Test that mke2fs does not allow to create file system with both
resize_inode and meta_bg features enabled.
This was fixes with commit 42e77d5d ("libext2fs: don't create
filesystems with meta_bg and resize_inode").
Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Test if the e2fsck can fix file system with resize_inode and meta_bg
features enabled simultaneously.
Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Make generated test scripts read-only, to avoid errors by developers
editing the generated test scripts and then having them accidentally
clobbered when "make" is run again.
Change-Id: I60d417b816b7c559b5e05baf4167fc2cf2a871cf
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
We do not allow initialized blocks to exist past i_size as this could
lead to stale data exposure.
Remove test f_pgsize_gt_blksize because it is testing for the scenario
that not allowed. f_eofblocks is already testing for this.
Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
If the developer is running with SELinux enabled on /tmp, the
f_extent_htree regression test will fail because mke2fs by default
copies the extended attributes into the newly created file system (if
a directory hierarchy is specified via the -d option).
Fix this by adding a new extended option to mke2fs, -E no_copy_xattrs
and using it in f_extent_htree's test script.
Reported-by: Lukas Czerner <lczerner@redhat.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Resize2fs will not enable lazy_itable if the kernel apparently does
not support that feature. This will cause spurious test failures when
the tests are running on such a system (or where sysfs is not
mounted).
So for the purposes of the regression test we need to force the use of
lazy_itable so that the results conform to expected golden output.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Previously we were always retrying the ext2fs_open2() call with the
EXT2_FLAG_IGNORE_SB_ERRORS flag. It only makes sense to do this if
the superblock is reported as corrupt. Otherwise, it's a waste of
time, and results in printing an extra set of error messages to annoy
the user.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
The test script was not properly checking e2image's ability to convert
a qcow2 image back to a raw image. Fix this, and also change it to
use /tmp for the temporary files to speed things up systems with a
ramdisk for /tmp.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Mark device inodes bad if they have the EXT4_EXTENTS_FL set, since
this should never happen and likely shows the inode is corrupt.
Change-Id: I85ab667b39ff57c658a779e59f692a080217690e
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Verify correct operation in the case of writing files with allocated
blocks at the end of the file beyond i_size. This can happen for
PAGE_SIZE > blocksize, or through fallocate().
The f_eofblocks thest checks quite the same situation, but for broken
file system. This test is about correct file.
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Artem Blagodarenko <artem.blagodarenko@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Fix handling of block preallocation support in cases where the kernel
PAGE_SIZE is larger than the filesystem blocksize.
Signed-off-by: Kalpak Shah <kalpak@clusterfs.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
e2mmpstatus is a Multi-Mount Protection (MMP) helper utility to read
an MMP block to see if it is being updated. It can also output the
latest update time, nodename, and device from the MMP block.
This is useful for HA and other maintenance scripts to determine if
the filesystem is in use on another node, and which node it is.
Signed-off-by: Shuichi Ihara <sihara@ddn.com>
Signed-off-by: Li Xi <lixi@ddn.com>
Signed-off-by: Wang Shilong <wshilong@ddn.com>
Moved e2mmpstatus checking/dumping code to be part of dumpe2fs rather
than a standalone program, using the "-m" option to check MMP status,
and "-i" to dump info. If dumpe2fs is called as "e2mmpstatus" (and
also "mmpstatus" for compatibility reasons), assume "-m" is specified.
Re-use the existing MMP block handing routines (with some changes) to
check and dump the MMP block, rather than adding duplicate versions.
Modify dumpe2fs to exit with a non-zero error code if there is an
error while reading the filesystem metadata or MMP block, or if
"-m" is used with the "mmp" feature and is in use by another node.
Add a configure check for gethostname() rather than depending on
_BSD_SOURCE or _XOPEN_SOURCE to be set.
Update the f_mmp, m_mmp, m_mmp_bad_csum, and m_mmp_bad_magic tests
to use e2mmpstatus to check and dump the MMP state before and after
e2fsck is run to verify that the tool is working correctly.
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
If there is a directory with more than EXT2_LINK_MAX (65000)
subdirectories, but the DIR_NLINK feature is not set in the
superblock, the feature should be set before continuing on
to change the on-disk directory link count to 1.
While most filesystems should have DIR_NLINK set (it was set
by default for all ext4 filesystems, and all kernels between
2.6.23 and 4.12 automatically set it if the directory link
count grew too large), it is possible that this flag is lost
due to disk corruption or for an upgraded filesystem. We no
longer want kernels to automatically enable features.
