platform/upstream/btrfs-progs.git
7 years agobtrfs-progs: add balance status structures
David Sterba [Tue, 1 Nov 2016 23:53:54 +0000 (00:53 +0100)]
btrfs-progs: add balance status structures

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: teach print_leaf about temporary item subtypes
David Sterba [Tue, 1 Nov 2016 12:39:44 +0000 (13:39 +0100)]
btrfs-progs: teach print_leaf about temporary item subtypes

[ kernel patch 9f07e1d76eaeeddedcf9651395b4a8f870df31f0 ]
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: teach print_leaf about permanent item subtypes
David Sterba [Tue, 1 Nov 2016 12:36:58 +0000 (13:36 +0100)]
btrfs-progs: teach print_leaf about permanent item subtypes

[ kernel patch 585a3d0d238dfe42909cb58b0d05f644365fed14 ]
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: switch dev stats item to the permanent item key
David Sterba [Tue, 1 Nov 2016 12:34:13 +0000 (13:34 +0100)]
btrfs-progs: switch dev stats item to the permanent item key

[ kernel patch 242e2956e4afde7727fffe21adc0a7b198b77f96 ]
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: introduce key type for persistent permanent items
David Sterba [Tue, 1 Nov 2016 12:32:26 +0000 (13:32 +0100)]
btrfs-progs: introduce key type for persistent permanent items

The number of distinct key types is not that big that we could waste one
for something new we want to store in the tree.

Similar to the temporary items, we'll introduce a new name for an
existing key value and use the objectid for further extension.  The
victim is the BTRFS_DEV_STATS_KEY (248).

The device stats are an example of a permanent item.

[ kernel patch 50c2d5abe64c1726b48d292a2ab04f60e8238933 ]
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: print-tree: extract offset from the item key
David Sterba [Tue, 1 Nov 2016 12:40:19 +0000 (13:40 +0100)]
btrfs-progs: print-tree: extract offset from the item key

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: introduce key type for persistent temporary items
David Sterba [Tue, 1 Nov 2016 12:29:25 +0000 (13:29 +0100)]
btrfs-progs: introduce key type for persistent temporary items

The number of distinct key types is not that big that we could waste one
for something new we want to store in the tree. We'll introduce a new
name for an existing key value and use the objectid for further
extension.  The victim is the BTRFS_BALANCE_ITEM_KEY (248).

The nature of the balance status item is a good example of the temporary
item. It exists from beginning of the balance, keeps the status until it
finishes.

[ kernel patch 0bbbccb17fea86818e1a058faf5903aefd20b31a ]
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: mkfs: store creation time of the toplevel subvolume
David Sterba [Mon, 31 Oct 2016 16:06:15 +0000 (17:06 +0100)]
btrfs-progs: mkfs: store creation time of the toplevel subvolume

For unkonwn reasons, the creation time of the toplevel subvolume inode
item hasn't been set.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: dump-tree: print missing data for file extent item
David Sterba [Mon, 31 Oct 2016 15:31:15 +0000 (16:31 +0100)]
btrfs-progs: dump-tree: print missing data for file extent item

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: dump-tree: print more root_item data
David Sterba [Mon, 31 Oct 2016 15:22:17 +0000 (16:22 +0100)]
btrfs-progs: dump-tree: print more root_item data

Print a few more missing items of root_item (there are still some
remaining). Drop key is printend unconditionally.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: dump-tree: print missing dir_item data
David Sterba [Mon, 31 Oct 2016 13:25:03 +0000 (14:25 +0100)]
btrfs-progs: dump-tree: print missing dir_item data

Add items from dir_item that are not printed, plus other adjustments.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: dump-tree: print missing inode_item data
David Sterba [Mon, 31 Oct 2016 13:25:03 +0000 (14:25 +0100)]
btrfs-progs: dump-tree: print missing inode_item data

Add items from inode_item that are not printed.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: dump-tree: add helper to print timespec
David Sterba [Mon, 31 Oct 2016 15:00:53 +0000 (16:00 +0100)]
btrfs-progs: dump-tree: add helper to print timespec

