As an object lesson in why autoreconf is fundamentally unsafe, the
newer version of nls.m4 no longer handles @MKINSTALLDIRS@. So add
this back, since our Makefiles depend on it.
mke2fs: add support to align hugefiles relative to beginning of the disk
Add the mke2fs.conf configuration option which causes the hugefiles to
be aligned to the beginning of the disk. This is important if the the
reason for aligning the hugefiles is to support hard-drive specific
features such as Shingled Magnetic Recording (SMR).
e2fsck: reopen the file system with saved flags after a journal replay
After a journal replay, we close and reopen the file system so that
any changes in the superblock can get reflected in the libext2fs's
internal data structures. We need to save the flags passed to
ext2fs_open() that we used when we originally opened the file system.
Otherwise we will end up not be able to repair a file system which
requires a journal replay and which has bigalloc enabled or which has
more than 2**32 blocks; e2fsck will abort with the error message:
fsck.ext4: Filesystem too large to use legacy bitmaps while trying to re-open
Addresses-Debian-Bug: 744953 Cc: Андрей Василишин <a.vasilishin@kpi.ua> Cc: Jon Severinsson <jon@severinsson.net> Cc: 744953@bugs.debian.org
mke2fs: prevent creation of unmountable ext4 with large flex_bg count
In mke2fs command, if flex_bg count is too large to filesystem blocks
count, unmountable ext4 which has the out of filesystem block offset
is created (Case1). Moreover this large flex_bg count causes an
unintentional metadata layout (bmap-imap-itable-bmap-imap-itable .. in
block group) (Case2).
To fix these issues and keep healthy flex_bg layout, disallow creating
ext4 with obviously large flex_bg count to filesystem blocks count.
Steps to reproduce:
(Case1)
1.
# mke2fs -t ext4 -b 4096 -O ^resize_inode -G $((2**20)) DEV 2130483
2.
# mount -t ext4 DEV MP
mount: wrong fs type, bad option, bad superblock on /dev/sdb4,
missing codepage or helper program, or other error
In some cases useful info is found in syslog - try
dmesg | tail or so
3.
# dumpe2fs DEV
...
Block count: 2130483
...
Flex block group size: 1048576
...
Group 65: (Blocks 2129920-2130482) [INODE_UNINIT]
Checksum 0x4cb3, unused inodes 8080
Block bitmap at 67 (bg #0 + 67), Inode bitmap at 1048643 (bg #32 + 67)
Inode table at 2129979-2130483 (+59)
^^^^^^^ 2130483 is out of FS!
65535 free blocks, 8080 free inodes, 0 directories, 8080 unused inodes
Free blocks:
Free inodes: 525201-533280
2.
# debugfs -R stats DEV
...
Block count: 786432
...
Flex block group size: 2147483648
...
Group 0: block bitmap at 193, inode bitmap at 194, inode table at 195
...
Group 1: block bitmap at 707, inode bitmap at 708, inode table at 709
...
Group 2: block bitmap at 1221, inode bitmap at 1222, inode table at 1223
...
We can set flex_bg count only up to 2^30 with profile
because get_int_from_profile can handle it to 2^31-1.
Add get_uint_from_profile to read unsigned int value
so that mke2fs with profile can handle up to 2^31 flex_bg same as -G option.
mke2fs -G option allows root user to set flex_bg count (power of 2).
However ext4 has bad metadata layout if we specify more than or equal to
2^32 to mke2fs -G, because of the 32bit shift operation
in ext2fs_allocate_group_table().
And the maximum block group count of ext4 is 2^32 -1 (ext4_group_t
s_groups_count), so diallow more than 2^32 flex_bg count.
Andreas Dilger [Fri, 28 Feb 2014 20:15:45 +0000 (13:15 -0700)]
mke2fs: handle flex_bg collision with backup descriptors
If a large flex_bg factor is specified and the block allocator was
laying out block or inode bitmaps or inode tables, and collides with
previously allocated metadata (for example the backup superblock or
group descriptors) it would reset the allocator back to the beginning
of the flex_bg instead of continuing past the obstruction.
