platform/upstream/btrfs-progs.git
7 years agobtrfs-progs: convert: move migrate_reserved_ranges
David Sterba [Tue, 23 Aug 2016 15:43:21 +0000 (17:43 +0200)]
btrfs-progs: convert: move migrate_reserved_ranges

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert: move migrate_one_reserved_range
David Sterba [Tue, 23 Aug 2016 15:42:23 +0000 (17:42 +0200)]
btrfs-progs: convert: move migrate_one_reserved_range

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert: move create_image_file_range
David Sterba [Tue, 23 Aug 2016 15:42:01 +0000 (17:42 +0200)]
btrfs-progs: convert: move create_image_file_range

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert: move block_iterate_proc
David Sterba [Tue, 23 Aug 2016 15:41:20 +0000 (17:41 +0200)]
btrfs-progs: convert: move block_iterate_proc

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert: move record_file_blocks
David Sterba [Tue, 23 Aug 2016 15:41:00 +0000 (17:41 +0200)]
btrfs-progs: convert: move record_file_blocks

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert: move init_blk_iterate_data
David Sterba [Tue, 23 Aug 2016 15:40:08 +0000 (17:40 +0200)]
btrfs-progs: convert: move init_blk_iterate_data

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert: move csum_disk_extent
David Sterba [Tue, 23 Aug 2016 15:38:55 +0000 (17:38 +0200)]
btrfs-progs: convert: move csum_disk_extent

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert: move read_disk_extent
David Sterba [Tue, 23 Aug 2016 15:36:13 +0000 (17:36 +0200)]
btrfs-progs: convert: move read_disk_extent

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert: move convert_insert_dirent
David Sterba [Tue, 23 Aug 2016 15:35:05 +0000 (17:35 +0200)]
btrfs-progs: convert: move convert_insert_dirent

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert: move intersect_with_sb
David Sterba [Tue, 23 Aug 2016 15:34:43 +0000 (17:34 +0200)]
btrfs-progs: convert: move intersect_with_sb

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert: prefix all ext2-related helpers
David Sterba [Tue, 23 Aug 2016 15:31:34 +0000 (17:31 +0200)]
btrfs-progs: convert: prefix all ext2-related helpers

Add ext2_ prefix to all functions that seem to be ext2-specific.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: build: extend convert options
David Sterba [Tue, 23 Aug 2016 15:21:30 +0000 (17:21 +0200)]
btrfs-progs: build: extend convert options

Add --with-convert[=VALUE] option to configure. Accepts ext2, auto, yes,
or no, but will be extended to more in the future. The configure-time
defines are not used in the code, ext2 is built-in unconditionally.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: build: add per-tool cflags variable
David Sterba [Tue, 23 Aug 2016 16:04:37 +0000 (18:04 +0200)]
btrfs-progs: build: add per-tool cflags variable

The standalone tools are built from pattern rules, add support for
per-tool cflaags, like btrfs_something_clfags.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: mkfs: cleanup, open_ctree to fs_info and use directly
David Sterba [Tue, 23 Aug 2016 14:18:33 +0000 (16:18 +0200)]
btrfs-progs: mkfs: cleanup, open_ctree to fs_info and use directly

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: mkfs: kill print_version
David Sterba [Tue, 23 Aug 2016 14:12:36 +0000 (16:12 +0200)]
btrfs-progs: mkfs: kill print_version

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: mkfs: clean up make_image
David Sterba [Mon, 22 Aug 2016 17:14:20 +0000 (19:14 +0200)]
btrfs-progs: mkfs: clean up make_image

The exact errors are printed, the removed message does not seem to be
necessary. Return proper errors.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: mkfs: help and usage now to to stdout
David Sterba [Mon, 22 Aug 2016 16:18:54 +0000 (18:18 +0200)]
btrfs-progs: mkfs: help and usage now to to stdout

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: mkfs: handle and report transaction commit failures
David Sterba [Mon, 22 Aug 2016 16:18:14 +0000 (18:18 +0200)]
btrfs-progs: mkfs: handle and report transaction commit failures

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: mkfs: switch BUG_ON to error handling in traverse_directory
David Sterba [Mon, 22 Aug 2016 16:10:43 +0000 (18:10 +0200)]
btrfs-progs: mkfs: switch BUG_ON to error handling in traverse_directory

