This reuses the existing code for checking the free space cache, we just
need to load the free space tree. While we do that, we check a couple of
invariants on the free space tree itself. This requires pulling in some
code from the kernel to exclude the super stripes.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
If a file is linked from more than one directory and only one
of the links is corrupted, btrfs check dose not reset the nlink
properly. Actually it can go into infinite loop to link the broken file
into lost+found.
This patch fix two part of the code. The first one delay the freeing
valid (no error, found inode ref, directory index, and directory
item) backrefs. Freeing valid backrefs earier prevent reset_nlink() to
add back all valid links.
The second fix is obvious: passing `ref_type' to btrfs_add_link() is just
wrong. It should be `filetype' instead. The current code can break all valid
file links.
Signed-off-by: Naohiro Aota <naota@elisp.net>
Reviewed-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
In process_extent_item(), it gives 'metadata' initial value 0, but for
non-skinny-metadata case, metadata extent can't be judged just from key
type and it forgot that case.
This causes a lot of false alert in non-skinny-metadata filesystem.
Fix it by set correct metadata value before calling add_extent_rec().
Reported-by: Christoph Anton Mitterer <calestyo@scientia.net>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Since we have btrfs_close_all_devices() in btrfs's main entrance,
it is not necessary to call btrfs_close_all_devices() separately
in each sub-command.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The same code is executed when the condition "ret" is true or false,
because the code in the if-then branch and after the if statement is
identical.
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The variable "err" is assigned to "ret" then "ret" gets overwritten by
check_extent_refs() before "ret" can be used. Reported by Coverity.
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This patch is generated from a coccinelle semantic patch:
identifier t;
expression e;
statement s;
@@
-t = malloc(e);
+t = calloc(1, e);
(
if (!t) s
|
if (t == NULL) s
|
)
-memset(t, 0, e);
Signed-off-by: Silvio Fricke <silvio.fricke@gmail.com>
[squashed patches into one]
Signed-off-by: David Sterba <dsterba@suse.com>
There is a bug in btrfs-convert in 4.1.2, even we don't allow mixed
block group for converted image, btrfs-convert will still create image
with data and metadata inside one chunk.
And further more, the chunk type is still DATA or METADATA, not
DATA|METADATA (not mixed).
So add btrfsck check for it right now.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Added a missing newline to some error messages.
Also printf() was changed to fprintf(stderr) for error messages.
Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Reviewed-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
mkfs creates more than one fs_devices in fs_uuids.
1: one is for file system being created
2: others are created in test_dev_for_mkfs in order to check mount point
test_dev_for_mkfs()-> ... -> btrfs_scan_one_device()
Current code only closes 1, and this patch also closes in case 2.
Similar problem exist in other tools, eg.::
cmd-check.c: the function is:
cmd_check()->check_mounted()-> ... -> btrfs_scan_one_device()
...
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
rec->crossing_stripes is not initialized in allocate place,
and have possibility causing wrong report for normal tree block.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We need not check path before btrfs_free_path() is called because
path is checked in btrfs_free_path().
Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
For a special case, discount file extent repair function will cause
infinite loop.
The case is, if the file loses all its extents, we won't have a hole
to fill, causing repair function doing nothing. Since the
I_ERR_DISCOUNT doesn't disappear, fsck will do an infinite loop.
For such case, just puch hole to fill the whole range to fix it.
Reported-by: Robert Munteanu <robert.munteanu@gmail.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
If a file lost all its extents, fsck will unable to print out the hole.
Add an extra check to print out the hole range.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Kernel btrfs_map_block() function has a limitation that it can only
map BTRFS_STRIPE_LEN size.
That will cause scrub fails to scrub tree block which crosses strip
boundary, causing BUG_ON().
Normally, it's OK as metadata is always in metadata chunk and
BTRFS_STRIPE_LEN can always be divided by node/leaf size.
So without mixed block group, tree block won't cross stripe boundary.
But for mixed block group, especially for btrfs converted from ext4,
it's almost sure one or more tree blocks are not aligned with node size
and cross stripe boundary.
Causing bug with kernel scrub.
This patch will report the problem, although we don't have a good idea
how to fix it in user space until we add the ability to relocate tree
block in user space.
Also, kernel code should also be checked for such tree block alloc
problems.
Reported-by: Chris Murphy <lists@colorremedies.com>
Reported-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Some unknown kernel bug makes inode nbytes modification out of sync with
file extent update.
But it's quite easy to fix in btrfs-progs anyway.
So just fix it by adding a new function repair_inode_nbytes by using the
found_size in inode_record.
Reported-by: Christian <cdysthe@gmail.com>
Reported-by: Chris Murphy <lists@colorremedies.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
In copy_inode_rec(), a shallow copy happens on rec->holes rb_root.
