coverity ID: 4: Deadcode
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Theodore Ts'o [Sun, 18 Mar 2007 15:21:44 +0000 (11:21 -0400)]
[COVERITY] Allow blkid_dev_has_tag to check if a tag exists when value==NULL
blkid_dev_has_tag() will immediately return -1 (an error if value is
NULL. Thus at the test later on value cannot be NULL. There are two
possible ways to go about fixing this. The first would be to remove the
first NULL check for value. The second one would be to remove the
second check (and the deadcode).
I chose the second path because the functionality added is something
which a programmer could reasonably expect given the function name, and
it is highly unlikely any existing code is depending on the fact that
blkid_dev_has_tag() will return an error if value is NULL.
Theodore Ts'o [Sun, 4 Mar 2007 13:17:18 +0000 (08:17 -0500)]
Add versioned dependency in Debian's shlibs for libblkid
Add dependency to account for the new blkid function blkid_probe_all_new()
was added to e2fsprogs 1.39. The lack of this versioned dependency
caused a failure in rpc.mountd as detailed in Debian Bug #413057.
Theodore Ts'o [Tue, 26 Dec 2006 08:38:07 +0000 (03:38 -0500)]
Use the newer add/remove_error_table com_err interfaces
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.
Theodore Ts'o [Fri, 22 Dec 2006 18:38:38 +0000 (13:38 -0500)]
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.
projects / thirdparty / e2fsprogs.git / log