Addresses: https://bugzilla.kernel.org/show_bug.cgi?id=196405
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
The f_extent_htree test depends on the "mke2fs -d" functionality
that did not exist in the maint branch when the patch was landed.
On master this functionality exists and should be used instead of
a pre-generated image. That exercises the "mke2fs -d" functionality
as well as e2fsck better.
[ Fixed to add missing expect.pre.? files --tytso ]
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Don't call sed multiple times on the output, and avoid the use
of temporary files, or if possible. It would be convenient to
use "sed -i" to only update the output file once, but this is
not portable to all platforms.
[ Fixed a few test regression failures --tytso ]
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Now that the majority of device name filtering is in filter.sed,
it does not need to be specified explicitly for every test.
Fix the error message printed in debugfs when opening the device
to match that used in other tools. This simplifies the filtering,
and will be helpful if debugfs messages are internationalized.
[ Fixed up some test failures in m_hugefile t_uninit_bg_rm --tytso ]
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Instead of putting the entire test script under an implicit
"if test -x $DEBUGFS_EXE" conditional (sometimes indenting
the code, and sometimes not), rather check for the reverse
and exit the test script early if $DEBUGFS_EXE is missing.
In some places, tests were running $DEBUGFS_EXE directly,
when they should be running $DEBUGFS (which will run with
Fortify, or other options).
[ Fixed up missing exit statement in f_detect_junk. --tytso ]
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Don't register a trap to unlink $TMPFILE at the test exit if
SKIP_UNLINK is set. Otherwise, this makes it difficult to
debug a failing test.
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
If the TEST_BITS file is sparse, then the "debugfs -R write"
command may skip holes in the file when copying it into the
test image (depending on whether SEEK_HOLE/SEEK_DATA and/or
FIEMAP are available in the copy_file() function).
This was causing test failures on MacOS in the f_dup_resize
and d_loaddump tests because the TEST_BITS file was the
compiled "debugfs" binary, which apparently has holes when
built on MacOS, and the number of blocks allocated in the
test image was reduced as a result. This caused the expect
output to differ in the summary line and resulted in failure.
Instead of using the debugfs binary for TEST_BITS, generate
a temporary file using /dev/urandom, if available. If not,
fall back to the old behaviour or using debugfs.
[ Fixed up use of non-POSIX /bin/sh constructs and add test_data.tmp to
.gitignore and files to be deleted on a make clean. --tytso ]
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Google-Bug-Id: 64109868
Test: e2fsck -E unshare_blocks does a full scan
Change-Id: Idc36ceba3bf24e1fb1487feedefe9a68f9acc7f3
From AOSP commit: 7c180d6598363722de6195d142d7677bbc2b0161
Add an -E unshare_blocks flag for unsharing blocks that were created for
a filesystem with block sharing enabled. If the filesystem does not have
this feature enabled, the flag has no effect. If the filesystem does not
have free space, e2fsck will error.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Google-Bug-Id: 64109868
Test: f_unshare_blocks_no_space, f_unshare_blocks_ok
Change-Id: I8821353e9e6200c6c0c71dd22f4f43d796fc720c
From AOSP commit: 8ba190e3135d61501d3a694b6960c2fbee98e7a6
If RO_COMPAT_SHARED_BLOCKS is set, the duplicate block pass is skipped
entirely.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Google-Bug-Id: 64109868
Test: f_shared_blocks_ok
Change-Id: I49a9d6c5012d5e76c9a47578a45cae7484df4c4a
From AOSP commit: 93d12bc16bffd39a56b1e08aefbc641eba298077
Add a flag to ext2fs_open() which allows to open a filesystem even if
superblock is somewhat inconsistent. Use this flag from e2fsck as a last
resort to try to fix the superblok.
Currently, the flag does nothing. We'll relax checks in ext2fs_open()
once e2fsck is able to handle corresponding corruption gracefully.
Reviewed-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Make sure we can handle a maliciously created file system containing
an inode containing an extended attribute whose e_value_inum points
back at itself.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Don't use "debugfs -f /dev/stdin" for portability reasons --- not all
systems have /dev/stdin. Simply dropping "-f /dev/stdin" works just
fine.
Signed-off-by: Ingo Brückl <ib@wupperonline.de>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
The tests f_bigalloc_badinode and f_bigalloc_orphan_list were not
using the version of mke2fs and debugfs from the build tree, and if
mke2fs and debugfs are not in the user's PATH, these tests would fail.
Fix this.
