If the caller isn't planning on modifying the block group descriptors,
there's no need to pass in a pointer to a struct buffer_head. Nuking
this saves a tiny amount of CPU time and stack space usage.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The __ext4_write_dirty_metadata() function was introduced by commit
0390131b, "ext4: Allow ext4 to run without a journal", but nothing
ever used the function, either then or since. So let's remove it and
save a bit of space.
Cc: Frank Mayhar <fmayhar@google.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Until now we have had a 1:1 mapping between storage device physical
block size and the logical block sized used when addressing the device.
With SATA 4KB drives coming out that will no longer be the case. The
sector size will be 4KB but the logical block size will remain
512-bytes. Hence we need to distinguish between the physical block size
and the logical ditto.
This patch renames hardsect_size to logical_block_size.
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Not sure why I put this in as down_write originally; all we are
doing is walking the tree, nothing will change under us and
concurrent reads should be no problem.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
To catch filesystem bugs or corruption which could lead to the
filesystem getting severly damaged, this patch adds a facility for
tracking all of the filesystem metadata blocks by contiguous regions
in a red-black tree. This allows quick searching of the tree to
locate extents which might overlap with filesystem metadata blocks.
This facility is also used by the multi-block allocator to assure that
it is not allocating blocks out of the system zone, as well as by the
routines used when reading indirect blocks and extents information
from disk to make sure their contents are valid.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
If two CPU's simultaneously call ext4_ext_get_blocks() at the same
time, there is nothing protecting the i_cached_extent structure from
being used and updated at the same time. This could potentially cause
the wrong location on disk to be read or written to, including
potentially causing the corruption of the block group descriptors
and/or inode table.
This bug has been in the ext4 code since almost the very beginning of
ext4's development. Fortunately once the data is stored in the page
cache cache, ext4_get_blocks() doesn't need to be called, so trying to
replicate this problem to the point where we could identify its root
cause was *extremely* difficult. Many thanks to Kevin Shanahan for
working over several months to be able to reproduce this easily so we
could finally nail down the cause of the corruption.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Reviewed-by: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
The BH_Unwritten flag indicates that the buffer is allocated on disk
but has not been written; that is, the disk was part of a persistent
preallocation area. That flag should only be set when a get_blocks()
function is looking up a inode's logical to physical block mapping.
When ext4_get_blocks_wrap() is called with create=1, the uninitialized
extent is converted into an initialized one, so the BH_Unwritten flag
is no longer appropriate. Hence, we need to make sure the
BH_Unwritten is not left set, since the combination of BH_Mapped and
BH_Unwritten is not allowed; among other things, it will result ext4's
get_block() to be called over and over again during the write_begin
phase of write(2).
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The ext4_get_blocks() function was depending on the value of
bh_result->b_state as an input parameter to decide whether or not
update the delalloc accounting statistics by calling
ext4_da_update_reserve_space(). We now use a separate flag,
EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE, to requests this update, so that
all callers of ext4_get_blocks() can clear map_bh.b_state before
calling ext4_get_blocks() without worrying about any consistency
issues.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The static function ext4_da_get_block_write() was only used by
mpage_da_map_blocks(). So to simplify the code, merge that function
into mpage_da_map_blocks().
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Use a very large unsigned number (~0xffff) as as the fake block number
for the delayed new buffer. The VFS should never try to write out this
number, but if it does, this will make it obvious.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
We need to mark the buffer_head mapping preallocated space as new
during write_begin. Otherwise we don't zero out the page cache content
properly for a partial write. This will cause file corruption with
preallocation.
Now that we mark the buffer_head new we also need to have a valid
buffer_head blocknr so that unmap_underlying_metadata() unmaps the
correct block.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Enforce that noalloc_get_block_write() is only called to map one block
at a time, and that it always is successful in finding a mapping for
given an inode's logical block block number if it is called with
create == 1.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
This adds more documentation to various internal functions in
fs/ext4/inode.c, most notably ext4_ind_get_blocks(),
ext4_da_get_block_write(), ext4_da_get_block_prep(),
ext4_normal_get_block_write().