For example, with "-G 131072" the inode table will hit the backup
descriptors in groups 1, 3, 5, 7, 9 and start interleaving with the
block and inode bitmaps. That results in poorly allocated bitmaps
and inode tables that are interleaved and not contiguous as was
intended for flex_bg:
Instead, skip the intervening blocks if there aren't too many of them.
That mostly keeps the flex_bg allocations from colliding, though still
not perfect because there is still some overlap with the backups.
This patch addresses the majority of the problem, allowing about 124k
groups to be layed out perfectly, instead of less than 4k groups with
the previous code.
Signed-off-by: Andreas Dilger <adilger@dilger.ca> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
mke2fs: enable lazy_itable_init on newer kernel by default
Currently is used did not specified lazy_itable_init option we rely on
information from ext4 module exported via sysfs interface. However if
the ext4 module is not loaded it will not be enabled even though kernel
might support it.
With this commit we set the default according to the kernel version,
however we still allow it to be set manually via extended option or be
enabled in case that ext4 module advertise that it supports this
feature.
Currently there are many uses of ext2fs_close() which might be wrong.
First of all ext2fs_close() does not set the ext2_filsys pointer to NULL
so the caller is responsible for clearing it, however there are some
cases there we do not do it.
Second of all very small number of users of ext2fs_close() actually
check the return value. If there is a problem in ext2fs_close() it will
not even free the ext2_filsys structure, but majority of users expect it
to do so.
To fix both problems this commit introduces a new helper
ext2fs_close_free() which will not only check for the return value and
free the ext2_filsys structure if the call to ext2fs_close2() failed,
but it will also set the ext2_filsys pointer to NULL.
Replace every use of ext2fs_close() in e2fsprogs tools with
ext2fs_close_free() - there is no real reason to keep using
ext2fs_close().
Signed-off-by: Lukas Czerner <lczerner@redhat.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Commit 2500ebfc89 (util: fix make dependencies for subst) broke cross
compilation because it unconditionally used config.h without setting a
includes path so that the config.h file could be found.
The proposed fix of adding the include path (such as was proposed at
http://patchwork.ozlabs.org/patch/355662/ or in Debian Bug #753375)
isn't really the right way to go, since the information in config.h is
for the target environment, and not the build environment. So using
config.h when building helper programs used as part of the build can
potentially cause more problems than it solves.
In general, build helpers must be written to be as portable as
possible, and to not require any autoconf defined #ifdef's whenever
possible. The subst program broke this rule to (1) address a Coverity
security complaint by using futimes(2) instad of utimes(2) if present,
and (2) to preserve the nanosecond portion of the file timestamp.
Oh, well. We won't be able to do the latter when cross compiling, and
as to the former, if an attacker has write access to your build tree
while you are building programs that will be run as root, you've got
bigger problems. :-)
Fix the problem that commit 2500ebfc89 was trying to address by
explicitly adding @DEFS@ to CFLAGS, so that -DHAVE_CONFIG_H is passed
to make depend. This fixes up the make depend without forcing the use
of config.h when cross-compiling.
configure.in: fix external libblkid test for static link
External libblkid needs -luuid when linking statically.
Also fix up the bogus other-lib parameter in the libuuid test;
$LIBUUID is the null string, so it doesn't do anything other than
obfuscate the use of AC_CHECK_LIB.
Andreas Dilger [Sat, 5 Jul 2014 03:34:10 +0000 (23:34 -0400)]
blkid,ext2fs: avoid name clash with __u{8,16,32,64}
Try to avoid name clashes with definitions of __u8, __u16, __u32,
and __u64 in userspace, in case other headers also define these
types. Define HAVE___{S,U}{8,16,32,64} preprocessor macros to
show that these types are already defined.
This would avoid the need to check for _BLKID_TYPES_H in ext2_types.h
and _EXT2_TYPES_H in blkid_types.h, but since older versions of these
headers did not use HAVE___U8 et.al. keep these checks around for now.
Report an error if there are no 64-bit types available. The code
will not compile if these are not available.