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: dump-tree: factor inode_item dump to function
David Sterba [Mon, 31 Oct 2016 14:20:20 +0000 (15:20 +0100)]
btrfs-progs: dump-tree: factor inode_item dump to function

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: dump-tree: print missing dev_extent data
David Sterba [Mon, 31 Oct 2016 13:25:03 +0000 (14:25 +0100)]
btrfs-progs: dump-tree: print missing dev_extent data

Add items from chunk that are not printed, unify names according to
the member names and reorder a bit.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: dump-tree: print missing chunk data
David Sterba [Mon, 31 Oct 2016 13:25:03 +0000 (14:25 +0100)]
btrfs-progs: dump-tree: print missing chunk data

Add items from chunk that are not printed, unify names according to
the member names and reorder a bit.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: dump-tree: print missing dev_item data
David Sterba [Mon, 31 Oct 2016 13:25:03 +0000 (14:25 +0100)]
btrfs-progs: dump-tree: print missing dev_item data

Add items from dev_item that are not printed, unify names according to
the member names.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: subvol show: print more details about toplevel subvolume
David Sterba [Mon, 31 Oct 2016 12:56:28 +0000 (13:56 +0100)]
btrfs-progs: subvol show: print more details about toplevel subvolume

The toplevel subvolume is special and the other listing code leaves it
out so we have to add several special cases to handle it. There's no
backreference so the path is built artificially. New helper
btrfs_get_toplevel_subvol is a reduced version of btrfs_get_subvol.

There's some information usually missing for the toplevel subvolume, eg.
the uuid or creation info. This has to be fixed on the mkfs side, the
other subvolumes are created by kernel.

Example:
/mnt
        Name:                   <FS_TREE>
        UUID:                   -
        Parent UUID:            -
        Received UUID:          -
        Creation time:          -
        Subvolume ID:           5
        Generation:             233
        Gen at creation:        0
        Parent ID:              0
        Top level ID:           0
        Flags:                  -
        Snapshot(s):
                                subv1

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: subvol list: better error message if subvol insertion fails
David Sterba [Mon, 31 Oct 2016 11:53:35 +0000 (12:53 +0100)]
btrfs-progs: subvol list: better error message if subvol insertion fails

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: subvol list: remove ugly goto construct
David Sterba [Mon, 31 Oct 2016 11:46:52 +0000 (12:46 +0100)]
btrfs-progs: subvol list: remove ugly goto construct

Jumping into an if is not considered nice.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: remove trivial helper root_lookup_init
David Sterba [Mon, 31 Oct 2016 11:44:29 +0000 (12:44 +0100)]
btrfs-progs: remove trivial helper root_lookup_init

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: subvol list: consolidate uuid types accross functions
David Sterba [Mon, 31 Oct 2016 11:33:28 +0000 (12:33 +0100)]
btrfs-progs: subvol list: consolidate uuid types accross functions

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: subvol list: add simplified helper for adding root backrefs
David Sterba [Mon, 31 Oct 2016 11:27:08 +0000 (12:27 +0100)]
btrfs-progs: subvol list: add simplified helper for adding root backrefs

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: subvol list: consilidate naming of otime varaibles
David Sterba [Mon, 31 Oct 2016 11:19:50 +0000 (12:19 +0100)]
btrfs-progs: subvol list: consilidate naming of otime varaibles

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: subvol list: simplify value assignments
David Sterba [Mon, 31 Oct 2016 11:15:18 +0000 (12:15 +0100)]
btrfs-progs: subvol list: simplify value assignments

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: subvol list: remove useless comments
David Sterba [Mon, 31 Oct 2016 11:04:55 +0000 (12:04 +0100)]
btrfs-progs: subvol list: remove useless comments

The search ioctl follows a well known usage pattern.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: subvol list: cleanup layout argument setup
David Sterba [Mon, 31 Oct 2016 11:00:15 +0000 (12:00 +0100)]
btrfs-progs: subvol list: cleanup layout argument setup

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: return void from btrfs_list_setup_filter
David Sterba [Mon, 31 Oct 2016 10:51:54 +0000 (11:51 +0100)]
btrfs-progs: return void from btrfs_list_setup_filter