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: document all btrfs_open_ctree_flags
David Sterba [Mon, 22 Aug 2016 16:08:13 +0000 (18:08 +0200)]
btrfs-progs: document all btrfs_open_ctree_flags

Document and add unsigned type to the values.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: mkfs: improve error handling in recow_roots
David Sterba [Mon, 22 Aug 2016 15:55:16 +0000 (17:55 +0200)]
btrfs-progs: mkfs: improve error handling in recow_roots

Return and handle errors in the callchain.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: mkfs: improve error handling in main()
David Sterba [Mon, 22 Aug 2016 15:50:49 +0000 (17:50 +0200)]
btrfs-progs: mkfs: improve error handling in main()

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: mkfs: return errors from block group creation functions
David Sterba [Mon, 22 Aug 2016 14:57:15 +0000 (16:57 +0200)]
btrfs-progs: mkfs: return errors from block group creation functions

No more BUG_ONs, we don't care about cleanup as the filesystem is
supposed to be marked as partial.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: two staged filesystem creation
David Sterba [Mon, 22 Aug 2016 14:32:24 +0000 (16:32 +0200)]
btrfs-progs: two staged filesystem creation

The filesystem existence on a device is manifested by the signature,
during the mkfs process we write it first and then create other
structures. Such filesystem is not valid and should not be registered
during device scan nor listed among devices from blkid.

This patch will introduce two staged creation. In the first phase, the
signature is wrong, but recognized as a partially created filesystem (by
open or scan helpers). Once we successfully create and write everything,
we fixup the signature. At this point automated scanning should find
a valid filesystem on all devices.

We can also rely on the partially created filesystem to do better error
handling during creation. We can just bail out and do not need to clean
up.

The partial signature is '!BHRfS_M', can be shown by

  btrfs inspect-internal dump-super -F image

Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: mkfs: do not scan partially initialized devices
David Sterba [Mon, 22 Aug 2016 14:31:11 +0000 (16:31 +0200)]
btrfs-progs: mkfs: do not scan partially initialized devices

We call scan ioctl on the devices too early, when most of the filesystem
structures are not yet created. Move the registration to the end, after
the filesystem gets closed.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: introduce signature for a partially set up filesystem
David Sterba [Fri, 19 Aug 2016 14:40:14 +0000 (16:40 +0200)]
btrfs-progs: introduce signature for a partially set up filesystem

Currently the superblock is created first, with a valid signaure, but
the rest of the filesystem is missing. When the creation process is
interrupted, the filesystem still might be considered as valid.

To prevent that, create the filesytem with an invalid signature that
would be still recognized during the mkfs process, and finalize at the
end.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: make superblock reading/scanning api more generic
David Sterba [Fri, 19 Aug 2016 14:36:40 +0000 (16:36 +0200)]
btrfs-progs: make superblock reading/scanning api more generic

We'll add more modes that affect scanning.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: pass OPEN_CTREE flags as unsigned
David Sterba [Fri, 19 Aug 2016 14:20:36 +0000 (16:20 +0200)]
btrfs-progs: pass OPEN_CTREE flags as unsigned

As we're passing a set of flags, the enum type is not appropriate.

Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: build: add more debugging features
David Sterba [Fri, 19 Aug 2016 14:06:41 +0000 (16:06 +0200)]
btrfs-progs: build: add more debugging features

Add options to show file and line or stack trace for error/warning
messages that use the common helpers. Possible to let any error stop
execution for ease of analysis and debugging.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: switch ternary op to an if in cmd_subvol_show
David Sterba [Fri, 19 Aug 2016 13:45:04 +0000 (15:45 +0200)]
btrfs-progs: switch ternary op to an if in cmd_subvol_show

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: print-tree: Print hex and human readable root flags
Qu Wenruo [Fri, 19 Aug 2016 08:13:08 +0000 (16:13 +0800)]
btrfs-progs: print-tree: Print hex and human readable root flags

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: print-tree: Print human readable inode flags
Qu Wenruo [Fri, 19 Aug 2016 08:13:07 +0000 (16:13 +0800)]
btrfs-progs: print-tree: Print human readable inode flags

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert-test: Check if the ext2_save/image is read only
Qu Wenruo [Fri, 19 Aug 2016 08:13:06 +0000 (16:13 +0800)]
btrfs-progs: convert-test: Check if the ext2_save/image is read only

Old convert codes uses both 0400 permission and INODE_READONLY flag to
make the converted ext2 image readonly.

