In btrfs_set_block_flags() we want to check if the slot
in the leaf points to the first item in the leaf - if it
doesn't check if the previous item in the leaf is an extent
item. By removing this extra slot decrement we are indeed
checking the item right before the slot, and not the second
item before.
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Reviewed-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
If we did a tree search with the goal to find a metadata item
but the search failed with return value 1, we attempt to see
if in the same leaf there's a corresponding extent item, and if
there's one, just use it instead of doing another tree search
for this extent item. The check in the leaf was wrong because
it was seeking for a metadata item instead of an extent item.
This optimization was also being triggered incorrectly, as it
was evaluating path->slots which always evaluates to true. The
goal was to see if the leaf level slot was greater than zero
(i.e. not the first item in the leaf).
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Reviewed-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
Several function return values were being completely
ignored.
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
extent-tree.c: In function 'btrfs_free_block_groups':
extent-tree.c:3190:12: warning: cast to pointer from integer of
different size [-Wint-to-pointer-cast]
Signed-off-by: Wang Shilong <wangsl-fnst@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
extent_ref_type() contains inconsequential differences between
kernelspace and userspace, and has since the initial commits
to each. Just make userspace look like kernelspace.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
div_factor has been implemented for two times, cleanup it.
And I move them into a independent file named math.h because they are
common math functions.
[Eric Sandeen: port kernel commit 3fed40c to userspace]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
Remove some commented-out & #if 0'd code:
* close_blocks()
* btrfs_drop_snapshot()
* btrfs_realloc_node()
* btrfs_find_dead_roots()
There are still some #if 0'd functions in there, but I'm hedging
on those for now, they have been copied to cmds-check.c and I want
to see if they can be brough back into ctree.c eventually.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
Add chunk-recover program to check or rebuild chunk tree when the system
chunk array or chunk tree is broken.
Due to the importance of the system chunk array and chunk tree, if one of
them is broken, the whole btrfs will be broken even other data are OK.
But we have some hint(fsid, checksum...) to salvage the old metadata.
So this function will first scan the whole file system and collect the
needed data(chunk/block group/dev extent), and check for the references
between them. If the references are OK, the chunk tree can be rebuilt and
luckily the file system will be mountable.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
The allocator looks for these hints when moving on to another block group which
will make it reset which block group it looks at, when we've already searched
that block group and didn't find any space to allocate, we need to fix this by
just letting the allocator make the determination if the block group is good
enough. This also fixes a problem where if we couldn't find space in the block
group we were given we'd just error out instead of moving on to the next block
group. Previously I couldn't fix some file systems that were relatively full,
but with this patch I can now run fsck on them with no allocation errors.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
In some cases the extent tree can just be so gone there is no point in trying to
figure out how to put it back together. So add a --init-extent-tree mode which
will zero out the extent tree and then re-add extents for all of the blocks we
find. This will also undo any balance that was going on at the time of the
crash, this is needed because the reloc tree seems to confuse fsck at the
moment. With this patch I can put back together a users file system that was
completely gone. Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
The tree log bug I introduced could create inconsistent file extent entries in
the file system tree and in some worst cases even create multiple extent entries
for the same entry. To fix this we need to do a few things
1) Keep track of extent items that overlap and then pick the one that covers the
largest area and delete the rest of the items.
2) Keep track of file extent items that land in extent items but don't match
disk_bytenr/disk_num_bytes exactly. Once we find these we need to figure out
who is the right ref and then fix all of the other refs to agree.
Each of these cases require a complete rescan of all of the extents, so
unfortunately if you hit this particular problem the fsck is going to take quite
a while since it will likely rescan all the trees 2 or 3 times. With this patch
the broken file system a user sent me is fixed and a broken file system that was
created by my reproducer is also fixed. Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
In trying to track down a weird tree log problem I wanted to make sure that the
free space cache was actually valid, which we currently have no way of doing.
So this patch adds a bunch of support for the free space cache code and then a
checker to fsck. Basically we go through and if we can actually load the free
space cache then we will walk the extent tree and verify that the free space
cache exactly matches what is in the extent tree. Hopefully this will always be
correct, the only time it wouldn't is if the extent tree is corrupt or we have
some sort of awful bug in the free space cache. Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
This fixes up the progs to properly deal with skinny metadata. This adds the -x
option to mkfs and btrfstune for enabling the skinny metadata option. This also
makes changes to fsck so it can properly deal with the skinny metadata entries.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
I've been working on btrfs-image and I kept seeing these leaks pop up on
valgrind so I'm just fixing them. We don't properly cleanup the device cache,
the chunk tree mapping cache, or the space infos on close. With this patch
valgrind doesn't complain about any memory leaks running btrfs-image. Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Allocate fs_info::super_copy dynamically of full BTRFS_SUPER_INFO_SIZE
and use it directly for saving superblock to disk.