As an early setup function, failure uisng exit is ok.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: subvol list: setup list filters later
David Sterba [Mon, 31 Oct 2016 10:40:23 +0000 (11:40 +0100)]
btrfs-progs: subvol list: setup list filters later

First check for arguments and whether we can open the filesystem.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: list: switch to common message helpers
David Sterba [Mon, 31 Oct 2016 10:32:12 +0000 (11:32 +0100)]
btrfs-progs: list: switch to common message helpers

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: list: rename some helpers
David Sterba [Mon, 31 Oct 2016 10:17:02 +0000 (11:17 +0100)]
btrfs-progs: list: rename some helpers

Using _volume_ is confusing, it's for subvolumes.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: use enum for list layout type
David Sterba [Mon, 31 Oct 2016 10:11:50 +0000 (11:11 +0100)]
btrfs-progs: use enum for list layout type

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: list: constify prefix arugment
David Sterba [Mon, 31 Oct 2016 10:08:06 +0000 (11:08 +0100)]
btrfs-progs: list: constify prefix arugment

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: use symbolic tree name when searching
David Sterba [Mon, 31 Oct 2016 09:59:08 +0000 (10:59 +0100)]
btrfs-progs: use symbolic tree name when searching

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: remove leading underscores from several helper
David Sterba [Mon, 31 Oct 2016 09:52:26 +0000 (10:52 +0100)]
btrfs-progs: remove leading underscores from several helper

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: opencode trivial helper __free_all_subvolumn
David Sterba [Mon, 31 Oct 2016 09:51:52 +0000 (10:51 +0100)]
btrfs-progs: opencode trivial helper __free_all_subvolumn

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: use existing rootid resolving helper in btrfs_list_get_path_rootid
David Sterba [Mon, 31 Oct 2016 09:38:31 +0000 (10:38 +0100)]
btrfs-progs: use existing rootid resolving helper in btrfs_list_get_path_rootid

The utils helper is not verbose in case of an error, for now the helper
used for subvolume listing will print the error message but not
duplicate the ioctl anymore.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: rename lookup_ino_rootid
David Sterba [Mon, 31 Oct 2016 09:34:37 +0000 (10:34 +0100)]
btrfs-progs: rename lookup_ino_rootid

It does not resolve the inode number but path where fd has been opened.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: remove unused function btrfs_get_path_rootid
David Sterba [Mon, 31 Oct 2016 09:29:12 +0000 (10:29 +0100)]
btrfs-progs: remove unused function btrfs_get_path_rootid

Last user replaced with a more appropriate helper.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: qgroups show: handle errors when resolving root id
David Sterba [Mon, 31 Oct 2016 09:27:25 +0000 (10:27 +0100)]
btrfs-progs: qgroups show: handle errors when resolving root id

If btrfs_get_path_rootid fails, prints an error message but continues.
Replace with a helper that returns an error.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: remove send-test tool
Qu Wenruo [Tue, 1 Nov 2016 08:01:46 +0000 (16:01 +0800)]
btrfs-progs: remove send-test tool

Since new "receive --dump" has better output and structure, it's time
to remove old and function-weak send-test tool.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: Add travis CI build support
Roman Lebedev [Mon, 31 Oct 2016 09:08:46 +0000 (10:08 +0100)]
btrfs-progs: Add travis CI build support

Automated build checks for the devel branch and integration to coverity,
using travis.org. Roman's initial .yml version has been adjusted for
btrfs-progs needs.

Author: Roman Lebedev <lebedev.ri@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: convert: switch more messages to common helpers
David Sterba [Fri, 28 Oct 2016 16:34:03 +0000 (18:34 +0200)]
btrfs-progs: convert: switch more messages to common helpers

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: send: fix handling of multiple snapshots (-p option)
Tsutomu Itoh [Wed, 19 Oct 2016 02:35:03 +0000 (11:35 +0900)]
btrfs-progs: send: fix handling of multiple snapshots (-p option)

We cannot send multiple snapshots at once by -p option.