While new convert treat the inode just as normal inode, with no special
inode flag and uses 0600 permission.
This makes user able to modify converted image unintentionally and make
rollback fails.

This test case will test the regression.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert: Fix a regression that ext2_save/image is not readonly
Qu Wenruo [Fri, 19 Aug 2016 08:13:05 +0000 (16:13 +0800)]
btrfs-progs: convert: Fix a regression that ext2_save/image is not readonly

The new convert treats the convert image as a normal file, without any
special flags and permissions.

This is different from original code:
1) Permission changed from 0400 to 0600
2) Inode lacks READONLY flag

This makes we can read-write mount the ext2 image and cause rollback
failure.

Follow old code behavior, use 0400 permission and add back READONLY
flag to fix it.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: docs: refer to btrfs(5) from btrfs(8)
David Sterba [Thu, 18 Aug 2016 17:50:06 +0000 (19:50 +0200)]
btrfs-progs: docs: refer to btrfs(5) from btrfs(8)

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: docs: update btrfs-quota manual page, more sections
David Sterba [Thu, 18 Aug 2016 17:25:10 +0000 (19:25 +0200)]
btrfs-progs: docs: update btrfs-quota manual page, more sections

Add more overall sections.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: docs: update btrfs-quota manual page
David Sterba [Thu, 18 Aug 2016 17:25:10 +0000 (19:25 +0200)]
btrfs-progs: docs: update btrfs-quota manual page

Copy the intoductory and usecases from the text written by Arne Jansen,

https://git.kernel.org/cgit/linux/kernel/git/arne/qgroups-doc.git/

The graphics missing for now.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: mkfs: switch to new error message helpers
David Sterba [Thu, 18 Aug 2016 16:38:34 +0000 (18:38 +0200)]
btrfs-progs: mkfs: switch to new error message helpers

Do not use fprintf, adjust messages, add verbose errno or at least the
errorr code if there's no clear mapping to a string.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: no BUG_ON in close_ctree
David Sterba [Thu, 18 Aug 2016 15:44:18 +0000 (17:44 +0200)]
btrfs-progs: no BUG_ON in close_ctree

There's no reason for it.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: adjust command line options for the low-memory mode
David Sterba [Wed, 17 Aug 2016 17:33:24 +0000 (19:33 +0200)]
btrfs-progs: check: adjust command line options for the low-memory mode

Change the single-purpose option --low-memory to a generic option that
takes the mode. Currently supported are the original mode and the
low-memory in the same way.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce low memory mode
Lu Fengqi [Sun, 24 Apr 2016 07:47:12 +0000 (15:47 +0800)]
btrfs-progs: check: introduce low memory mode

Introduce a new fsck mode: low memory mode.

Old btrfsck is working efficiently but uses some memory for each extent
item.  This method will ensure extents are only iterated once at
extent/chunk tree check process.

But since it uses some memory for each extent item, for a large fs with
several TB metadata, this can easily eat up memory and cause OOM.

To handle such limitation and improve scalability, the new low-memory
mode will not use any heap memory to record which extent is checked.
Instead it will use extent backref to avoid most of uneeded checks on
shared fs/subvolume tree blocks.
And with the use forward and backward reference cross check, we can also
ensure every tree block is at least checked once.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce traversal function for fsck
Lu Fengqi [Sun, 24 Apr 2016 07:35:41 +0000 (15:35 +0800)]
btrfs-progs: check: introduce traversal function for fsck

Introduce a new function traverse_tree_block() to do pre-order
traversal, to co-operate with new fs/subvolume tree skip function.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to speed up fs tree check
Lu Fengqi [Sun, 24 Apr 2016 07:24:56 +0000 (15:24 +0800)]
btrfs-progs: check: introduce function to speed up fs tree check

Introduce function should_check() to reduced duplicated tree block check
for fs/subvolume tree.

The idea is, we only check the fs/subvolue tree block if we have the
lowest referencer rootid, according to extent tree.

In that case, we can skip a lot of fs/subvolume tree block checks if
there are a lot of snapshots.