This fixes incorrect superblock checksum after mkfs.
Signed-off-by: David Sterba <dsterba@suse.cz>
David Woodhouse originally contributed this code, and Chris Mason
changed it around to reflect the current design goals for raid56.
The original code expected all metadata and data writes to be full
stripes. This meant metadata block size == stripe size, and had a few
other restrictions.
This version allows metadata blocks smaller than the stripe size. It
implements both raid5 and raid6, although it does not have code to
rebuild from parity if one of the drives is missing or incorrect.
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
This was a bug from long time ago that never actually got fixed. We start
with bytenr 0 when looping through all of the block groups, but
btrfs_lookup_block_group will bail out since it couldn't find a block group
with 0 as the bytenr. Btrfs_lookup_first_block_group will be nice and
adjust the start up to the right value, so this way we reset all the block
groups properly and not screw up the users block group accounting. Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
We need to align earlier to make sure we're getting things
properly setup against the raid56 stripes.
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
This is mostly disabled, but it is step one in handling
corrupted block groups in the extent allocation tree.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
During btrfsck --repair, we make an index of extents that have incorrect
reference counts. Once we've collect the whole index, we go through
and modify the extent allocation tree to reflect the correct results.
Changing the extent allocation tree may free blocks, and so it may
end up removing a block that had a missing reference structure. The
fsck code may then circle back around and add the reference back.
The result is an extent that isn't actually used, but is recorded in the
extent allocation tree.
This commit adds a hook called as extents are freed. The hook searches
the index of incorrect references and updates it to reflect the freeing
of the extent.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The block group accounting is fixed after we check the extent back
references. This makes sure the accounting is fixed unless we
were not able to repair the backrefs.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The code that corrects the count of bytes used in each block group
was only marking block groups dirty when they contained extents. This
fixes things to dirty all the block groups, so any empty block groups
are written with their correct (zero) count.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This also includes a new --repair btrfsck option. For now it can
only fix errors in the extent allocation tree.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
When creating a mixed fs with mkfs, an extra metadata chunk got allocated.
This is because btrfs_reserve_extent calls do_chunk_alloc for METADATA,
which in turn wasn't able to find the proper space_info, as __find_space_info
did a hard compare of the flags. It is now sufficient for the space_info to
include the proper flag. This reflects the change done to the kernel code
to support mixed chunks.
Also for a subsequent chunk allocation (which should not be hit in the mkfs
case), the chunk is now created with the flags from the space_info instead
of the requested flags. A better solution would be to pull the full changeset
for the mixed case from the kernel into the user mode (or, even better, share
the code)
The additional chunk probably confused block_rsv calculation, which in turn
led to severeal ENOSPC Oopses.
Signed-off-by: Arne Jansen <sensille@gmx.net>
Signed-off-by: Hugo Mills <hugo@carfax.org.uk>
When building on ppc64 I hit a number of warnings in printf:
btrfs-map-logical.c:69: error: format ‘%Lu’ expects type ‘long long
unsigned int’, but argument 4 has type ‘u64’
Fix them.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Hugo Mills <hugo@carfax.org.uk>
gcc-4.6 (as shipped in Fedora) turns on -Wunused-but-set-variable by
default, which breaks the build when combined with -Wall, e.g.:
debug-tree.c: In function ‘print_extent_leaf’:
debug-tree.c:45:13: error: variable ‘last_len’ set but not used [-Werror=unused-but-set-variable]
debug-tree.c:44:13: error: variable ‘last’ set but not used [-Werror=unused-but-set-variable]
debug-tree.c:41:21: error: variable ‘item’ set but not used [-Werror=unused-but-set-variable]
cc1: all warnings being treated as errors
This patch fixes the errors by removing the unused variables.
Signed-off-by: Chris Ball <cjb@laptop.org>
Signed-off-by: Hugo Mills <hugo@carfax.org.uk>
When raid is setup with mkfs, it is supposed to cow the initial filesystem
it creates up to the desired raid level. RAID10 was not in the list
of RAID levels it checked for, so the initial FS created for RAID10
actually only lived on the first disk.
