platform/upstream/btrfs-progs.git
8 years agobtrfs-progs: build: check if FIEMAP_EXTENT_SHARED is defined
David Sterba [Fri, 24 Jun 2016 12:04:04 +0000 (14:04 +0200)]
btrfs-progs: build: check if FIEMAP_EXTENT_SHARED is defined

Detect the macro at configure time rather than during compilation.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: build: add m4 macros for AC_CHECK_DEFINE
David Sterba [Fri, 24 Jun 2016 12:03:36 +0000 (14:03 +0200)]
btrfs-progs: build: add m4 macros for AC_CHECK_DEFINE

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: build: let autotools include m4
David Sterba [Fri, 24 Jun 2016 11:54:45 +0000 (13:54 +0200)]
btrfs-progs: build: let autotools include m4

We'll need some non-standard macro definitions.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: fi show: print error message if no valid Btrfs is specified
Satoru Takeuchi [Fri, 24 Jun 2016 08:24:47 +0000 (17:24 +0900)]
btrfs-progs: fi show: print error message if no valid Btrfs is specified

* Before this patch

 ===============================
 # ./btrfs fi show foo      # "foo" doesn't mean any valid Btrfs
 #                          # no error message
 # echo $?
 1
 ===============================

* After this patch

 ===============================
 # ./btrfs fi show foo
 ERROR: foo is not a valid Btrfs
 #
 # echo $?
 1
 ===============================

Signed-off-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: fix filesytem type creation for convert tests
David Sterba [Wed, 22 Jun 2016 13:52:17 +0000 (15:52 +0200)]
btrfs-progs: tests: fix filesytem type creation for convert tests

The extN filesystem type was lost when the separate tests were created
and we've been testing only ext2. The tests pass for ext3 and ext4
though.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: add 005-delete-all-rollback
David Sterba [Wed, 22 Jun 2016 13:50:17 +0000 (15:50 +0200)]
btrfs-progs: tests: add 005-delete-all-rollback

Test if a rollback works after deleing all files from btrfs.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: update README
David Sterba [Wed, 22 Jun 2016 12:28:23 +0000 (14:28 +0200)]
btrfs-progs: tests: update README

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: add test console
David Sterba [Wed, 22 Jun 2016 12:07:46 +0000 (14:07 +0200)]
btrfs-progs: tests: add test console

Add a wrapper that sets up environment the same way a test would use it.
Use it for quick prototyping or testing, the commands and output is
logged.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: print shorter test name in the output
David Sterba [Wed, 22 Jun 2016 09:30:24 +0000 (11:30 +0200)]
btrfs-progs: tests: print shorter test name in the output

The full path is printed, we're interested in the last path component
only.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: split convert_test
David Sterba [Wed, 22 Jun 2016 09:25:33 +0000 (11:25 +0200)]
btrfs-progs: tests: split convert_test

Split the big function to several helpers so we can use them separately.
Add comments and do minor tweaks.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: update 'btrfs-device' manual page
David Sterba [Tue, 21 Jun 2016 00:48:39 +0000 (02:48 +0200)]
btrfs-progs: docs: update 'btrfs-device' manual page

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: free qgroup counts in btrfsck
Mark Fasheh [Wed, 15 Jun 2016 22:50:01 +0000 (15:50 -0700)]
btrfs-progs: free qgroup counts in btrfsck

Signed-off-by: Mark Fasheh <mfasheh@suse.de>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agoBtrfs-progs: Initialize stripesize to the value of sectorsize
Chandan Rajendra [Fri, 17 Jun 2016 05:37:54 +0000 (11:07 +0530)]
Btrfs-progs: Initialize stripesize to the value of sectorsize

stripesize should ideally be set to the value of sectorsize. However
previous versions of btrfs-progs/mkfs.btrfs had set stripesize to a
value of 4096. On machines with PAGE_SIZE other than 4096, This could
lead to the following scenario,

- /dev/loop0, /dev/loop1 and /dev/loop2 are mounted as a single
  filesystem. The filesystem was created by an older version of mkfs.btrfs
  which set stripesize to 4k.
- losetup -a
   /dev/loop0: [0030]:19477 (/root/disk-imgs/file-0.img)
   /dev/loop1: [0030]:16577 (/root/disk-imgs/file-1.img)
   /dev/loop2: [64770]:3423229 (/root/disk-imgs/file-2.img)
- /etc/mtab lists only /dev/loop0
- losetup /dev/loop4 /root/disk-imgs/file-1.img
  The new mkfs.btrfs invoked as 'mkfs.btrfs -f /dev/loop4' succeeds even
  though /dev/loop1 has already been mounted and has
  /root/disk-imgs/file-1.img as its backing file.