Although we will do a lot of extent tree searches for it.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce main entry function for checking leaf items
Lu Fengqi [Sun, 24 Apr 2016 07:07:59 +0000 (15:07 +0800)]
btrfs-progs: check: introduce main entry function for checking leaf items

Introduce an entry function, check_leaf_items() to check all
known/valuable items and update related accounting like total_bytes and
csum_bytes.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to check chunk item
Lu Fengqi [Sun, 24 Apr 2016 07:00:24 +0000 (15:00 +0800)]
btrfs-progs: check: introduce function to check chunk item

Introduce function check_chunk_item() to check a chunk item.
It will check all chunk stripes with dev extents and the corresponding
block group item.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to check block group item
Lu Fengqi [Sun, 24 Apr 2016 06:41:41 +0000 (14:41 +0800)]
btrfs-progs: check: introduce function to check block group item

Introduce function check_block_group_item() to check a block group item.
It will check the referencer chunk and the used space accounting with
extent tree.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to check dev used space
Lu Fengqi [Sun, 24 Apr 2016 06:17:10 +0000 (14:17 +0800)]
btrfs-progs: check: introduce function to check dev used space

Introduce function check_dev_item() to check used space with dev extent
items.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to check dev extent item
Lu Fengqi [Sun, 24 Apr 2016 05:54:28 +0000 (13:54 +0800)]
btrfs-progs: check: introduce function to check dev extent item

Introduce function check_dev_extent_item() to find its referencer chunk.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to check an extent
Lu Fengqi [Sun, 24 Apr 2016 05:45:28 +0000 (13:45 +0800)]
btrfs-progs: check: introduce function to check an extent

Introduce function check_extent_item() using previously introduced
functions.

With previous function to check referencer and backref, this function
can be quite easy.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to check shared data backref
Lu Fengqi [Sun, 24 Apr 2016 05:33:47 +0000 (13:33 +0800)]
btrfs-progs: check: introduce function to check shared data backref

Introduce the function check_shared_data_backref() to check the
referencer of a given shared data backref.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to check referencer for data backref
Lu Fengqi [Sat, 23 Apr 2016 08:58:24 +0000 (16:58 +0800)]
btrfs-progs: check: introduce function to check referencer for data backref

Introduce new function check_extent_data_backref() to search referencer
for a given data backref.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to check shared block ref
Lu Fengqi [Sat, 23 Apr 2016 08:35:04 +0000 (16:35 +0800)]
btrfs-progs: check: introduce function to check shared block ref

Introduce function check_shared_block_backref() to check shared block
ref.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to check referencer of a backref
Lu Fengqi [Sat, 23 Apr 2016 08:08:40 +0000 (16:08 +0800)]
btrfs-progs: check: introduce function to check referencer of a backref

Introduce a new function check_tree_block_backref() to check if a
backref points to correct referencer.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to query tree block level
Lu Fengqi [Sat, 23 Apr 2016 07:56:50 +0000 (15:56 +0800)]
btrfs-progs: check: introduce function to query tree block level

Introduce function query_tree_block_level() to resolve tree block level
by the following method:
1) tree block backref level
2) tree block header level

And only when header level == backref level, and transid matches, it will
return the tree block level.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to check data backref in extent tree
Lu Fengqi [Sat, 23 Apr 2016 07:26:25 +0000 (15:26 +0800)]
btrfs-progs: check: introduce function to check data backref in extent tree

Introduce a new function check_data_extent_item() to check if the
corresponding data backref exists in extent tree.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: introduce function to check tree block backref in extent tree
Lu Fengqi [Sat, 23 Apr 2016 06:40:37 +0000 (14:40 +0800)]
btrfs-progs: check: introduce function to check tree block backref in extent tree

Introduce function check_tree_block_ref() to check whether a tree block
has correct backref in extent tree.

Unlike old extent tree check method, we only use search_slot() to search
reference, no extra structure will be allocated in heap to record what we
have checked.

