When resize2fs exits early, perhaps because of an error, we should
free the file system so that if MMP is in use, the MMP block is reset.
This also releases the memory to avoid memory leak reports.
Addresses-Debian-Bug: #984472
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
When resizing past the point where the reserve inode has reserved
space for the block group descriptors to expand, and resize2fs (in an
offline resize) needs to move the allocation bitmaps and/or inode
table around, it's possible for resize2fs to allocate over the MMP
block, which would be bad.
Prevent this from happening by reserving the MMP block as a file
system metadata block (which it is) in resize2fs's accounting.
Addresses-Debian-Bug: #984472
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
When unix_io does not use pread/pread64 (which is the case the bounce
buffer is in use, either when Direct I/O is in use or the
IO_FLAG_FORCE_BOUNCE in enabled), there are races between the llseek
and and read or write system calls. Fix this by using the BOUNCE_MTX
so only one thread is using the file descriptor at a time.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
The bounce read/write code would crash with a floating point exception
if alignment is set to 0.
Fixes: c001596110 ("libext2fs: fix unix_io's Direct I/O support")
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
The previous Direct I/O support worked on HDD's with 512 byte logical
sector sizes, and on FreeBSD which required 4k aligned memory buffers.
However, it was incomplete and was not correctly working on HDD's with
4k logical sector sizes (aka Advanced Format Disks).
Based on a patch from Alexey Lyashkov <alexey.lyashkov@hpe.com> but
rewritten to work with the latest e2fsprogs and to minimize changes to
make it easier to review.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reported-by: Artem Blagodarenko <artem.blagodarenko@gmail.com>
The iscan program program isn't built by default, and was relying on
e2fsck's util.c, so it had suffered bitrot as e2fsck/util.c had
evolved. Fix it so that iscan builds correct.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Currently the mmp block is read using O_DIRECT to avoid any caching that
may be done by the VM. However when working with regular files this
creates alignment issues when the device of the host file system has
sector size larger than the blocksize of the file system in the file
we're working with.
This can be reproduced with t_mmp_fail test when run on the device with
4k sector size because the mke2fs fails when trying to read the mmp
block.
Fix it by disabling O_DIRECT when working with regular files. I don't
think there is any risk of doing so since the file system layer, unlike
shared block device, should guarantee cache consistency.
Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
This commit fixes a bug where if a small file system is resized to
ridiculous sizes, such that the size of the resized block group
descriptor blocks exceed the original file system, and this would
result in resize2fs triggering a large number of scary warning
messages:
Illegal block number passed to ext2fs_test_block_bitmap #12440
for block bitmap for broken.img
This can be replicated via:
./misc/mke2fs broken.img -b 1024 -T default
truncate -s +1500G broken.img
./resize/resize2fs broken.img
Fortunately, aside from triggering these warning messages, the bug had
no other bad effects.
https://github.com/tytso/e2fsprogs/issues/60
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
This adds support for setting/querying the FS_NOCOMP_FL/EXT2_NOCOMPR_FL
file flag to chattr/lsattr. I picked the character "m" because it was
so far unused and all other characters that were more obvious candidates
were already taken.
The flag is available on btrfs, and with this patch it is possible to
manage it correctly.
Signed-off-by: Lennart Poettering <lennart@poettering.net>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
In read_list func, if strtoull() fails in while loop,
we will return the error code directly. Then, memory of
variable lst will be leaked without setting to *list.
Signed-off-by: Zhiqiang Liu <liuzhiqiang26@huawei.com>
Signed-off-by: linfeilong <linfeilong@huawei.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Initialize ext2fs_ex variable in ext4_fc_replay_scan() before first
use. Also make sure ext2fs_decode_extent() completely overwrites the
extent structure passed to it as argument to prevent potential future
bugs for the users of the function.
Signed-off-by: Harshad Shirwadkar <harshadshirwadkar@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
During fast commit replay scan phase, in ext4_fc_replay_scan(), we
want to fallthrough in switch case for EXT4_FC_TAG_ADD_RANGE case. Add
a comment for that.
Signed-off-by: Harshad Shirwadkar <harshadshirwadkar@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Replace the remaining loff_t uses with ext2_loff_t, as
was done in patch 1df6a4555, since loff_t is a GCC'ism
and is not portable.
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Follow the example in do_mkdir_internal, and do_symlink_internal,
and find the correct parent directory if the destination
is not just a plain basename.
https://github.com/tytso/e2fsprogs/issues/61
Signed-off-by: Earl Chew <earl_chew@yahoo.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
The fatal_error() function in e2fsck can call ext2fs_mmp_stop() on a
file system where MMP hasn't yet been initialized. When that happens,
instead of crashing, have ext2fs_mmp_stop() return success, since mmp
doesn't need to be stopped if it hasn't even been initialized yet.
Addresses-Debian-Bug: #696609
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Also remove a dead assignment (the value of try is overwritten after
the continue statement).
Addresses-Coverity-Bug: 1297515
Addresses-Coverity-Bug: 1464573
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
This function can't actually fail today, but in the future it could
return an error, so it's better to add the appropriate error check.
Addresses-Coverity-Bug: #1464579
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
On most systems where we compile e2fsprogs, the u64 type is an
unsigned long long. However, there are platforms (such as the
PowerPC) where a long 64-bits and so u64 is typedef'ed to be unsigned
long instead of a unsigned long long. Fix this by using explicit
casts in printf statements. For scanf calls, we need to receive the
value into a unsigned long long, and then assign it to a u64, after
doing range checks.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
The test script had a dependency on the group id for the "adm" group
being 4. Fixed by hard-coding the group id for the default acl for
the acl_dir directory.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