[before]
# btrfs send -f /tmp/data0 -p Snap0 Snap[12]
At subvol Snap1
At subvol Snap2
ERROR: parent determination failed for 0
#

[after]
# btrfs send -f /tmp/data0 -p Snap0 Snap[12]
At subvol Snap1
At subvol Snap2
#

Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: Ignore clang complete file
Qu Wenruo [Thu, 27 Oct 2016 02:32:48 +0000 (10:32 +0800)]
btrfs-progs: Ignore clang complete file

While most guys are using ctags and cscope with vim, new completion tool
like vim-clang_completion is gaining its popularity, due to its compiler
level accuracy simpleness to use.

Since ctags and cscope are already in .gitignore, I see no reason to
reject .clang_complete.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: tests: switch to dump- commands from inspect
David Sterba [Wed, 26 Oct 2016 14:41:32 +0000 (16:41 +0200)]
btrfs-progs: tests: switch to dump- commands from inspect

The dump-super and debug-tree commands are replacements for the
standalone tools.

Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: fsck-tests: Check if clear space cache works
Qu Wenruo [Thu, 13 Oct 2016 09:22:27 +0000 (17:22 +0800)]
btrfs-progs: fsck-tests: Check if clear space cache works

Add test case to check the basic function of --clear-space-cache.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
7 years agobtrfs-progs: check: add support to clear v1 free space cache
Qu Wenruo [Thu, 13 Oct 2016 09:22:26 +0000 (17:22 +0800)]
btrfs-progs: check: add support to clear v1 free space cache

Kernel clear_cache mount option will only rebuild free space cache if
the used space of that chunk has changed.

So it won't ensure any corrupted free space cache get cleared.

So add a new option "--clear-space-cache v1|v2" to btrfsck, to
completely wipe out free space cache.
So kernel won't complain again.

Reported-by: Ivan P <chrnosphered@gmail.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[ adjusted error messages, doc wording changes ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agoBtrfs progs v4.8.2
David Sterba [Wed, 26 Oct 2016 17:43:34 +0000 (19:43 +0200)]
Btrfs progs v4.8.2

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: update CHANGES for v4.8.2
David Sterba [Mon, 24 Oct 2016 13:40:37 +0000 (15:40 +0200)]
btrfs-progs: update CHANGES for v4.8.2

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: check: release path after usage
Goldwyn Rodrigues [Mon, 24 Oct 2016 15:18:14 +0000 (10:18 -0500)]
btrfs-progs: check: release path after usage

While performing an fsck, an assertion failure occurs because of reusing path in a loop.
ctree.c:1112: btrfs_search_slot: Warning: assertion `p->nodes[0] != NULL` failed, value 0

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Fix wrong tree block alignment for unalianged block group
Qu Wenruo [Mon, 24 Oct 2016 07:22:33 +0000 (15:22 +0800)]
btrfs-progs: Fix wrong tree block alignment for unalianged block group

Commit 854437ca(btrfs-progs: extent-tree: avoid allocating tree block
that crosses stripe boundary) introduces check for logical bytenr not
crossing stripe boundary.

However that check is not completely correct.
It only checks if the logical bytenr and length agaist absolute logical
offset.
That's to say, it only check if a tree block lies in 64K logical stripe.

But in fact, it's possible a block group starts at bytenr unaligned with
64K, just like the following case.

Then btrfsck will give false alert.

0       32K       64K       96K        128K         160K ...
        |--------------- Block group A ---------------------
|<-----TB 32K------>|
        |/Scrub stripe unit/|
|    WRONG UNIT   |

In that case, TB(tree block) at bytenr 32K in fact fits into the kernel
scrub stripe unit.
But doesn't fit into the pure logical 64K stripe.

Fix check_crossing_stripes() to compare bytenr to block group start, not
to absolute logical bytenr.

Reported-by: Jussi Kansanen <jussi.kansanen@gmail.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: ioctl: fix build failure if BUILD_ASSERT is not defined
David Sterba [Mon, 24 Oct 2016 12:46:26 +0000 (14:46 +0200)]
btrfs-progs: ioctl: fix build failure if BUILD_ASSERT is not defined

The build fails if ioctl.h is included without kerncompat.h which
defines BUILD_ASSERT, as reported. The proposed patches add kerncompat.h
but I don't want to pull that header via ioctl.h, so the fix will
provide an empty macro instead.