This method may cause a little more IO, but should work for super large
fs without triggering OOM.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: eliminate some unnecessary btrfs_lookup_extent_info() calls in walk_down...
Wang Xiaoguang [Tue, 16 Aug 2016 06:17:57 +0000 (14:17 +0800)]
btrfs-progs: eliminate some unnecessary btrfs_lookup_extent_info() calls in walk_down_tree()

In walk_down_tree(), we may call btrfs_lookup_extent_info() for same tree
block many times, obviously unnecessary. Here we define a simple struct to
record whether we already have gotten tree block's refs:
        struct node_refs {
                u64 bytenr[BTRFS_MAX_LEVEL];
                u64 refs[BTRFS_MAX_LEVEL];
        };

I fill a disk partition with linux kernel source codes and use below
test script to have performance test.
#!/bin/bash

echo 3 > /proc/sys/vm/drop_caches
for ((i = 0; i < 20; i++)); do
time ./btrfsck  /dev/sdc5
done 2>&1 | grep real | awk -F "[ms]" '{run_time += $2} END{print run_time / 20}'

Before this patch, it averagely took 0.8447s for every btrfsck execution,
and with this patch, it averagely took 0.7807s.

Signed-off-by: Wang Xiaoguang <wangxg.fnst@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agoBtrfs progs v4.7
David Sterba [Fri, 29 Jul 2016 13:06:42 +0000 (15:06 +0200)]
Btrfs progs v4.7

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: update CHANGES for 4.7
David Sterba [Thu, 28 Jul 2016 12:18:59 +0000 (14:18 +0200)]
btrfs-progs: update CHANGES for 4.7

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: fix memory leak with missing device
Justin Maggard [Thu, 28 Jul 2016 17:44:11 +0000 (10:44 -0700)]
btrfs-progs: fix memory leak with missing device

In read_one_chunk(), we may add an empty entry for a missing device.
However, this entry wasn't being added to the dev_list, and so it never
got freed.

Signed-off-by: Justin Maggard <jmaggard@netgear.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: fix unaligned access in raid6 calculations
David Sterba [Thu, 28 Jul 2016 17:56:20 +0000 (19:56 +0200)]
btrfs-progs: fix unaligned access in raid6 calculations

The raid6 code matches kernel implementation that also does the
unaligned access. So to keep the code close, add helpers for unaligned
native access and use them. The helpers are local as we don't plan to
use them elsewhere.

Reported-by: Anatoly Pugachev <matorola@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: use proper unaligned helper in btrfs_csum_final
David Sterba [Thu, 28 Jul 2016 17:36:50 +0000 (19:36 +0200)]
btrfs-progs: use proper unaligned helper in btrfs_csum_final

This will not cause unaligned access as the checksum is at the beginning
of btrfs_header and thus aligned to a page, but for clarity use the
helper.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: fi defrag: change default extent target size to 32 MiB
David Sterba [Thu, 28 Jul 2016 11:14:08 +0000 (13:14 +0200)]
btrfs-progs: fi defrag: change default extent target size to 32 MiB

The kernel default is too low, 32 MiB is recommended and should give
better results.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: Doc: Fix format error in btrfs-send
Qu Wenruo [Thu, 28 Jul 2016 08:28:25 +0000 (16:28 +0800)]
btrfs-progs: Doc: Fix format error in btrfs-send

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: handler memory allocation error in write_data_to_disk
David Sterba [Thu, 28 Jul 2016 00:16:43 +0000 (02:16 +0200)]
btrfs-progs: handler memory allocation error in write_data_to_disk

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: handle memory allocation error in __alloc_extent_buffer
David Sterba [Thu, 28 Jul 2016 00:04:30 +0000 (02:04 +0200)]
btrfs-progs: handle memory allocation error in __alloc_extent_buffer

Drop the BUG() as all callers handle errors.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: update values of EXTENT_* bits
David Sterba [Wed, 27 Jul 2016 23:56:06 +0000 (01:56 +0200)]
btrfs-progs: update values of EXTENT_* bits

Make the values unsigned as we do various bit operations.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: refactor and extend btrfs_prepare_device arguments
David Sterba [Wed, 27 Jul 2016 23:47:40 +0000 (01:47 +0200)]
btrfs-progs: refactor and extend btrfs_prepare_device arguments

The message about discard is printed unconditionally and does not
conform to the --quite option eg. in mkfs. Consolidate the operation
flags into one argument and add support for verbosity.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: use values directly for BLOCK_GROUP_ macros
David Sterba [Wed, 27 Jul 2016 21:27:05 +0000 (23:27 +0200)]
btrfs-progs: use values directly for BLOCK_GROUP_ macros

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: use unsigned type for extent_buffer flags
David Sterba [Wed, 27 Jul 2016 21:19:20 +0000 (23:19 +0200)]
btrfs-progs: use unsigned type for extent_buffer flags

We're doing bit operations.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: fix unaligned access calculating raid56 data
David Sterba [Wed, 27 Jul 2016 20:30:52 +0000 (22:30 +0200)]
btrfs-progs: fix unaligned access calculating raid56 data

The extent_buffer::data might be unaligned wrt unsigned long, depends on
acutal layout of the structure and width of the int types. Use explicit
unaligned access helpers.