The above behaviour occurs because check_super() function returns an
error code (due to stripesize not being set to 4096) and hence
check_mounted_where() function treats /dev/loop1 as a disk containing a
filesystem other than Btrfs.

Hence as a workaround this commit allows 4096 as a valid stripesize.

Signed-off-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: qgroup: Fix a bug that fails to skip rescan running case
Qu Wenruo [Thu, 16 Jun 2016 01:15:51 +0000 (09:15 +0800)]
btrfs-progs: qgroup: Fix a bug that fails to skip rescan running case

Commit 6bdf962fe35a8648d(btrfs-progs: Read qgroup status for qgroup
verify) will read qgroup status, and then use it to skip qgroup
reporting.

However since the rescan_running/inconsistent flags are only 1 bit long,
while qgroup flags & BTRFS_QGROUP_FLAGS returns value longer than 1bit,
it doesn't work.

Fix it by doing double negation on (flags & BTRFS_QGROUP_FLAGS) to
convert it to either 1 or 0.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: update btrfs-restore manual page
David Sterba [Thu, 16 Jun 2016 16:48:44 +0000 (18:48 +0200)]
btrfs-progs: docs: update btrfs-restore manual page

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: add 003-fi-resize-args
David Sterba [Thu, 16 Jun 2016 11:50:18 +0000 (13:50 +0200)]
btrfs-progs: tests: add 003-fi-resize-args

Check various resize option combinations.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: fi resize: fix parsing of negative argument
David Sterba [Thu, 16 Jun 2016 11:27:26 +0000 (13:27 +0200)]
btrfs-progs: fi resize: fix parsing of negative argument

The cleanup of option parsing caused a regression where the negative
resize argument is recognized as short options and the command fails.
Use the new helper to allow that.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=119971
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: new helper for option parsing, more permissive for "no options"
David Sterba [Thu, 16 Jun 2016 11:27:18 +0000 (13:27 +0200)]
btrfs-progs: new helper for option parsing, more permissive for "no options"

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: doc: correct the destination of btrfs-receive
Satoru Takeuchi [Tue, 14 Jun 2016 05:50:19 +0000 (14:50 +0900)]
btrfs-progs: doc: correct the destination of btrfs-receive

We can set not only btrfs mount point but also any path belong to
btrfs mount point as btrfs-receive's destination.

Signed-off-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: doc: add missing newline in btrfs-convert
Noah Massey [Fri, 10 Jun 2016 13:57:55 +0000 (09:57 -0400)]
btrfs-progs: doc: add missing newline in btrfs-convert

Signed-off-by: Noah Massey <noah.massey@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agoBtrfs progs v4.6
David Sterba [Fri, 10 Jun 2016 13:03:18 +0000 (15:03 +0200)]
Btrfs progs v4.6

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: update CHANGES for 4.6
David Sterba [Wed, 8 Jun 2016 09:40:02 +0000 (11:40 +0200)]
btrfs-progs: update CHANGES for 4.6

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Fix a bug which fails to insert hole file extent
Qu Wenruo [Thu, 2 Jun 2016 07:22:49 +0000 (15:22 +0800)]
btrfs-progs: convert: Fix a bug which fails to insert hole file extent

When copying inode, if there is a file referring part of a hole range,
convert will fail.

The problem is, when calculating real extent bytenr, it doesn't check if
the original extent is a hole.

In case the orinal extent is a hole, we still calculate bytenr using
file_pos - found_extent_file_pos, causing non-zero value, and later
btrfs_record_file_extent() detects that we are pointing to non-exist
extent and aborts convert.

Fix it by checking the disk_bytenr before calculating real disk bytenr.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Insert needed holes for superblock migration
Qu Wenruo [Wed, 1 Jun 2016 08:29:43 +0000 (16:29 +0800)]
btrfs-progs: convert: Insert needed holes for superblock migration

New convert doesn't insert holes for superblock migration range.