In addition, the static function ext4_normal_get_block_write() has
been renamed noalloc_get_block_write(), since it is used in many
places far beyond ext4_normal_writepage().
Plenty of warnings have been added to the noalloc_get_block_write()
function, since the way it is used is amazingly fragile.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The functions ext4_get_blocks(), ext4_ext_get_blocks(), and
ext4_ind_get_blocks() used an ad-hoc set of integer variables used as
boolean flags passed in as arguments. Use a single flags parameter
and a setandard set of bitfield flags instead. This saves space on
the call stack, and it also makes the code a bit more understandable.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Another function rename for clarity's sake. The _wrap prefix simply
confuses people, and didn't add much people trying to follow the code
paths.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The static function ext4_get_blocks_handle() is badly named. Of
*course* it takes a handle. Since its counterpart for extent-based
file is ext4_ext_get_blocks(), rename it to be ext4_ind_get_blocks().
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The function ext4_da_get_block_write() is called in exactly one write,
and the last argument, create, is always 1. Remove it to simplify the
code slightly.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
On UP systems without DEBUG_SPINLOCK, ext4_is_group_locked always fails
which triggers a BUG_ON() call.
This patch fixes it by using assert_spin_locked instead.
Signed-off-by: Vincent Minet <vincent@vincent-minet.net>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
We have sb_bgl_lock() and ext4_group_info.bb_state
bit spinlock to protech group information. The later is only
used within mballoc code. Consolidate them to use sb_bgl_lock().
This makes the mballoc.c code much simpler and also avoid
confusion with two locks protecting same info.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
If the file's blocks have not yet been allocated because of delayed
allocation, the length of the extent returned by fiemap is incorrect.
This commit fixes this bug.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Carl Henrik Lunde reported and debugged this; the test for the
last allocated block was comparing bytes to blocks in this test:
if (logical + length - 1 == EXT_MAX_BLOCK ||
ext4_ext_next_allocated_block(path) == EXT_MAX_BLOCK)
flags |= FIEMAP_EXTENT_LAST;
so any extent which ended right at 4G was stopping the extent
walk. Just replacing these values with the extent block &
length should fix it.
Also give blksize_bits a saner type, and reverse the order
of the tests to make the more likely case tested first.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reported-by: Carl Henrik Lunde <chlunde@ping.uio.no>
Tested-by: Carl Henrik Lunde <chlunde@ping.uio.no>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
In memory-constrained systems with many partitions, the ~68K for each
partition for the mb_history buffer can be excessive.
This patch adds a new mount option, mb_history_length, as well as a
way of setting the default via a module parameter (or via a sysfs
parameter in /sys/module/ext4/parameter/default_mb_history_length).
If the mb_history_length is set to zero, the mb_history facility is
disabled entirely.
Signed-off-by: Curt Wohlgemuth <curtw@google.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The fs/ext4/namei.h header file had only a single function
declaration, and should have never been a standalone file. Move it
into ext4.h, where should have been from the beginning.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
There is no longer a reason for a separate ext4_sb.h header file, so
move it into ext4.h just to make life easier for developers to find
the relevant data structures and typedefs. Should also speed up
compiles slightly, too.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
There is no longer a reason for a separate ext4_i.h header file, so
move it into ext4.h just to make life easier for developers to find
the relevant data structures and typedefs. Should also speed up
compiles slightly, too.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
By avoiding the use of not-yet-used block groups (i.e., block groups
with the BLOCK_UNINIT flag), mballoc had a tendency to create large
files with large non-contiguous gaps. In addition avoiding the use of
new block groups had a tendency to push regular file data into the
first block group in a flex_bg group, which slows down the speed of
e2fsck pass 2, since it has a tendency to seek much more. For
example:
Before Patch After Patch
Time in seconds Time in seconds
Real / User/ Sys MB/s Real / User/ Sys MB/s
Pass 1 8.52 / 2.21 / 0.46 20.43 8.84 / 4.97 / 1.11 19.68
Pass 2 21.16 / 1.02 / 1.86 11.30 6.54 / 1.77 / 1.78 36.39
Pass 3 0.01 / 0.00 / 0.00 139.00 0.01 / 0.01 / 0.00 128.90
Pass 4 0.16 / 0.15 / 0.00 0.00 0.17 / 0.17 / 0.00 0.00
Pass 5 2.52 / 1.99 / 0.09 0.79 2.31 / 1.78 / 0.06 0.86
Total 32.40 / 5.11 / 2.49 12.81 17.99 / 8.75 / 2.98 23.01
This was on a sample 80 gig root filesystem which was approximately
50% full. Note the improved e2fsck pass 2 performance, by over a
factor of 3, due to a decreased number of seeks. (The total amount of
I/O in pass 2 was unchanged; the layout of the directory blocks was
simply much better from e2fsck's's perspective.)