Reported-by: Anatoly Pugachev <matorola@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: kerncompat: introduce put_unaligned_x helpers
David Sterba [Wed, 27 Jul 2016 21:13:09 +0000 (23:13 +0200)]
btrfs-progs: kerncompat: introduce put_unaligned_x helpers

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: docs: update btrfs-balance manual page
David Sterba [Tue, 26 Jul 2016 17:33:12 +0000 (19:33 +0200)]
btrfs-progs: docs: update btrfs-balance manual page

Update the new options.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: balance: add another (shorter) option for background
David Sterba [Tue, 26 Jul 2016 17:30:39 +0000 (19:30 +0200)]
btrfs-progs: balance: add another (shorter) option for background

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: balance: silence compiler warning
David Sterba [Tue, 26 Jul 2016 17:11:42 +0000 (19:11 +0200)]
btrfs-progs: balance: silence compiler warning

cmds-balance.c: In function 'cmd_balance_start':
cmds-balance.c:654:6: warning: ignoring return value of 'chdir', declared with
attribute warn_unused_result [-Wunused-result]
chdir("/");

Reported-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: add option to run balance as daemon
Austin S. Hemmelgarn [Tue, 21 Jun 2016 15:16:59 +0000 (11:16 -0400)]
btrfs-progs: add option to run balance as daemon

Currently, balance operations are run synchronously in the foreground.
This is nice for interactive management, but is kind of crappy when you
start looking at automation and similar things.

This patch adds an option to `btrfs balance start` to tell it to
daemonize prior to running the balance operation, thus allowing us to
preform balances asynchronously.  The two biggest use cases I have for
this are starting a balance on a remote server without establishing a
full shell session, and being able to background the balance in a
recovery shell (which usually has no job control) so I can still get
progress information.

Because it simply daemonizes prior to calling the balance ioctl, this
doesn't actually need any kernel support.

Signed-off-by: Austin S. Hemmelgarn <ahferroin7@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: fix btrfsck of space_cache=v2 bitmaps on big-endian
Omar Sandoval [Fri, 15 Jul 2016 19:12:48 +0000 (12:12 -0700)]
btrfs-progs: fix btrfsck of space_cache=v2 bitmaps on big-endian

Copy le_test_bit() from the kernel and use that for the free space tree
bitmaps.

Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: corrupt block: handle eb read and write errors
David Sterba [Fri, 15 Jul 2016 15:00:56 +0000 (17:00 +0200)]
btrfs-progs: corrupt block: handle eb read and write errors

Resolves-coverity-id: 1261558
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: corrupt block: handle block mapping errors in debug_corrupt_block
David Sterba [Fri, 15 Jul 2016 14:46:08 +0000 (16:46 +0200)]
btrfs-progs: corrupt block: handle block mapping errors in debug_corrupt_block

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: corrupt block: pass eb as argument to debug_corrupt_block
David Sterba [Fri, 15 Jul 2016 14:38:07 +0000 (16:38 +0200)]
btrfs-progs: corrupt block: pass eb as argument to debug_corrupt_block

Allocate the eb externally so we can handle the easy errors in advance.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: image: fix minor resource leak
David Sterba [Fri, 15 Jul 2016 14:27:06 +0000 (16:27 +0200)]
btrfs-progs: image: fix minor resource leak

It's on error exit path.