Unlike the old design, which only relocates 4K (superblock size) to
other places.
In the new design, to make sure convert can handle different page sizes
and align chunks bytenr, we relocate the whole 64K range.

And if there is only a 4K used block inside 64K superblock migration
range, it will make converted the fs have discontiguous file extents.

This patch will fix it by inserting needed holes to avoid such
discontinuous error.

Reported-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Fix bugs in backup superblock migration
Qu Wenruo [Tue, 31 May 2016 08:49:33 +0000 (16:49 +0800)]
btrfs-progs: convert: Fix bugs in backup superblock migration

New convert has several bugs with backup superblock migration

1) Backup superblocks are not migrated due to incorrect condition
   Two wrong checks cause backup superblocks not to be migrated at all

2) Converted ext* image doesn't keep hole for backup superblocks
   Since we are creating file extents according to tmp_used, which has
   wiped out backup superblock ranges.
   In that case, later superblock migration will fail, since migration
   will insert file extent range into ext* image.

Fix above bugs will make convert on ext2 image filled about 100M data
successful.

Reported-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Fix a bug that makes convert asserts at scan time
Qu Wenruo [Tue, 31 May 2016 07:51:13 +0000 (15:51 +0800)]
btrfs-progs: convert: Fix a bug that makes convert asserts at scan time

When a ext2 fs filled with a 57M file, it's possible that convert fails
with assert in add_merge_cache_extent().

The problem is that the ext2 used space just takes some of the second
superblock.
And due to a bug in reserving superblock space, it corrupted used space
tree and cause assert.

Fix in by doing better used space merging for case where superblock
range is inside the ext2 used space.

Reported-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Add support to rollback new convert image
Qu Wenruo [Tue, 31 May 2016 02:46:00 +0000 (10:46 +0800)]
btrfs-progs: convert: Add support to rollback new convert image

For new btrfs-convert, it's less restrict for metadata chunk allocation.
While the may_rollback() function is still following the restrict 1:1
mapping check for all chunks, it will not allow some new convert image
to be rolled back.

Add new per extent check for new convert to allow it to be rolled back.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Cleanup old btrfs-convert
Qu Wenruo [Fri, 29 Jan 2016 05:03:32 +0000 (13:03 +0800)]
btrfs-progs: Cleanup old btrfs-convert

Cleanup all the old btrfs-convert facilities, including:
1) btrfs_convert_operations->alloc/free/test_extents*
   No need to do non-standard extent allocation.
   After init_btrfs() everything can be done by normal routine.

   Now only 4 functions are needed in btrfs_convert_operations.
   1) open_fs
   2) read_used_space
   3) copy_inodes
   4) close_fs

2) fs_info->extent_ops
   Same as above.

3) Old init_btrfs(), create_image(), create_file_image_range()
   Replaced with newer and cleaner one.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Strictly avoid meta or system chunk allocation
Qu Wenruo [Fri, 29 Jan 2016 05:03:31 +0000 (13:03 +0800)]
btrfs-progs: convert: Strictly avoid meta or system chunk allocation

Before this patch, btrfs-convert only rely on large enough initial
system/metadata chunk size to ensure no newer system/meta chunk will be
created.

But that's not safe enough. So add two new members in fs_info,
avoid_sys/meta_chunk_alloc flags to prevent any newer system or meta
chunks to be created before init_btrfs_v2().

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Convert: Add support for rollback new convert behavior
Qu Wenruo [Fri, 29 Jan 2016 05:03:30 +0000 (13:03 +0800)]
btrfs-progs: Convert: Add support for rollback new convert behavior

Add support to rollback new btrfs-convert.

The support is quite easy unlike the new convert behavior, which in fact
makes the rollback less restricted.

The core is, rollback must support both old and new convert result.

For old convert, all fs image extents can be covered by any chunk type.
But any extents inside reserved range must be covered by chunk.
For new convert, all fs image extents are covered by data chunk.

Allowing any condition will just make another fail to pass.
So make the convert condition a little less restricted to allow both can
be converted.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Introduce do_convert_v2 function
Qu Wenruo [Fri, 29 Jan 2016 05:03:29 +0000 (13:03 +0800)]
btrfs-progs: Introduce do_convert_v2 function

Introduce new function do_convert_v2() to do new convert.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[ use __strncpy_null for label ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Introduce init_btrfs_v2 function.
Qu Wenruo [Fri, 29 Jan 2016 05:03:28 +0000 (13:03 +0800)]
btrfs-progs: convert: Introduce init_btrfs_v2 function.