Reported-by: Somchai Smythe <buraphalinuxserver@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
If e2fsck encounters a filesystem that supports internal quotas, it is
responsible for adjusting the quota counters if it decides to clear any
orphaned inodes. Therefore, we must read the quota files, adjust the
counters, and write the quota files back out when we are done.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
On the Hurd, mke2fs will force the use of 4k block sizes by default
because the Hurd's implemntation of ext2 doesn't support any other
block sizes. This causes spurious test failures. Since these test
are testing e2fsprogs functionality, force the use of 1k block sizes
so the test output matches the expected output.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
While deleting a bad inode in fsck pass2, we should remove clusters
only once. We do it by remembering last released cluster while
deleting clusters one by one.
Signed-off-by: Harshad Shirwadkar <harshads@google.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
This is a test to make sure e2image and its associated library
routines can correctly read the inode and block allocation bitmaps.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
This is a regression test for the bugfix found in commit 707599bf:
"e2fsck: release clusters only once in release_inode_blocks"
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Add a test which verifies that e2fsck will detect a variety of bad
symlinks, both fast and slow, and with several combinations of
filesystem features including extents, encryption, and inline_data.
There was already a similar test (f_badsymlinks), but it's an old test
that doesn't use any of these newer filesystem features.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Both the kernel and e2fsck expect that if an inode has inline data, then
it contains a "system.data" xattr -- even if i_size <= 60 so the data
fits entirely in i_block.
But if a symlink of exactly 60 bytes (not counting a NUL terminator) was
created using ext2fs_symlink() and the inline data feature was enabled,
then the symlink inode ended up with inline data but without a
system.data xattr. This is possible because "fast" symlinks store a NUL
terminator but inline data symlinks do not. So 60 bytes is too long for
a real fast symlink, but still short enough to fit the entire target in
i_block as a "slow" symlink using inline data.
Some places use ext2fs_inline_data_init() to ensure the system.data
xattr is created, but the symlink path does not.
Fix this by making ext2fs_inline_data_set() set system.data to an empty
string when i_size <= 60.
Fixes: 54e880b870 ("libext2fs: handle inline data in read/write function")
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
ext2fs_symlink() didn't count the NUL terminator when limiting the
symlink target length to fs->blocksize, so it could create symlinks
using the entire block. Such symlinks are problematic because if
block_size == PAGE_SIZE (as is usually the case), then when following
such a symlink the kernel will truncate the last byte because it needs
to ensure the symlink page is NUL-terminated (see page_get_link()).
Perhaps for that reason, e2fsck considers such symlinks to be invalid
(although only if they use the traditional block list rather than an
extent tree, which seems to be another bug).
Fix this by counting the NUL terminator, thereby decreasing the maximum
accepted symlink target length by 1. Note that this matches the
kernel's behavior in ext4_symlink().
This breaks the test 'f_create_symlinks' which was using debugfs to
create a 1024-byte symlink (not counting the NUL terminator) on a
1024-byte block filesystem. Fix it by removing the leading '/' from the
test's symlink targets so that their lengths are decreased by 1.
Fixes: f01c1a6bce ("libext2fs: add the ext2fs_symlink() function")
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Also change the t_replay_and_set test so that we do something more
innocuous, such as setting the file system label, instead of something
much more dangerous such as removing metadata_csum feature (which
requires rewriting the metadat checksums, and this will fail
catastrophically after the test corrupts the inode bitmaps and we
perform the journal replay correctly).
Reported-by: NeilBrown <neilb@suse.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Create a .slow file for tests that take too long to run, so it is
possible to find them easily after all of the tests have run.
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
While not critical to functionality, it is easier to see the pass/fail
status of each test if they are not wrapped onto the next line.
Shorten test descriptions so the name+description fit in 80 columns.
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
These two tests could leave an temp file behind if the test is
skipped. In general test script files should avoid using 'exit',
since this bypasses the test cleanup scripting.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Don't leave temp files around after test has completed successfully.
Some test scripts were creating their own temporary files, and in
some cases test_one was skipping a test completely (e.g. slow) after
the temp file had been created.
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
This a regression test for ca74ecf325: "libext2fs: skip start_blk
adjustment when stride and flex_bg is set"
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
This is a regression test for 0c12896e08: "libext2fs: prevent
allocating inode table from already used blocks"
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Print out a message if a test takes longer than 60s, with a
reminder to potentially add is_slow_test to that test, so
that it is easier to find which tests are taking a long time,
especially if running with "make -j check" or similar.
Add an exclusion for r_expand_full on MacOS since HFS+ does
not support sparse devices and doesn't like large test images.
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