Resolves-coverity-id: 1354246
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: handle ulist_add errors in qgroup-verify
David Sterba [Fri, 15 Jul 2016 14:12:00 +0000 (16:12 +0200)]
btrfs-progs: handle ulist_add errors in qgroup-verify

Resolves-coverity-id: 1364084
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agoBtrfs-progs: fix btrfs-map-logical to only print extent mapping info
Liu Bo [Fri, 15 Jul 2016 01:40:27 +0000 (18:40 -0700)]
Btrfs-progs: fix btrfs-map-logical to only print extent mapping info

I have a valid btrfs image which contains,
...
        item 10 key (1103101952 BLOCK_GROUP_ITEM 1288372224) itemoff 15947 itemsize 24
                block group used 655360 chunk_objectid 256 flags DATA|RAID5
        item 11 key (1103364096 EXTENT_ITEM 131072) itemoff 15894 itemsize 53
                extent refs 1 gen 11 flags DATA
                extent data backref root 5 objectid 258 offset 0 count 1
        item 12 key (1103888384 EXTENT_ITEM 262144) itemoff 15841 itemsize 53
                extent refs 1 gen 15 flags DATA
                extent data backref root 1 objectid 256 offset 0 count 1
        item 13 key (1104281600 EXTENT_ITEM 262144) itemoff 15788 itemsize 53
                extent refs 1 gen 15 flags DATA
                extent data backref root 1 objectid 257 offset 0 count 1
...

The extent [1103364096, 131072) has length 131072, but if we run

"btrfs-map-logical -l 1103364096 -b $((65536 * 3)) /dev/sda"

it will return mapping info 's of  non-existing extents.

It's because it assumes that extents's are contiguous on logical address,
when it's not true, after one loop (cur_logical += cur_len) and mapping
the next extent, we can get an extent that is out of our search range and
we end up with a negative @real_len and printing all mapping infos till
the disk end.

Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Reviewed-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: add 007-unsupported-block-sizes
David Sterba [Thu, 14 Jul 2016 23:25:12 +0000 (01:25 +0200)]
btrfs-progs: tests: add 007-unsupported-block-sizes

Check if block sizes smaller than 4k expectedly fail to convert.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: update btrfs manual page
David Sterba [Thu, 14 Jul 2016 21:43:38 +0000 (23:43 +0200)]
btrfs-progs: docs: update btrfs manual page

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: update btrfs-convert manual page
David Sterba [Thu, 14 Jul 2016 21:43:38 +0000 (23:43 +0200)]
btrfs-progs: docs: update btrfs-convert manual page

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: update btrfs-filesystem manual page
David Sterba [Thu, 14 Jul 2016 21:43:38 +0000 (23:43 +0200)]
btrfs-progs: docs: update btrfs-filesystem manual page

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: update btrfs-scrub manual page
David Sterba [Thu, 14 Jul 2016 21:35:26 +0000 (23:35 +0200)]
btrfs-progs: docs: update btrfs-scrub manual page

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: do not set optind if not necessary
David Sterba [Wed, 13 Jul 2016 17:41:55 +0000 (19:41 +0200)]
btrfs-progs: do not set optind if not necessary

In the subcommand callbacks that are called just once, we don't need to
explicitly reset optind.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: balance: use errno directly
David Sterba [Wed, 13 Jul 2016 17:05:23 +0000 (19:05 +0200)]
btrfs-progs: balance: use errno directly

No need to store the errno in a local variable.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: balance: cleanup, switch to common exit block
David Sterba [Wed, 13 Jul 2016 17:00:01 +0000 (19:00 +0200)]
btrfs-progs: balance: cleanup, switch to common exit block

Call close_file_or_dir at the end of the function and replace returns by
gotos to the exit block.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: update btrfs-subvolume manual page
David Sterba [Tue, 12 Jul 2016 15:47:23 +0000 (17:47 +0200)]
btrfs-progs: docs: update btrfs-subvolume manual page

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: update btrfs-balance manual page
David Sterba [Tue, 12 Jul 2016 15:13:16 +0000 (17:13 +0200)]
btrfs-progs: docs: update btrfs-balance manual page

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: du: fix to skip not btrfs dir/file
Wang Shilong [Thu, 7 Jul 2016 09:43:38 +0000 (18:43 +0900)]
btrfs-progs: du: fix to skip not btrfs dir/file

'btrfs file du' is a very useful tool to watch my system
file usage information with snapshot aware.

