While it may be true that playing games with old_fs' block count
during a grow operation shuts up a bunch of warnings, resize2fs
doesn't actually expand the group descriptor array to match the size
we're artificially stuffing into old_fs, which means that if we
actually need to allocate a block out of the larger fs (i.e. we're in
desperation mode), ext2fs_block_alloc_stats2() scribbles on the heap,
leading to crashes if you're lucky and FS corruption if not.
So, rip that piece out and turn off com_err warnings properly and add
a test case to deal with growing a nearly full filesystem.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Recalculate the unused inode count and the block/inode uninit flags
when resizing a filesystem. This can speed up future e2fsck runs
considerably and will reduce mount times.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
When shrinking a filesystem, resize2fs wants to free per-bg metadata
blocks that are no longer needed. This behavior is gated on whether
there's a superblock in the group as told by new_fs. The check really
should be against old_fs, since we're effectively freeing blocks out
of old_fs in the transition to new_fs, but prior to sparse_super2 this
didn't matter since superblocks didn't move, so it didn't matter.
Under sparse_super2, however, there's a superblock in the last group,
so now we need to change the test to use old_fs as it should.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
When we're moving an inode on a metadata_csum filesystem, we need to
rewrite the checksum of all interior nodes of the extent tree. The
current code does this inefficiently via set_bmap, but we can do this
more efficiently through direct iteration of the extent tree.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
If we're shrinking a sparse_super2 filesystem to a single block group,
the superblock will be in block 0. This is perfectly valid (for block
group 0 with a blocksize > 1024) so don't exit.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Fix some gcc-4.8 warnings and other problems that broke the build.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Convert all call sites that write zero blocks to disk to use
ext2fs_zero_blocks2() since it can use Linux's zero out feature to do
the writes more quickly. Reclaim the zero buffer at freefs time and
make the write-zeroes fallback use a larger buffer.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
When shrinking a file system, if the number block groups drops below
the point where we started using the meta_bg layout, disable the
meta_bg feature and set s_first_meta_bg to zero. This is necessary to
avoid creating an invalid/corrupted file system after the shrink.
Addresses-Debian-Bug: #756922
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reported-by: Marcin Wolcendorf <antymat+debian@chelmska.waw.pl>
Tested-by: Marcin Wolcendorf <antymat+debian@chelmska.waw.pl>
There are a number of places where we need convert groups to blocks or
clusters by multiply the groups by blocks/clusters per group.
Unfortunately, both quantities are 32-bit, but the result needs to be
64-bit, and very often the cast to 64-bit gets lost.
Fix this by adding new macros, EXT2_GROUPS_TO_BLOCKS() and
EXT2_GROUPS_TO_CLUSTERS().
This should fix a bug where resizing a 64bit file system can result in
calculate_minimum_resize_size() looping forever.
Addresses-Launchpad-Bug: #1321958
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
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>
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>
In the case where the new location of the inode table is before the
old inode table, the optimization which tries to optimize zero block
moves breaks. Fix it.
This fixes a bug that was tickled by the reproduction described in the
previous commit.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The previous commit exposed bugs in the calculation for flex_bg file
systems. The problem is that since (by default) we keep the metadata
blocks for the flex_bg in the first block group of the flex_bg, and
because we don't want to overwrite metadata blocks used by the
original file system with data blocks make life easier in case the
resize is aborted for some reason, we need to treat all of the
metadata blocks in the existing flex_bg has in use for the purposes of
calculate_minimum_resize_size().
Even though this means we need to reserve more data blocks to avoid
running out of space, the net result of these two commits is a net
savings in how much we can shrink a file system.
Using the following test sequence:
mke2fs -F -t ext4 /tmp/foo.img 2T
resize2fs -M /tmp/foo.img
resize2fs -M /tmp/foo.img
resize2fs -M /tmp/foo.img
Here is the comparison in the resulting file systems between the old
and new resize2fs (units are in 4k blocks):
resize #1 resize #2 resize #3
old resize2fs 1117186 45679 43536
new resize2fs 48784 37413 37392
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
For extent-mapped file systems, we need to reserve some extra space in
case we need to grow the extent tree. Calculate the safety margin
more intelligently, so we don't overestimate the amount of space
required.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Reported-by: Dmitry Monakhov <dmonakhov@openvz.org>
When we're looking for directory blocks for the inode remapping step,
we need to include inline_data directories in the remap process.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
When we're moving blocks around the filesystem, ensure that freeing
the old blocks only frees the clusters if they're not in use by other
metadata.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
When freeing a block group descriptor block, be careful not to free
metadata clusters belonging to other groups!
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
ext2fs_free() does not set the ext2_filsys pointer to null so the
caller is responsible to setting it himself if it is needed.
This patch fixes some places where caller did not set ext2_filsys
pointer to NULL after ext2fs_free() which might result in use after
free. Fix it.
Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
In practice, it is **extremely** rare for users to try to use more
than the first backup superblock located at the beginning of block
group #1. (i.e., at block number 32768 for file systems with a 4k
block size). This new compat feature restricts the backup superblock
to block group #1 and the last block group in the file system.
Aside from reducing the overhead of the file system by a small number
of blocks, by eliminating the rest of the backup superblocks, it
allows us to have a much more flexible metadata layout. For example,
we can force all of the allocation bitmaps and inode table blocks to
the beginning of the disk, which allows most of the disk to be
exclusively used for contiguous data blocks.
This simplifies taking advantage of certain HDD specific features,
such as Shingled Magnetic Recording (aka Shingled Drives), and the
TCG's OPAL Storage Specification where having a simple mapping between
LBA block ranges and the data blocks used by the file system can make
life much simpler.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Mostly by adding static and removing excess extern qualifiers. Also
convert a few remaining non-ANSI function declarations to ANSI.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Compiling with LLVM generates a large number of warnings due
to the use of _() for wrapping strings for i18n:
warning: format string is not a string literal
(potentially insecure) [-Wformat-security]
./nls-enable.h:4:14: note: expanded from macro '_'
#define _(a) (gettext (a))
^~~~~~~~~~~~
These warnings are fixed by using "%s" as the format string,
and then _() is used as the string argument.
Signed-off-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
With the advent of metadata_csum, we now tie extent and directory
blocks to the associated inode number (and generation). Therefore, we
must be careful when remapping inodes. At that point in the resize
process, all the blocks that are going away have been duplicated
elsewhere in the FS (albeit with checksums based on the old inode
numbers). If we're moving the inode, then do that and remember that
new inode number. Now we can update the block mappings for each inode
with the final inode number, and schedule directory blocks for mass
inode relocation. We also have to recalculate the EA block checksum.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
metadata_csum implies uninit_bg, and in fact forces the bit off for
rocompat with older implementations. Therefore, to detect the
presence of checksums, we should use the predicate function to decide
if group descriptor checksums are turned on, not open-coded flag
tests.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
If the file system is being shrunk, and a block group's inode table
falls beyond the end of the inode table, we need to try to relocate
the inode table blocks.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
If the file system's inode table blocks in the last block group are
located in the middle or the end of the block group, it's possible for
resize2fs -M to use a size which will require relocating the inode
table blocks in the last block group. This can lead to all sorts of
problems, so solve it by simply guaranteeing that we will never do
that.
Reported-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
free_gdp_blocks needs to be taught to use 64-bit fields and the appropriate
getters, otherwise it'll truncate high block numbers (when, say, resizing a
>16T fs) and mark the low numbered group descriptor blocks as free. Yikes.
Reported-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
It is possible to have a flex_bg filesystem with block groups
which have inode & block bitmaps at some point well past the
start of the group.
If an offline shrink puts the new size somewhere between
the start of the block group and the (old) location of
the bitmaps, they can be left beyond the end of the filesystem,
i.e. result in fs corruption.
Check each remaining block group for whether its bitmaps
are beyond the end of the new filesystem, and reallocate
them in a new location if needed.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Accessing name_len (and file_type) in ext4_dir_entry structure is
somewhat problematic because on big endian architecture we need to now
whether we are really dealing with ext4_dir_entry (which has u16
name_len which needs byte swapping) or ext4_dir_entry_2 (which has u8
name_len which must not be byte swapped).
Currently the code is somewhat surprising and name_len is always
treated as u16 and byte swapped (flag EXT2_DIRBLOCK_V2_STRUCT isn't
ever used) and then masking of name_len is used to access real
name_len or file_type. Doing things this way in applications using
libext2fs is unexpected to say the least (more natural is to type
struct ext4_dir_entry * to struct ext4_dir_entry_2 * but that gives
wrong results on big endian architectures. So provide helper functions
that give endian-safe access to these fields. Also convert users in
e2fsprogs to use these functions.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
When doing an off-line resize2fs of an initially very small file
system, it's possible to run out of reserved gdt blocks (which are
reserved via the resize inode). Once we run out, we need to move the
allocation bitmaps and inode table out of the way to grow the gdt
blocks. Unfortunately, when moving these metadata blocks, it was
possible that a block that had been just been newly allocated for a
new block group could also get allocated for a metadata block for an
existing block group that was being moved.
To prevent this, after we grow the gdt blocks and allocate the
metadata blocks for the new block groups, make sure all of these
blocks are marked as reserved.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Reported-by: John Jolly <john.jolly@gmail.com>
Fixes resize2fs so it correctly calculates the number of free clusters
in each block group for file systems with the bigalloc feature
enabled.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
There are a number of places where we multiply a dgrp_t with
s_blocks_per_group expecting that we will get a blk64_t. This
requires a cast, or using the convenience function
ext2fs_group_first_block2().
This audit was suggested by Eric Sandeen.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Use ext2fs_[un]mark_block_range2() functions to reduce the CPU
overhead of resizing large file systems by 45%, primarily by
reducing the time spent in fix_uninit_block_bitmaps().
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Darrick J. Wong