Introduce new init_btrfs_v2() function for later newer do_convert().

Since we have good enough chunk allocation, a lot of wired chunk hack
won't ever be used.
We only need to insert data chunks and create needed subvolume.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Enhance record_file_blocks to handle reserved ranges
Qu Wenruo [Fri, 29 Jan 2016 05:03:27 +0000 (13:03 +0800)]
btrfs-progs: convert: Enhance record_file_blocks to handle reserved ranges

Enhance record_file_blocks() to handle reserved ranges.

Old file system can use the space in btrfs reserved ranges.
So we could not use the bytenr of old filesystem directly.

Thanks to previous patches, we have a full fs image in convert_root, and
it has already relocated the blocks in reserved ranges.
So here we just search the convert_root to get correct disk_bytenr and
use it.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Introduce function to migrate reserved ranges
Qu Wenruo [Fri, 29 Jan 2016 05:03:26 +0000 (13:03 +0800)]
btrfs-progs: convert: Introduce function to migrate reserved ranges

Introduce new function, migrate_reserved_ranges() to migrate used fs
data in btrfs reserved space.

Unlike old implementation, which will need to relocate all the
complicated csum and reference relocation, previous patches already
ensure such reserved ranges won't be allocated.  So here we only need
copy these data out and create new extent/csum/reference.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Introduce new function to create converted image
Qu Wenruo [Fri, 29 Jan 2016 05:03:25 +0000 (13:03 +0800)]
btrfs-progs: convert: Introduce new function to create converted image

Use new function, create_convert_image_v2() to create snapshot of old
filesystem.

Unlike old function which is called after copying all inodes, this
function need to be called before copying inodes.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: extent-tree: Enhance btrfs_record_file_extent
Qu Wenruo [Fri, 29 Jan 2016 05:03:24 +0000 (13:03 +0800)]
btrfs-progs: extent-tree: Enhance btrfs_record_file_extent

Btrfs_record_file_extent() has some small problems like:
1) Can't handle overlapping extents
2) May create extent larger than BTRFS_MAX_EXTENT_SIZE

So enhance it using previously added facilites.
This is used for later btrfs-convert, as for new convert, we create
saved image first, then copy inode.
Which will also cause extent overlapping.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: extent-tree: Introduce function to find the first overlapping extent
Qu Wenruo [Fri, 29 Jan 2016 05:03:23 +0000 (13:03 +0800)]
btrfs-progs: extent-tree: Introduce function to find the first overlapping extent

Introduce a new function, btrfs_search_overlap_extent() to find the first
overlapping extent.

It's useful for later btrfs-convert rework.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Introduce function to create convert data chunks
Qu Wenruo [Fri, 29 Jan 2016 05:03:22 +0000 (13:03 +0800)]
btrfs-progs: Introduce function to create convert data chunks

Introduce new function, make_convert_data_chunks(), to build up data
chunks for convert.

It will call a modified version of btrfs_alloc_data_chunk() to force
data chunks to covert all known ext* data.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Introduce function to setup temporary extent tree
Qu Wenruo [Fri, 29 Jan 2016 05:03:21 +0000 (13:03 +0800)]
btrfs-progs: Introduce function to setup temporary extent tree

Introduce new function, setup_temporary_extent_tree() to build a
temporary extent tree for make_btrfs_v2().

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Introduce function to initialize csum tree
Qu Wenruo [Fri, 29 Jan 2016 05:03:20 +0000 (13:03 +0800)]
btrfs-progs: Introduce function to initialize csum tree

Introduce new function, setup_temp_csum_tree(), to setup temporary
csum tree for make_btrfs_v2().

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Introduce function to initialize fs tree
Qu Wenruo [Fri, 29 Jan 2016 05:03:19 +0000 (13:03 +0800)]
btrfs-progs: Introduce function to initialize fs tree

Introudce new function, setup_temp_fs_tree(), to setup temporary fs tree
for make_btrfs_v2().

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Introduce function to initialize device tree
Qu Wenruo [Fri, 29 Jan 2016 05:03:18 +0000 (13:03 +0800)]
btrfs-progs: Introduce function to initialize device tree

Introduce new function, setup_temp_device_tree(), to setup temporary
device tree for make_btrfs_v2().