Reported-by: Eric Sandeen <sandeen@redhat.com>
Reported-by: Sergei Trofimovich <slyfox@gentoo.org>
Reported-by: Mikhail Pukhlikov <cynede@gentoo.org>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: move btrfs_extref_hash() to hash.h
Lu Fengqi [Fri, 15 Jul 2016 03:00:28 +0000 (11:00 +0800)]
btrfs-progs: move btrfs_extref_hash() to hash.h

Move btrfs_extref_hash() from inode-item.c to hash.h,
so that the function can be called elsewhere.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
8 years agobtrfs-progs: send: remove unnecessary code
Tsutomu Itoh [Wed, 19 Oct 2016 03:45:57 +0000 (12:45 +0900)]
btrfs-progs: send: remove unnecessary code

Some unnecessary codes are deleted.

 - the setting of subvol is duplicated
 - read only check was already done by the previous loop

Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: dump-tree: fix crash on unrecognized tree id
David Sterba [Wed, 19 Oct 2016 10:38:10 +0000 (12:38 +0200)]
btrfs-progs: dump-tree: fix crash on unrecognized tree id

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: rename btrfs_scan_lblkid() to btrfs_scan_devices()
Karel Zak [Tue, 11 Nov 2014 11:35:58 +0000 (12:35 +0100)]
btrfs-progs: rename btrfs_scan_lblkid() to btrfs_scan_devices()

It seems like bad idea to use a library name (lblkid) within generic
function name. The currently used scanning library is implementation
detail and this detail should be hidden for rest of the code.

Signed-off-by: Karel Zak <kzak@redhat.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
8 years agobtrfs-progs: build: detect fiemap shared flag but don't fail build
David Sterba [Fri, 14 Oct 2016 16:39:40 +0000 (18:39 +0200)]
btrfs-progs: build: detect fiemap shared flag but don't fail build

The FIEMAP_EXTENT_SHARED fiemap flag was introduced in 2.6.33. If the
headers do not provide the definition, the build will fail. The support
of the fiemap sharing depends on the running kernel. There are still
systems with 2.6.32 kernel headers but running newer versions.

To support such environment, don't fail build, provide own defintion of
the structure and detect if there's an old kernel in use in the relevant
command (btrfs fi du).

Reported-by: Abhay Sachan <lkp.abhay@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: mkfs: detect version of running kernel
David Sterba [Tue, 9 Feb 2016 23:23:29 +0000 (00:23 +0100)]
btrfs-progs: mkfs: detect version of running kernel

Use the uname syscall and parse the string.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: volumes: Remove BUG_ON in raid56 write routine
Qu Wenruo [Tue, 25 Oct 2016 02:11:04 +0000 (10:11 +0800)]
btrfs-progs: volumes: Remove BUG_ON in raid56 write routine

Remove various BUG_ON in raid56 write routine, including:
1) Memory allocation error
   Old codes allocates memory when code needs new memory in a loop, and
   catch the error using BUG_ON().
   New codes allocates memory in a allocation loop first, if any failure
   is caught, freeing already allocated memories then throw -ENOMEM.

2) Write error
   Change BUG_ON() to correct return value.

3) Validation check
   Same as write error.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: rename raid6.c to raid56.c
Qu Wenruo [Fri, 30 Sep 2016 05:04:54 +0000 (13:04 +0800)]
btrfs-progs: rename raid6.c to raid56.c

This allows us to put raid5 codes into that file other than creating a
new raid5.c.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: test: Add image for quota verify stack overflow
Qu Wenruo [Fri, 7 Oct 2016 07:23:00 +0000 (15:23 +0800)]
btrfs-progs: test: Add image for quota verify stack overflow

For image with tree reloc tree, if its height is over level 2, the root
node's backref will point to itself.  It's valid for kernel, but quota
verify code can't handle it and cause a infinite call, overflowing the
stack.

Add minimal image to reproduce the bug, as regression test.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[ minor cleanups in test.sh ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: test: Add test image for btrfsck qgroup rescan detection
Qu Wenruo [Fri, 7 Oct 2016 07:22:59 +0000 (15:22 +0800)]
btrfs-progs: test: Add test image for btrfsck qgroup rescan detection

Fixed by commit 7c646c538e74 btrfs-progs: qgroup: Fix regression leads
to corrupted qgroup status.