Other changes as a result of this patch on this sample filesystem:
Before Patch After Patch
# of non-contig files 762 779
# of non-contig directories 571 570
# of BLOCK_UNINIT bg's 307 293
# of INODE_UNINIT bg's 503 503
Out of 640 block groups, of which 333 were in use, this patch caused
an extra 14 block groups to be utilized. The number of non-contiguous
files did go up slightly, but when measured against the 99.9% of the
files (603,154) which were contiguously allocated, this is pretty
insignificant.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: Andreas Dilger <adilger@sun.com>
If a filesystem supports POSIX ACL's, the VFS layer expects the filesystem
to do POSIX ACL checks on any files not owned by the caller, and it does
this for every single pathname component that it looks up.
That obviously can be pretty expensive if the filesystem isn't careful
about it, especially with locking. That's doubly sad, since the common
case tends to be that there are no ACL's associated with the files in
question.
ext4 already caches the ACL data so that it doesn't have to look it up
over and over again, but it does so by taking the inode->i_lock spinlock
on every lookup. Which is a noticeable overhead even if it's a private
lock, especially on CPU's where the serialization is expensive (eg Intel
Netburst aka 'P4').
For the special case of not actually having any ACL's, all that locking is
unnecessary. Even if somebody else were to be changing the ACL's on
another CPU, we simply don't care - if we've seen a NULL ACL, we might as
well use it.
So just load the ACL speculatively without any locking, and if it was
NULL, just use it. If it's non-NULL (either because we had a cached
entry, or because the cache hasn't been filled in at all), it means that
we'll need to get the lock and re-load it properly.
(This commit was ported from a patch originally authored by Linus for
ext3.)
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Use a separate lock to protect s_groups_count and the other block
group descriptors which get changed via an on-line resize operation,
so we can stop overloading the use of lock_super().
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The function ext4_mark_recovery_complete() is called from two call
paths: either (a) while mounting the filesystem, in which case there's
no danger of any other CPU calling write_super() until the mount is
completed, and (b) while remounting the filesystem read-write, in
which case the fs core has already locked the superblock. This also
allows us to take out a very vile unlock_super()/lock_super() pair in
ext4_remount().
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
ext4_fill_super() is no longer called by read_super(), and it is no
longer called with the superblock locked. The
unlock_super()/lock_super() is no longer present, so this comment is
entirely superfluous.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Ext4's on-line resizing adds a new block group and then, only at the
last step adjusts s_groups_count. However, it's possible on SMP
systems that another CPU could see the updated the s_group_count and
not see the newly initialized data structures for the just-added block
group. For this reason, it's important to insert a SMP read barrier
after reading s_groups_count and before reading any (for example) the
new block group descriptors allowed by the increased value of
s_groups_count.
Unfortunately, we rather blatently violate this locking protocol
documented in fs/ext4/resize.c. Fortunately, (1) on-line resizes
happen relatively rarely, and (2) it seems rare that the filesystem
code will immediately try to use just-added block group before any
memory ordering issues resolve themselves. So apparently problems
here are relatively hard to hit, since ext3 has been vulnerable to the
same issue for years with no one apparently complaining.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
By using a separate super_operations structure for filesystems that
have and don't have journals, we can simply ext4_write_super() ---
which is only needed when no journal is present --- and ext4_freeze(),
ext4_unfreeze(), and ext4_sync_fs(), which are only needed when the
journal is present.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The s_dirt flag wasn't completely handled correctly, but it didn't
really matter when journalling was enabled. It turns out that when
ext4 runs without a journal, we don't clear s_dirt in places where we
should have, with the result that the high-level write_super()
function was writing the superblock when it wasn't necessary.