Signed-off-by: Andreas Dilger <adilger@dilger.ca> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Andreas Dilger [Wed, 11 Jun 2014 18:59:05 +0000 (12:59 -0600)]
blkid: remove unnecessary header and comment
The LIST_HEAD macro is not directly used in getsize.c, so
<sys/queue.h> is not needed at all, and could cause confusion at
some later point if the Linux-style list macros are ever used.
Build was verified on MacOS which defined HAVE_SYS_DISK_H true.
I manually inspected the sources for recent *BSD headers to check
if this was needed there or not. MacOS and FreeBSD <sys/disk.h>
do not use lists at all. NetBSD and OpenBSD <sys/disk.h> and all
of the <sys/mount.h> headers include <sys/queue.h> internally.
I used http://fxr.watson.org/fxr/source/sys/mount.h?v={OSTYPE}
as a reference, checking both old and new *BSD versions.
Signed-off-by: Andreas Dilger <adilger@dilger.ca> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Eric Sandeen [Sat, 5 Jul 2014 03:07:36 +0000 (23:07 -0400)]
e2fsprogs: add mount options to ext4.5
This is a straight cut and paste from the util-linux
mount manpage to ext4.5 (with commented-out lines
removed).
It's pretty much impossible for util-linux to keep up
with every filesystem out there, and Karel has more than
once expressed a wish that mount options move into fs-specific
manpages.
So, here we go.
Signed-off-by: Eric Sandeen <sandeen@redhat.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Eric Sandeen [Sat, 5 Jul 2014 03:03:14 +0000 (23:03 -0400)]
e2fsprogs: revise and extend chattr(1) and chattr usage()
The chattr(1) manpage and chattr usage() output were missing some flags.
Add those, and make some other minor cosmetic fixes.
(I've left out the 'B' (EXT2_COMPRBLK_FL) flag, because
it's not actually used anywhere, and I can't figure out
how it differs from 'c' (EXT2_COMPR_FL))
Also, because the matrix of filesystems & flags is quite large,
refer to filesystem-specific manpages for detailed discussion
of flags supported by those filesystems, rather than trying to
cover it all in this manpage. I'll send those manpage
updates to the appropriate lists a bit later.
Signed-off-by: Eric Sandeen <sandeen@redhat.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Reviewed-by: Andreas Dilger <adilger@dilger.ca>
Eric Sandeen [Sat, 5 Jul 2014 03:02:59 +0000 (23:02 -0400)]
e2fsprogs: reorder flags in chattr(1)
The flags described in chattr usage() and the chattr(1) manpage
were in semi-random order, which makes it hard to ascertain
which flags might be missing or undocumented, and to locate
flags within the manpage.
Re-order the list of flags in alphanumeric order, and do
the same for the flag descriptions in the body of the manpage.
There should be no content changes here, just reordering
for consistency.
Signed-off-by: Eric Sandeen <sandeen@redhat.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Reviewed-by: Andreas Dilger <adilger@dilger.ca>
Jan Kara [Fri, 4 Jul 2014 20:24:18 +0000 (16:24 -0400)]
e2fsck: fix last mount time and last write time in preen mode
Fixing last mount time and last write time is safe - there's no risk of
loosing any important information or making corruption significantly
worse even if we get it wrong. So let's just fix these times in preen
mode. This allows initrd to automatically check and mount root
filesystem in case system clock is wrong without having to manually set
broken_system_clock variable (openSUSE uses broken_system_clock by default
to avoid these problems during boot but this disables time-based checks
even on systems where clock is fine so that's not ideal either).
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Ross Burton [Fri, 4 Jul 2014 03:19:38 +0000 (23:19 -0400)]
move e2fsprogs-specific macros from aclocal.m4 to acinclude.m4
Autoconf says that aclocal.m4 is a machine-generated copy of
system-wide macros, whereas acinclude.m4 is for project-specific
macros. Reflect this division and ease building by anyone who needs
to re-generate the autotooling by splitting AX_TLS and CHECK_GNU_MAKE
to acinclude.m4.
Signed-off-by: Ross Burton <ross.burton@intel.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Eric Sandeen [Mon, 9 Jun 2014 13:52:19 +0000 (09:52 -0400)]
resize2fs: don't attempt to calculate minimum size on fs with errors
My old patch:
resize2fs: don't print minimum size if fs is not clean
almost did this, but it still calculated the size; it just didn't print
it. Which is a bit silly.