Add minimal test image for that fix.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[ minor cleanups in test.sh ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Fix stack overflow for checking qgroup on tree reloc tree
Qu Wenruo [Thu, 6 Oct 2016 09:13:32 +0000 (17:13 +0800)]
btrfs-progs: Fix stack overflow for checking qgroup on tree reloc tree

For tree reloc tree whose level is >= 2, the root node's parent will
point to itself.
In this case it will make btrfsck overflow its stack and cause segfault.

While for tree reloc tree, it doesn't affect qgroup and kernel can
handle it well.

So add tree reloc tree check for qgroup-verify.c and fix the bug.

Test case will follow soon after I make a minimal image for it.
Current xz ziped image is still over 10M for a 512M fs.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: image: fix compiler warning
Tsutomu Itoh [Wed, 5 Oct 2016 08:07:48 +0000 (17:07 +0900)]
btrfs-progs: image: fix compiler warning

In v4.8-rc1, gcc 5.3.1 gives following warning. Fixed it.

    [CC]     btrfs-image.o
btrfs-image.c: In function 'flush_pending':
btrfs-image.c:708:17: warning: 'start' may be used uninitialized in this function [-Wmaybe-uninitialized]
  header->bytenr = cpu_to_le64(start);
                 ^
btrfs-image.c:927:6: note: 'start' was declared here
  u64 start;
      ^

Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert-test: Add test case for common inode flags
Qu Wenruo [Mon, 10 Oct 2016 03:00:13 +0000 (11:00 +0800)]
btrfs-progs: convert-test: Add test case for common inode flags

Add a new test case to check if btrfs-convert copies common inode flags
like append(a), immutable(i).

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[ minor fixes ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Convert ext inode flags to btrfs inode flags
Qu Wenruo [Mon, 10 Oct 2016 02:30:34 +0000 (10:30 +0800)]
btrfs-progs: convert: Convert ext inode flags to btrfs inode flags

Before this patch, btrfs-convert never copy ext* inode flags to
corresponding btrfs inode flags.

This makes common flags like APPEND/SYNC/SYNCDIR/IMMUTABLE not copied to
btrfs inode.

This patch introduces ext2_convert_inode_flags() function to handle the
convert, so btrfs-convert can copy as many inode flags as possible.

Reported-by: Lakshmipathi.G <lakshmipathi.g@gmail.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Make btrfs-debug-tree print all readable strings for inode flags
Qu Wenruo [Mon, 10 Oct 2016 02:12:22 +0000 (10:12 +0800)]
btrfs-progs: Make btrfs-debug-tree print all readable strings for inode flags

Before this patch, only 3 inode flags have readable string: NODATACOW,
NODATASUM, READONLY.

This patch will output all readable strings for remaining inode flags,
making debug-tree tool more handy.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[ minor adjustments ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Copy btrfs inode flags from kernel header
Qu Wenruo [Mon, 10 Oct 2016 01:47:39 +0000 (09:47 +0800)]
btrfs-progs: Copy btrfs inode flags from kernel header

Btrfs-progs header lacks quite a lot inode flags.
Copy them from kernel for later use.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Add fast,slow symlinks, fifo types to convert test
Lakshmipathi.G [Sun, 9 Oct 2016 15:08:12 +0000 (17:08 +0200)]
btrfs-progs: Add fast,slow symlinks, fifo types to convert test

Signed-off-by: Lakshmipathi.G <Lakshmipathi.G@giis.co.in>
[ fix root helper use in fast_symlink ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: btrfs-debugfs: cleanup unused variables reported by pylint
Lakshmipathi.G [Mon, 10 Oct 2016 12:04:55 +0000 (14:04 +0200)]
btrfs-progs: btrfs-debugfs: cleanup unused variables reported by pylint

Signed-off-by: Lakshmipathi.G <Lakshmipathi.G@giis.co.in>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: test: add default ioctl-test build
David Sterba [Fri, 7 Oct 2016 09:55:50 +0000 (11:55 +0200)]
btrfs-progs: test: add default ioctl-test build