when trying to run following commands:
[root@localhost btrfs-progs]# btrfs file du /
     Total   Exclusive  Set shared  Filename
ERROR: Failed to lookup root id - Inappropriate ioctl for device
ERROR: cannot check space of '/': Unknown error -1

and My Filesystem looks like this:
[root@localhost btrfs-progs]# df -Th
Filesystem     Type      Size  Used Avail Use% Mounted on
devtmpfs       devtmpfs   16G     0   16G   0% /dev
tmpfs          tmpfs      16G  368K   16G   1% /dev/shm
tmpfs          tmpfs      16G  1.4M   16G   1% /run
tmpfs          tmpfs      16G     0   16G   0% /sys/fs/cgroup
/dev/sda3      btrfs      60G   19G   40G  33% /
tmpfs          tmpfs      16G  332K   16G   1% /tmp
/dev/sdc       btrfs     2.8T  166G  1.7T   9% /data
/dev/sda2      xfs       2.0G  452M  1.6G  23% /boot
/dev/sda1      vfat      1.9G   11M  1.9G   1% /boot/efi
tmpfs          tmpfs     3.2G   24K  3.2G   1% /run/user/1000

So I installed Btrfs as my root partition, but boot partition
can be other fs.

We can Let btrfs tool aware of this is not a btrfs file or
directory and skip those files, so that someone like me
could just run 'btrfs file du /' to scan all btrfs filesystems.

After patch, it will look like:
   Total   Exclusive  Set shared  Filename
     0.00B       0.00B           -  //root/.bash_logout
     0.00B       0.00B           -  //root/.bash_profile
     0.00B       0.00B           -  //root/.bashrc
     0.00B       0.00B           -  //root/.cshrc
     0.00B       0.00B           -  //root/.tcshrc

This works for me to analysis system usage and analysis
performaces.

Signed-off-by: Wang Shilong <wangshilong1991@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: add the error message when open fails
Tsutomu Itoh [Thu, 9 Jun 2016 01:23:15 +0000 (10:23 +0900)]
btrfs-progs: add the error message when open fails

When open in btrfs_open_devices failed, only the following message is
displayed. Therefore the user doesn't understand the reason why open
failed.

  # btrfs check /dev/sdb8
  Couldn't open file system

This patch adds the error message when open fails.

Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: man5, document control device
David Sterba [Fri, 8 Jul 2016 15:04:25 +0000 (17:04 +0200)]
btrfs-progs: docs: man5, document control device

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: 006-image-on-missing-device: fix btrfs tool path
Luis Henriques [Mon, 4 Jul 2016 22:48:47 +0000 (23:48 +0100)]
btrfs-progs: tests: 006-image-on-missing-device: fix btrfs tool path

If btrfs isn't in the path, this test will fail with:

    [TEST/misc]   006-image-on-missing-device
failed: btrfs fi show /dev/loop0
test failed for case 006-image-on-missing-device
Makefile:226: recipe for target 'test-misc' failed
make: *** [test-misc] Error 1

Fix the test script by adding $TOP to the path.

Signed-off-by: Luis Henriques <henrix@camandro.org>
[ updated to full command names ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: check: write corrected qgroup info to disk
Mark Fasheh [Mon, 4 Jul 2016 12:57:38 +0000 (14:57 +0200)]
btrfs-progs: check: write corrected qgroup info to disk

Now that we can verify all qgroups, we can write the corrected qgroups out
to disk when '--repair' is specified. The qgroup status item is also updated
to clear any out-of-date state. The repair_ functions were modeled after the
inode repair code in cmds-check.c.

I also renamed the 'scan' member of qgroup_status_item to 'rescan' in order
to keep consistency with the kernel.

Testing this was easy, I just reproduced qgroup inconsistencies via the
usual routes and had btrfsck fix them.

Signed-off-by: Mark Fasheh <mfasheh@suse.de>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: check: verify qgroups above level 0
Mark Fasheh [Fri, 17 Jun 2016 20:37:48 +0000 (13:37 -0700)]
btrfs-progs: check: verify qgroups above level 0

At the moment we only check subvolume quota groups (level 0). With this
patch we can check groups above 0, thus verifying the entire qgroup
hierarchy on a file system.  The accounting portion of this patch is modeled
after the kernel - we are essentially reimplementing the 'quota rescan' case
here. Most other sections of this code went unchanged, in particular the
root counting works independently of the accounting.

Signed-off-by: Mark Fasheh <mfasheh@suse.de>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: factor out repair mode
David Sterba [Mon, 4 Jul 2016 13:22:53 +0000 (15:22 +0200)]
btrfs-progs: factor out repair mode

Signed-off-by: David Sterba <dsterba@suse.com>