So for shared inode case, new rec->holes still points to old rb_root,
and when the old inode record is freed, the new inode_rec->holes will
points to garbage and cause segfault when we try to free new
inode_rec->holes.
Fix it by calling copy_file_extent_holes() to do deep copy.
Reported-by: Eric Sandeen <sandeen@redhat.com>
Reported-by: Filipe David Manana <fdmanana@gmail.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
make[1]: Nothing to be done for `all'.
cmds-check.c: In function ‘del_file_extent_hole’:
cmds-check.c:289:26: warning: ‘prev.start’ may be used uninitialized in this function
cmds-check.c:289:26: warning: ‘prev.len’ may be used uninitialized in this function
cmds-check.c:290:26: warning: ‘next.start’ may be used uninitialized in this function
cmds-check.c:290:26: warning: ‘next.len’ may be used uninitialized in this function
Reported-by: Anand Jain <Anand.Jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
We're not using it anywhere. The best practice is to add enums with
values > 255 for the long options, option index counting is error prone.
Signed-off-by: David Sterba <dsterba@suse.cz>
Before this patch, csum tree rebuild will not work with extent tree
rebuild, since extent tree rebuild will only build up basic block
groups, but csum tree rebuild needs data extents to rebuild.
So if one use btrfsck with --init-csum-tree and --init-extent-tree, csum
tree will be empty and tons of "missing csum" error will be outputted.
This patch allows csum tree rebuild get its data from fs/subvol trees
using regular file extents (which is also the only one using csum tree
currently).
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[renamed to fill_csum_tree_from_one_fs_root]
Signed-off-by: David Sterba <dsterba@suse.cz>
There is a case that can cause nlink fix function.
For example, lost+found dir already has the following files:
---------------------------
|ino |filename |
|-------------------------|
|258 |normal_file |
|259 |normal_file.260 |
---------------------------
The next inode to be fixed is the following:
---------------------------
|260 |normail_file |
---------------------------
And when trying to move inode to lost+found dir, its file name conflicts
with inode 258, and even add ".INO" suffix, it still conflicts with
inode 259.
Since the move failed, the LINK_COUNT_ERR flag is not cleared, the inode
record will not be freed, btrfsck will try fix it again and again,
causing the infinite loop.
The patch will first change the ".INO" suffix naming to a loop behavior,
and clear the LINK_COUNT_ERR flag anyway to avoid infinite loop.
Reported-by: Naohiro Aota <naota@elisp.net>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
We can have FULL_BACKREF set or not set when we need the opposite, this patch
fixes this problem by setting a bit when the flag is set improperly. This way
we can either correct the problem when we re-create the extent item if the
backrefs are also wrong, or we can just set the flag properly in the extent
item. Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
We don't want to keep extent records pinned down if we fix stuff as we may need
the space and we can be pretty sure that these records are correct. Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
We hold a transaction open for the entirety of fixing extent refs. This works
out ok most of the time but we can be tight on space and run out of space when
fixing things. To get around this just push down the transaction starting dance
into the functions that actually fix things. This keeps us from ending up with
ENOSPC because we pinned everything and allows the code to be a bit simpler.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
The METADUMP super flag makes us skip doing the chunk tree reading which isn't
helpful for the new restore since we have a valid chunk tree. But we still want
to have a way for the kernel to know that this is a metadump restore so it
doesn't do things like verify data checksums. We also want to skip some of the
device extent checks in fsck since those will obviously not match. Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
The data reloc root is weird with it's csums. It'll copy an entire extent and
then log any csums it finds, which makes it look weird when it comes to prealloc
extents. So just skip the data reloc tree, it's special and we just don't need
to worry about it. Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
We have logic to fix the root locations for roots in response to a corruption
bug we had earlier. However this work doesn't apply to reloc roots and can
screw things up worse, so make sure we skip any reloc roots that we find.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
If we fix bad blocks during run_next_block we will return -EAGAIN to loop around
and start again. The deal_with_roots work messed up this handling, this patch
fixes it. With this patch we can properly deal with broken tree blocks.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
Previously we used to just set FULL_BACKREF if we couldn't lookup an extent info
for an extent. Now we just bail out if we can't lookup the extent info, which
is less than good since fsck is supposed to fix these very problems. So instead
figure out the flag we are supposed to use and pass that along instead. This
patch also provides a test image to test this functionality. Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
Allow read_tree_block() and read_node_slot() to return error pointer.
This should help caller to get more specified error number.
For existing callers, change (!eb) judgmentt to
(!extent_buffer_uptodate(eb)) to keep the compatibility, and for caller
missing the check, use PTR_ERR(eb) if possible.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Since orphan extents are handled in previous patches, now just punch
holes to fill the file extents hole.
Also since now btrfsck is aware of whether the inode has orphan file
extent, allow repair_inode_no_item() to determine filetype according to
orphan file extent.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
David Sterba