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Introduce function to setup temporary chunk root
Qu Wenruo [Fri, 29 Jan 2016 05:03:17 +0000 (13:03 +0800)]
btrfs-progs: Introduce function to setup temporary chunk root

Introduce new function, setup_temp_chunk_root(), to initialize temporary
chunk root for make_btrfs_v2().

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Introduce function to setup temporary tree root
Qu Wenruo [Fri, 29 Jan 2016 05:03:16 +0000 (13:03 +0800)]
btrfs-progs: Introduce function to setup temporary tree root

Introduce new function, setup_temp_tree_root(), to initialize temporary
tree root for make_btrfs_v2().

The new function will setup tree root at metadata chunk and ensure data
won't be written into metadata chunk.

Also, new make_btrfs_v2() will have a much better code structure than
old make_btrfs().

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Introduce function to setup temporary superblock
Qu Wenruo [Fri, 29 Jan 2016 05:03:15 +0000 (13:03 +0800)]
btrfs-progs: Introduce function to setup temporary superblock

Introduce a new function, setup_temp_super(), to setup temporary super
for make_btrfs_v2().

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[ use __strncpy_null in setup_temp_super ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: utils: Introduce new function for convert
Qu Wenruo [Fri, 29 Jan 2016 05:03:14 +0000 (13:03 +0800)]
btrfs-progs: utils: Introduce new function for convert

Introduce new function make_convert_btrfs() for convert.

This new function will have the following features:
1) Allocate temporary sb/metadata/system chunk, avoiding old used data
2) More structured functions
   No more over 1000 lines function, better function split and code
   reuse

This will finally replace current make_btrfs(), but now only used for
convert.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Introduce function to calculate the available space
Qu Wenruo [Fri, 29 Jan 2016 05:03:13 +0000 (13:03 +0800)]
btrfs-progs: convert: Introduce function to calculate the available space

Introduce a new function, calculate_available_space() to get available
space cache_tree data_chunks cache_tree.

Unlike old implementation, this function will do the new work:

1) batch used ext* data space.
   To ensure data chunks will recovery them all.
   And restore the result into mkfs_cfg->convert_data_chunks for later
   use.

2) avoid SB and reserved space at chunk level
   Both batched data space or free space will not cover reserved space,
   like SB or the first 1M.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Introduce new function to remove reserved ranges
Qu Wenruo [Fri, 29 Jan 2016 05:03:12 +0000 (13:03 +0800)]
btrfs-progs: convert: Introduce new function to remove reserved ranges

Introduce functions to remove reserved ranges for later btrfs-convert
rework.

The reserved ranges includes:
1. [0,1M)
2. [btrfs_sb_offset(1), +BTRFS_STRIPE_LEN)
3. [btrfs_sb_offset(2), +BTRFS_STRIPE_LEN)

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: Introduce functions to read used space
Qu Wenruo [Fri, 29 Jan 2016 05:03:11 +0000 (13:03 +0800)]
btrfs-progs: convert: Introduce functions to read used space

Before we do real convert, we need to read and build up used space cache
tree for later data/meta separate chunk layout.

This patch will iterate all used blocks in ext2 filesystem and record it
into cctx->used cache tree, for later use.

This provides the very basic of later btrfs-convert rework.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: add compatibility layer for e2fsprogs < 1.42
Qu Wenruo [Mon, 9 May 2016 04:46:46 +0000 (12:46 +0800)]
btrfs-progs: convert: add compatibility layer for e2fsprogs < 1.42

The new convert framework copies code from current dumpe2fs, which uses
BIGALLOC feature introduced in e2fsprogs v1.42.

While there are a lot of enterprise distributions which are still using
v1.41 e2fsprogs, this will cause compile error for them.