Add another target that'll use default compiler options, eg. if we don't
know the bitness of the architecture or gcc does not recognize the
-m32/-m64 options.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: make the ioctl-test actually useful
David Sterba [Thu, 6 Oct 2016 11:40:24 +0000 (13:40 +0200)]
btrfs-progs: tests: make the ioctl-test actually useful

Enhance the test to verify ioctl uniqueness, compare the defined values
against the hardcoded expected values, and take care of the 32bit/64bit
compatibility workarounds.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: ioctl: add 64bit compat for SEND
David Sterba [Thu, 6 Oct 2016 11:11:17 +0000 (13:11 +0200)]
btrfs-progs: ioctl: add 64bit compat for SEND

The ioctl value of SEND will be different on 32bit userspace and 64bit
kernel due to different pointer type width, that unfortunatelly made it
into the structure definition.

To maintain backward compatibility, we must do it in the 64bit->32bit
way, because we don't have the kernel side workardound like
SET_RECEIVED_SUBVOL has.  Changing value of SEND would then break
existing users of the raw ioctl.

The compatibility structure and ioctl should not be used, exists for
documentation, and testing.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: ioctl: add 32bit compat for SET_RECEIVED_SUBVOL
David Sterba [Thu, 6 Oct 2016 11:06:18 +0000 (13:06 +0200)]
btrfs-progs: ioctl: add 32bit compat for SET_RECEIVED_SUBVOL

The ioctl value of SET_RECEIVED_SUBVOL will be different on 32bit
userspace and 64bit kernel. This is transparently handled on the kernel
side. We now define the same structure so we can verify the ioctl value.

The defined structure and ioctl should not be used. It is exists only
for testing purposes, documenting the mess and as a warning for the
future.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agoBtrfs progs v4.8.1
David Sterba [Wed, 12 Oct 2016 13:20:29 +0000 (15:20 +0200)]
Btrfs progs v4.8.1

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: update CHANGES for v4.8.1
David Sterba [Fri, 7 Oct 2016 12:23:35 +0000 (14:23 +0200)]
btrfs-progs: update CHANGES for v4.8.1

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: ioctl: make build-time structure size checks optional
David Sterba [Mon, 10 Oct 2016 07:59:18 +0000 (09:59 +0200)]
btrfs-progs: ioctl: make build-time structure size checks optional

Temporarily make the build checks optional. The structure sizes could
change on arches due to alignment requirements or padding inserted into
the structures. We need more extensive tests to make sure we'd not break
ioctls.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: fix printf format of sizeof on 32bit build
David Sterba [Thu, 6 Oct 2016 16:04:51 +0000 (18:04 +0200)]
btrfs-progs: fix printf format of sizeof on 32bit build

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: add script to help build coverage
David Sterba [Thu, 6 Oct 2016 15:35:33 +0000 (17:35 +0200)]
btrfs-progs: tests: add script to help build coverage

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Fix warning_trace compile error if backtrace is disabled
Qu Wenruo [Thu, 6 Oct 2016 08:47:19 +0000 (16:47 +0800)]
btrfs-progs: Fix warning_trace compile error if backtrace is disabled

If we disable backtrace, btrfs-progs can't be compiled since we don't
have warning_trace defined.

Fix by move it out of #ifndef BTRFS_DISABLE_BACKTRACE block.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[ move warning_trace to the right place ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: build: add explicit option separator to rm commands
David Sterba [Thu, 6 Oct 2016 10:56:02 +0000 (12:56 +0200)]
btrfs-progs: build: add explicit option separator to rm commands

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agoBtrfs progs v4.8
David Sterba [Wed, 5 Oct 2016 10:47:51 +0000 (12:47 +0200)]
Btrfs progs v4.8

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: update CHANGES for v4.8
David Sterba [Thu, 28 Jul 2016 12:18:59 +0000 (14:18 +0200)]
btrfs-progs: update CHANGES for v4.8

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: add build-time checks for size of ioctl structures
David Sterba [Tue, 4 Oct 2016 22:01:31 +0000 (00:01 +0200)]
btrfs-progs: add build-time checks for size of ioctl structures