Jes had a pretty badly corrupted image which made the minimum size
calculation go off into the weeds. It was corrupted, and also marked
as having an error.
We'll eventually bail out for an unmounted filesystem if it's marked
as being in an error state anyway; just move that test & bail-out
to a much earlier point, and remove the now-duplicate one under the
print_min_size block.
This will catch & block all resize operations on an offline filesystem
with errors, in one central place.
29758d2 filefrag: exit with error code if an error is hit
introduced a couple errors; in one case it missed returning
a value, and possibly picked up errno from (unchecked) close(),
and in the other used a test where it needed an
assignment. So capture the error, move perror() directly
after the failed call in both cases, and fix the assignment.
Also fix a precedence problem with:
if (fe_flags & mask == 0)
which is equivalent to:
if (fe_flags & (mask == 0))
but we need:
if ((fe_flags & mask) == 0)
Signed-off-by: Eric Sandeen <sandeen@redhat.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Reviewed-by: Andreas Dilger <adilger@dilger.ca>
e2fsck: fully and properly gettextize two error messages
The phrases "mounted" and "in use" were filled in untranslated into
the messages. But it is better to gettextize entire sentences, and
not synthesize them from fragments.
Andreas Dilger [Tue, 27 May 2014 16:56:45 +0000 (12:56 -0400)]
filefrag: exit with error code if an error is hit
If an error is hit during filefrag operation, it will continue to run
(if multiple files are specified on the command-line), but will exit
with a non-zero value, so that callers can determine that some error
was hit.
Clean up the printing of FIEMAP flags and print some newer flags that
were missing. Also print unknown flags as hex values.
Signed-off-by: Andreas Dilger <adilger@dilger.ca> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Andreas Dilger [Tue, 20 May 2014 22:30:35 +0000 (16:30 -0600)]
debugfs: fix filehandle leak in copy_file()
Fix a file handle leak for the target file in copy_file() when error
handlers return without closing the file. Instead, clean up at the
end of the function to handle cleanup in normal and error cases.
Minor other code style cleanups.
Signed-off-by: Andreas Dilger <adilger@dilger.ca> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Andreas Dilger [Tue, 20 May 2014 22:31:10 +0000 (16:31 -0600)]
debugfs: fix string_to_time for MacOS strptime()
The strptime() function does not update fields in struct tm that are
not specified in the input format. The glibc implementation sets the
tm_yday field (%j) when any of the year (%Y), month (%m), or day (%d)
fields are changed, but the MacOS strptime() does not set tm_yday in
this case. This caused string_to_time() to calculate the wrong Unix
epoch on MacOS. If tm_yday is unset, compute it in string_to_time().
This also fixes test regression failures for FreeBSD.
Signed-off-by: Andreas Dilger <adilger@dilger.ca> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Darrick J. Wong [Mon, 12 May 2014 04:21:29 +0000 (00:21 -0400)]
mke2fs: set gdt csum when creating packed fs
When we're creating a fs with metadata blocks packed at the beginning
(packed_meta_blocks=1 in mke2fs.conf), set the group descriptor
checksum or else we create DOA filesystems with checksum errors.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Lukas Czerner <lczerner@redhat.com>
Darrick J. Wong [Mon, 12 May 2014 03:33:08 +0000 (23:33 -0400)]
resize2fs: fix sanity check in reserve_sparse_super2_last_group()
In reserve_sparse_super2_last_group, the old_desc check should only be
performed if ext2fs_super_and_bgd_loc2() gave us a location -- a
return value of 0 means that there is no old-style GDT block.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Theodore Ts'o [Sun, 11 May 2014 04:23:49 +0000 (00:23 -0400)]
quota: remove mke2fs's and tune2fs's warning messages regarding quota
We no longer need to reference https://ext4.wiki.kernel.org/index.php/Quota
since we've fixed the nasty bugs associated with e2fsck and the quota
feature. The wiki page will be updated once we've done a release that
includes these fixes indicated the verison which these problems have
been fixed.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Aditya Kali <adityakali@google.com>
Theodore Ts'o [Sat, 10 May 2014 22:55:20 +0000 (18:55 -0400)]
quota: integrate mkquota.h into quotaio.h
There are interfaces that are used by mke2fs.c and tune2fs.c which are
in quotaio.h, and some future changes will be much simpler if we can
combine the two header files together. Also the guard #ifdef for
mkquota.h was incorrect, which caused problems when both header files
needed to be included.