This patch introduces backward compatibility for new convert framework,
by manually introduce macros for ext2 BIGALLOC feature.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: 004-ext2-backup-superblock-ranges, use common helpers for image...
David Sterba [Fri, 3 Jun 2016 15:08:43 +0000 (17:08 +0200)]
btrfs-progs: tests: 004-ext2-backup-superblock-ranges, use common helpers for image loop

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: 004-ext2-backup-superblock-ranges, drop unnecessary root privs
David Sterba [Fri, 3 Jun 2016 14:52:57 +0000 (16:52 +0200)]
btrfs-progs: tests: 004-ext2-backup-superblock-ranges, drop unnecessary root privs

We really use root only for mount/umount and access to the ext2_saved
image (that has 0600). Also switch to common variable so we can use
helpers.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: unify test drivers
David Sterba [Fri, 3 Jun 2016 14:01:51 +0000 (16:01 +0200)]
btrfs-progs: tests: unify test drivers

Remove unnecessary code, add exports to all common variables.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: convert: set common variables
David Sterba [Fri, 3 Jun 2016 13:54:32 +0000 (15:54 +0200)]
btrfs-progs: tests: convert: set common variables

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: clean up the test driver of convert tests
David Sterba [Fri, 3 Jun 2016 13:46:39 +0000 (15:46 +0200)]
btrfs-progs: tests: clean up the test driver of convert tests

Everything is now in separate tests, and TEST=mask now works.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: convert: separate ext4 tests
David Sterba [Fri, 3 Jun 2016 13:37:25 +0000 (15:37 +0200)]
btrfs-progs: tests: convert: separate ext4 tests

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: convert: separate ext3 tests
David Sterba [Fri, 3 Jun 2016 13:37:25 +0000 (15:37 +0200)]
btrfs-progs: tests: convert: separate ext3 tests

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: convert: separate ext2 tests
David Sterba [Fri, 3 Jun 2016 13:37:25 +0000 (15:37 +0200)]
btrfs-progs: tests: convert: separate ext2 tests

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: move convert helpers to a separate file
David Sterba [Fri, 3 Jun 2016 13:28:44 +0000 (15:28 +0200)]
btrfs-progs: tests: move convert helpers to a separate file

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert-tests: Add test case for backup superblock migration
Qu Wenruo [Fri, 3 Jun 2016 02:34:26 +0000 (10:34 +0800)]
btrfs-progs: convert-tests: Add test case for backup superblock migration

New convert introduced simpler chunk/extent allocation algorithm, at the
cost of complex backup superblock migration codes.

Use specially built ext2 images to test if btrfs-convert can convert and
rollback images without problem.

All these special ext2 image have blocks/holes across 2nd btrfs backup
superblock.

The naming of test image is like the following:
|<------superblock migration range----->|
64M 64M + 64K
|-Data--|-Data--|/Hole//|-Data--|/Hole//|-Data--|--Data--| = drdhdhdrd

These test cases should check all typical layouts and make sure new
convert works.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert-tests: Add support for custom test scripts
Qu Wenruo [Fri, 3 Jun 2016 02:34:25 +0000 (10:34 +0800)]
btrfs-progs: convert-tests: Add support for custom test scripts

Add support for custom convert test scripts, just like fsck tests.

Instead of generic convert tests, we need more specifically created images
for new convert tests.

This patch provide the needed infrastructure for later convert test
cases.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: btrfs-crc: make argc check more strict
Satoru Takeuchi [Thu, 2 Jun 2016 08:14:57 +0000 (17:14 +0900)]
btrfs-progs: btrfs-crc: make argc check more strict

Signed-off-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: btrfs-crc: improve usage message
Satoru Takeuchi [Thu, 2 Jun 2016 08:13:10 +0000 (17:13 +0900)]
btrfs-progs: btrfs-crc: improve usage message

- If -c is set, filename argument is ignored.
- Describe about -h option

Signed-off-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: btrfs-crc: print usage on receiving invalid arguments
Satoru Takeuchi [Thu, 2 Jun 2016 08:11:51 +0000 (17:11 +0900)]
btrfs-progs: btrfs-crc: print usage on receiving invalid arguments

Usage is only printed if -h option is set. However it's nice to
do it when wrong option is set or the number of argument is wrong.

Signed-off-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: btrfs-crc should be ignored by git
Satoru Takeuchi [Thu, 2 Jun 2016 08:10:18 +0000 (17:10 +0900)]
btrfs-progs: btrfs-crc should be ignored by git

It's a binary built from btrfs-crc.c

Signed-off-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: btrfs-crc: fix build error
Satoru Takeuchi [Thu, 2 Jun 2016 08:06:37 +0000 (17:06 +0900)]
btrfs-progs: btrfs-crc: fix build error

Remove the following build error.