All structures that become part of ioctl definition must never change
size. Add the build time checks for that.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: mkfs: fix warning for printf format on 32bit
David Sterba [Tue, 4 Oct 2016 21:00:20 +0000 (23:00 +0200)]
btrfs-progs: mkfs: fix warning for printf format on 32bit

Compiler complains about nlink_t and %ld format on 32bit build. Add
typecast and fix the format.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: kerncompat: add build-time assertion support
David Sterba [Tue, 4 Oct 2016 20:14:12 +0000 (22:14 +0200)]
btrfs-progs: kerncompat: add build-time assertion support

Macro to verify compile-time conditions, like sie of structures or
types.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: build: add alias for check tests
David Sterba [Tue, 4 Oct 2016 17:09:23 +0000 (19:09 +0200)]
btrfs-progs: build: add alias for check tests

$ make test-check"

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: kerncompat: call abort at the end of assert_trace
David Sterba [Tue, 4 Oct 2016 17:02:37 +0000 (19:02 +0200)]
btrfs-progs: kerncompat: call abort at the end of assert_trace

The assert* helpers should not exit normally, mimick the behaviour of
the std library assert.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: kerncompat: make WARN_ON more verbose
David Sterba [Tue, 4 Oct 2016 16:55:09 +0000 (18:55 +0200)]
btrfs-progs: kerncompat: make WARN_ON more verbose

Curretnly WARN_ON would crash but that's not it's purpose. Add helper
that prints the warning, optionally with trace.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: teach run_mayfail about sigabrt
David Sterba [Tue, 4 Oct 2016 16:53:38 +0000 (18:53 +0200)]
btrfs-progs: tests: teach run_mayfail about sigabrt

Mayfail helper should stop when we encoutner an abort.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: dump-super: print chunks after validation
David Sterba [Tue, 4 Oct 2016 16:12:30 +0000 (18:12 +0200)]
btrfs-progs: dump-super: print chunks after validation

The fuzz test 005-simple-dump-super hits a crash, because we print the
chunks too early.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: add new callback to check state
David Sterba [Tue, 4 Oct 2016 14:24:00 +0000 (16:24 +0200)]
btrfs-progs: convert: add new callback to check state

Make the hardcoded ext2 call to a generic per-fs callback.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: rename __setup_root and change to return void
David Sterba [Tue, 4 Oct 2016 13:34:16 +0000 (15:34 +0200)]
btrfs-progs: rename __setup_root and change to return void

Exported function should use a prefix.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: don't treat segfault as ignorable error
David Sterba [Mon, 3 Oct 2016 15:55:49 +0000 (17:55 +0200)]
btrfs-progs: tests: don't treat segfault as ignorable error

Some fuzzed images cause various tools to crash but the mayfail helper
would not recognize that. We don't mind if the utility failes but it
must not crash.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: split test 004 to separate tests
David Sterba [Mon, 3 Oct 2016 15:54:05 +0000 (17:54 +0200)]
btrfs-progs: tests: split test 004 to separate tests

Makes testing specific tool easier.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: don't access freed memory in btrfs_close_devices
David Sterba [Mon, 3 Oct 2016 14:48:53 +0000 (16:48 +0200)]
btrfs-progs: don't access freed memory in btrfs_close_devices

Caught by ASAN, as we don't access fs_devices directly, there's no null
deref but 0x30 is still wrong.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: chunk-recover: handle duplicate cache entries
David Sterba [Mon, 3 Oct 2016 13:49:45 +0000 (15:49 +0200)]
btrfs-progs: chunk-recover: handle duplicate cache entries

Triggered by fuzzed image bko-155621-bad-block-group-offset.raw .

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: print value when assertion fails
David Sterba [Mon, 3 Oct 2016 13:41:25 +0000 (15:41 +0200)]
btrfs-progs: print value when assertion fails

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: image: drop useless bug_on
David Sterba [Mon, 3 Oct 2016 13:13:50 +0000 (15:13 +0200)]
btrfs-progs: image: drop useless bug_on

The callers touch 'target', no point checking it inside restore_metadump
again.

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