Also remove quota.pc and installation rules for libquota, since this
library is never going to be something that we can export externally
anyway. Eventually we'll want to clean up the interfaces and move the
external publishable interfaces to the libext2fs library, and then
rename what's left from libquota.a to libsupport.a for internal use
only.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Aditya Kali <adityakali@google.com>
Theodore Ts'o [Sat, 10 May 2014 21:29:18 +0000 (17:29 -0400)]
quota: fix memory leak in quota_compare_and_update()
The quota_handle wasn't getting closed in quota_compare_and_update().
Fix this, and also make sure that quota_file_close() doesn't
unnecessarily modify the quota inode if it's not necessary. Otherwise
e2fsck will claim that the file system is modified when it didn't need
to be.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Aditya Kali <adityakali@google.com>
Theodore Ts'o [Sat, 10 May 2014 05:27:31 +0000 (01:27 -0400)]
quota: fix e2fsck to notice missing quota entries
Previously if there was a missing quota entry --- i.e., if there were
files owned by group "eng", but there was no quota record for group
"eng", e2fsck would not notice the missing entry. This means that the
usage informtion would not be properly repaired. This is unfortunate.
Fix this by marking each quota record in quota_dict that has a
corresponding record on disk, and then check to see if there are any
records in quota_dict that have not been marked as having been seen.
In that case, we know we need to update the relevant quota inode.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Aditya Kali <adityakali@google.com>
Theodore Ts'o [Sat, 10 May 2014 05:11:41 +0000 (01:11 -0400)]
quota: fix e2fsck so we update the quota file correctly
In scan_dquota_callback() we were copying dqb_off from the on-disk
dquot structure to the dqot structure that we have constructed in
memory. This is a bad idea, because if we detect that the quota inode
is corrupted and needs to be replaced, when we later write out the
inode, the fact that we have a non-zero dqb_off value means that quota
routines will not bother updating the quota tree index, since
presumably the quota tree index already has an entry for this dqot.
The problem is that e2fsck, the only user of these functions, has
zapped the quota inode so it can be written from scratch. So this
means the index for the quota records are not written out, so the
kernel can not find any of the records in the quota inodes. Oops.
The fix is simple; there is no reason to copy the dqb_off field into
the quota_dict copy of struct dquot.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Aditya Kali <adityakali@google.com>
Niu Yawei [Thu, 8 May 2014 02:38:53 +0000 (10:38 +0800)]
libquota: fix dict_uint_cmp()
dict_uint_cmp() returns an usigned int value in int type, which
could mess the dict key comparison when the difference of two
keys is greater than INT_MAX.
Chen Qi [Tue, 6 May 2014 01:08:42 +0000 (21:08 -0400)]
Fix cross compilation problem in parse-types.sh
The checking of types in parse-types.sh doesn't make much sense in a
cross-compilation environment, because the generated binary is
executed on build machine.
So even if asm_types.h has got correct statements for types, it's
possible that the generated binary will report an error, because these
types are for the target machine.
Theodore Ts'o [Mon, 5 May 2014 04:20:31 +0000 (00:20 -0400)]
mke2fs: print extra information about existing ext2/3/4 file systems
The basic idea is to provide a bit more context in this situation:
% ./misc/mke2fs -t ext4 /dev/sdc3
mke2fs 1.42.9 (4-Feb-2014)
/dev/sdc3 contains a ext4 file system
Proceed anyway? (y,n)
... by adding this bit of context:
% ./misc/mke2fs -t ext4 /dev/sdc3
mke2fs 1.42.9 (4-Feb-2014)
/dev/sdc3 contains a ext4 file system
last mounted on /SOX-backups on Mon May 5 08:59:53 2014
Proceed anyway? (y,n)
projects / thirdparty / e2fsprogs.git / log