   ====================================
   $ make btrfs-crc
       [CC]     btrfs-crc.o
       [LD]     btrfs-crc
   btrfs-crc.o: In function `usage':
   /home/sat/src/btrfs-progs/btrfs-crc.c:26: multiple definition of `usage'
   help.o:/home/sat/src/btrfs-progs/help.c:125: first defined here
   collect2: error: ld returned 1 exit status
   Makefile:294: recipe for target 'btrfs-crc' failed
   make: *** [btrfs-crc] Error 1
   =====================================

Signed-off-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: use wider int type in btrfs_min_global_blk_rsv_size
David Sterba [Thu, 2 Jun 2016 16:18:41 +0000 (18:18 +0200)]
btrfs-progs: use wider int type in btrfs_min_global_blk_rsv_size

We know nodesize should not overflow with the shift, but le'ts make the
code correct if the resulting type can store the full value.

Resolves-coverity-id: 1358120
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Use new random number API
Qu Wenruo [Thu, 26 May 2016 02:56:51 +0000 (10:56 +0800)]
btrfs-progs: Use new random number API

Replace old and not so informal srand()/rand() calls to new random
number API.

Including btrfs-corrupt-block(main user), btrfs-image and btrfs-crc.

Some tests like dir-test/random-test/quick-test is not modified.
As random-test itself can't even pass build.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: utils: Introduce new pseudo random API
Qu Wenruo [Thu, 26 May 2016 02:56:50 +0000 (10:56 +0800)]
btrfs-progs: utils: Introduce new pseudo random API

David has reported some quite chaos usage of pseudo random numbers.
Like using static srand seed, or even calling rand() without setting
seed correctly.

The new pseudo random API will initialize the random seed on its first
calling and use uniformly distributed pseudo random number generator as
backend.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[ renamed variables and functions, added prefixes ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: corrupt-block: Add support to corrupt extent for skinny metadata
Qu Wenruo [Thu, 26 May 2016 02:36:02 +0000 (10:36 +0800)]
btrfs-progs: corrupt-block: Add support to corrupt extent for skinny metadata

For skinny metadata case, btrfs-corrupt-block can't corrupt a metadata
extent due to the type check missing BTRFS_METADATA_ITEM_KEY.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert-tests: Add test for backup superblock migration
Qu Wenruo [Wed, 1 Jun 2016 01:51:21 +0000 (09:51 +0800)]
btrfs-progs: convert-tests: Add test for backup superblock migration

New convert framework uses new and simpler chunk layout, while the cost
is the more complex superblock range migration logical, compared to old
convert.

Enhance the convert test script to create file which will takes up 2nd
backup superblock space, to ensure the superblock migration is working
as expected.

Suggested-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: document cli-tests in readme
David Sterba [Wed, 1 Jun 2016 11:22:44 +0000 (13:22 +0200)]
btrfs-progs: tests: document cli-tests in readme

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: convert: dump all superblocks after conversion
David Sterba [Tue, 31 May 2016 17:17:02 +0000 (19:17 +0200)]
btrfs-progs: tests: convert: dump all superblocks after conversion

We want to see all of them, even if they're not valid.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: make btrfs-image restore to support dup
Lu Fengqi [Thu, 26 May 2016 09:43:00 +0000 (17:43 +0800)]
btrfs-progs: make btrfs-image restore to support dup

Previously btrfs-image restore would set the chunk items to have 1 stripe,
even if the chunk is dup. If you use btrfsck on the restored file system,
some dev_extent will not find any relative chunk stripe, and the
bytes-used of dev_item will not equal to the dev_extents's total_bytes.
This patch store a additional physical just for the dup case when build
the in-memory chunk-tree.
Currently btrfsck on the restored file system, only single and dup is no
problem. raid* support should be added in the future.

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: run rollback after conversion
David Sterba [Mon, 30 May 2016 15:11:19 +0000 (17:11 +0200)]
btrfs-progs: tests: run rollback after conversion

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: convert, run md5sum with sudo helper
David Sterba [Mon, 30 May 2016 15:07:35 +0000 (17:07 +0200)]
btrfs-progs: tests: convert, run md5sum with sudo helper

Some of the files might not end up in the checksum list because of
permissions. This is reported by md5sum as incorrectly formatted lines.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: add 020-extent-ref-cases
Lu Fengqi [Mon, 30 May 2016 02:58:13 +0000 (10:58 +0800)]
btrfs-progs: tests: add 020-extent-ref-cases