This works well enough because all the roots get quickly cowed during the
first mount. The exception is the data relocation tree, which only gets
cowed when we do a balance.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
lookup_inline_extent_backref only checks for duplicate backref for data extent.
It assumes backrefs for tree block never conflict. This patch makes
lookup_inline_extent_backref check duplicate backrefs for both data and tree
block, so that we can detect potential bug earlier.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Write dirty block groups may make some block groups dirty.
This patch make btrfs_write_dirty_block_groups properly
handle the recursion.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This commit introduces a new kind of back reference for btrfs metadata.
Once a filesystem has been mounted with this commit, IT WILL NO LONGER
BE MOUNTABLE BY OLDER KERNELS.
The new back ref provides information about pointer's key, level and in which
tree the pointer lives. This information allow us to find the pointer by
searching the tree. The shortcoming of the new back ref is that it only works
for pointers in tree blocks referenced by their owner trees.
This is mostly a problem for snapshots, where resolving one of these fuzzy back
references would be O(number_of_snapshots) and quite slow. The solution used
here is to use the fuzzy back references in the common case where a given tree
block is only referenced by one root, and use the full back references when
multiple roots have a reference
There are still some warnings of the form:
format '%llu' expects type 'long long unsigned int' but argument has type 'u64'
In conjunction with -Werror, this is causing some build failures.
Now they're properly casted, avoiding compiler warnings.
Signed-off-by: Luca Bruno <lucab@debian.org>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
This patch makes btrfsck check more things, including
directory items, file extents, checksumming, inode link
counts etc.
The code for these checks is similar to the code verifies
extent back references. The main difference is that
shared tree blocks are treated specially. The partial
checking results(unresolved references and/or errors)
of shared sub-trees are cached. This avoids scanning
the shared blocks several times. Thank you,
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
This patch updates the ext3 to btrfs converter for the new
disk format. This mainly involves changing the convert's
data relocation and free space management code. This patch
also ports some functions from kernel module to btrfs-progs.
Thank you,
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
This patch updates btrfs-progs for superblock duplication.
Note: I didn't make this patch as complete as the one for
kernel since updating the converter requires changing the
code again. Thank you,
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
This patch does the following:
1) Update device management code to match the kernel code.
2) Allocator fixes.
3) Add a program called btrfstune to set/clear the SEEDING
super block flags.
The offset field in struct btrfs_extent_ref records the position
inside file that file extent is referenced by. In the new back
reference system, tree leaves holding reference to file extent
are recorded explicitly. We can quickly scan these tree leaves, so the
offset field is not required.
This patch also makes the back reference system check the objectid
when extents are being deleted
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
This patch makes the back reference system to explicit record the
location of parent node for all types of extents. The location of
parent node is placed into the offset field of backref key. Every
time a tree block is balanced, the back references for the affected
lower level extents are updated.
This patch improves converter's allocator and fixes a bug in data relocation
function. The new allocator caches free blocks as Btrfs's default allocator.
In testing here, the user CPU time reduced to half of the original when
checksum and small file packing was disabled. This patch also enlarges the
size of block groups created by the converter.
balance level starts by trying to empty the middle block, and then
pushes from the right to the middle. This might empty the right block
and leave a small number of pointers in the middle.
The main changes in this patch are adding chunk handing and data relocation
ability. In the last step of conversion, the converter relocates data in system
chunk and move chunk tree into system chunk. In the rollback process, the
converter remove chunk tree from system chunk and copy data back.
Regards
YZ
---
Block headers now store the chunk tree uuid
Chunk items records the device uuid for each stripes
Device extent items record better back refs to the chunk tree
Block groups record better back refs to the chunk tree
The chunk tree format has also changed. The objectid of BTRFS_CHUNK_ITEM_KEY
used to be the logical offset of the chunk. Now it is a chunk tree id,
with the logical offset being stored in the offset field of the key.
This allows a single chunk tree to record multiple logical address spaces,
upping the number of bytes indexed by a chunk tree from 2^64 to
2^128.
When a block is freed, it can be immediately reused if it is from
the current transaction. But, an extra check is required to make sure
the block had not been written yet. If it were reused after being written,
the transid in the block header might match the transid of the
next time the block was allocated.
The parent node records the transaction ID of the block it is pointing to,
and this is used as part of validating the block on reads. So, there
can only be one version of a block per transaction.
The mkfs code bootstraps the filesystem on a single device. Once
the raid block groups are setup, it needs to recow all of the blocks so
that each tree is properly allocated.