So we fix this by making ext4_commit_super() clear the s_dirt flag,
and removing many of the other places where s_dirt is manipulated.
When journalling is enabled, the s_dirt flag might be left set more
often, but s_dirt really doesn't matter when journalling is enabled.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The ext4_commit_super() function took both a struct super_block * and
a struct ext4_super_block *, but the struct ext4_super_block can be
derived from the struct super_block.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
For very large filesystems, the s_flex_groups array can get quite big.
For example, a filesystem that can be resized up to 16TB will have
8192 flex groups (assuming the default flex_bg size of 16), so the
array is 96k, which is *very* marginal for kmalloc(). On the other
hand, a 160GB filesystem without the resize_inode feature will only
require 960 bytes. So we try to allocate the array first using
kmalloc(), and if that fails, we'll try to use vmalloc() instead.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Setting BH_Unwritten buffer_heads as BH_Mapped avoids multiple
(unnecessary) calls to get_block() during the call to the write(2)
system call. Setting BH_Unwritten buffer heads as BH_Mapped requires
that the writepages() functions can handle BH_Unwritten buffer_heads.
After this commit, things work as follows:
ext4_ext_get_block() returns unmapped, unwritten, buffer head when
called with create = 0 for prealloc space. This makes sure we handle
the read path and non-delayed allocation case correctly. Even though
the buffer head is marked unmapped we have valid b_blocknr and b_bdev
values in the buffer_head.
ext4_da_get_block_prep() called for block resrevation will now return
mapped, unwritten, new buffer_head for prealloc space. This avoids
multiple calls to get_block() for write to same offset. By making such
buffers as BH_New, we also assure that sub-block zeroing of buffered
writes happens correctly.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
These struct buffer_heads are allocated on the stack (and hence are
initialized with stack garbage). They are only used to call a
get_blocks() function, so that's mostly OK, but b_state must be
initialized to be 0 so we don't have any unexpected BH_* flags set by
accident, such as BH_Unwritten or BH_Delay.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The EXTENTS_FL flag should never be set on special files, but if it
is, don't bother trying to validate that the extents tree is valid,
since only files, directories, and non-fast symlinks will ever have an
extent data structure. We perhaps should flag the filesystem as being
corrupted if we see a special file (named pipes, device nodes, Unix
domain sockets, etc.) with the EXTENTS_FL flag, but e2fsck doesn't
currently check this case, so we'll just ignore this for now, since
it's harmless.
Without this fix, a special device with the extents flag is flagged as
an error by the kernel, so it is impossible to access or delete the
inode, but e2fsck doesn't see it as a problem, leading to
confused/frustrated users.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Don't try to look at i_file_acl_high unless the INCOMPAT_64BIT feature
bit is set. The field is normally zero, but older versions of e2fsck
didn't automatically check to make sure of this, so in the spirit of
"be liberal in what you accept", don't look at i_file_acl_high unless
we are using a 64-bit filesystem.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
If the block containing external extended attributes (which is stored
in i_file_acl and i_file_acl_high) is larger than the on-disk
filesystem, the process which tried to access the extended attributes
will endlessly issue kernel printks complaining that
"__find_get_block_slow() failed", locking up that CPU until the system
is forcibly rebooted.
So when we read in the inode, make sure the i_file_acl value is legal,
and if not, flag the filesystem as being corrupted.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Instead of just checking that the extent block number is greater or
equal than s_first_data_block, make sure it it is not pointing into
the block group descriptors, since that is clearly wrong. This helps
prevent filesystem from getting very badly corrupted in case an extent
block is corrupted.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Missing braces caused the warning to print more than once.
Signed-Off-By: Chuck Ebbert <cebbert@redhat.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Theodore Ts'o