In order to confirm that btrfsck supports to check a variety of
refs, add the
following cases:
* keyed_block_ref
* keyed_data_ref
* shared_block_ref
* shared_data_ref
* no_inline_ref (a extent item without inline ref)
* no_skinny_ref

Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: doc: fix typo in btrfs-subvolume
Merlin Hartley [Thu, 19 May 2016 14:17:24 +0000 (15:17 +0100)]
btrfs-progs: doc: fix typo in btrfs-subvolume

Signed-off-by: Merlin Hartley <merlinhartley@hotmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: switch to common message helpers in utils.c
David Sterba [Fri, 20 May 2016 11:39:02 +0000 (13:39 +0200)]
btrfs-progs: switch to common message helpers in utils.c

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: fix type mismatch in backtrace dumping functions
David Sterba [Fri, 20 May 2016 11:20:28 +0000 (13:20 +0200)]
btrfs-progs: fix type mismatch in backtrace dumping functions

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: drop O_CREATE from open_ctree_fs_info
David Sterba [Fri, 20 May 2016 11:09:17 +0000 (13:09 +0200)]
btrfs-progs: drop O_CREATE from open_ctree_fs_info

We stat the filesystem path before trying to open it so there's no point
to pass O_CREAT ("btrfs-progs: add stat check in open_ctree_fs_info").

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Return earlier for previous item
Qu Wenruo [Thu, 19 May 2016 02:54:39 +0000 (10:54 +0800)]
btrfs-progs: Return earlier for previous item

Follow kernel code to return earlier for btrfs_previous_item() function.

Before this patch, btrfs_previous_item() doesn't use its min_objectid to
exit, this makes caller to check key to exit, and if caller doesn't
check, it will iterate all previous item.

This patch will check min_objectid and type, to early return and save
some time.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Enhance tree block check by checking empty leaf or node
Qu Wenruo [Thu, 19 May 2016 08:44:35 +0000 (16:44 +0800)]
btrfs-progs: Enhance tree block check by checking empty leaf or node

For btrfs, it's possible to have empty leaf, but empty node is not
possible.

Add check for empty node for tree blocks.

Suggested-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: makefile: add target for testing installation
David Sterba [Thu, 19 May 2016 12:20:51 +0000 (14:20 +0200)]
btrfs-progs: makefile: add target for testing installation

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: clean up commands.h
Anand Jain [Thu, 19 May 2016 05:13:14 +0000 (13:13 +0800)]
btrfs-progs: clean up commands.h

This function is declared in utils.h so remove it from commands.h int
test_issubvolume(const char *path);

This function does not exists delete the declaration
get_subvol_name(char *mnt, char *full_path);

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: makefile: add clean-all to the usage
Anand Jain [Thu, 19 May 2016 05:13:13 +0000 (13:13 +0800)]
btrfs-progs: makefile: add clean-all to the usage

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: device usage: report slack space
David Sterba [Tue, 17 May 2016 14:21:00 +0000 (16:21 +0200)]
btrfs-progs: device usage: report slack space

The total filesystem space on a given device might be smaller than the
device size. We should report that space as well. The original idea was
to report the 'occupied' size but the term was not all clear, so the
logic was reversed to report the slack space.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: compression is disabled with nodatasum/nodatacow
David Sterba [Tue, 17 May 2016 14:13:44 +0000 (16:13 +0200)]
btrfs-progs: docs: compression is disabled with nodatasum/nodatacow

Explicitly mention the constraints in all involved options.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: autogen: Make build success in CentOS 6 and 7
Zhao Lei [Fri, 13 May 2016 03:44:52 +0000 (11:44 +0800)]
btrfs-progs: autogen: Make build success in CentOS 6 and 7

btrfs-progs build failed in CentOS 6 and 7:
 #./autogen.sh
 ...
 configure.ac:131: error: possibly undefined macro: PKG_CHECK_VAR
      If this token and others are legitimate, please use m4_pattern_allow.
      See the Autoconf documentation.
 ...

Seems PKG_CHECK_VAR is new in pkgconfig 0.28 (24-Jan-2013):
http://redmine.audacious-media-player.org/boards/1/topics/736

And the max available version for CentOS 7 in yum-repo and
rpmfind.net is: pkgconfig-0.27.1-4.el7
http://rpmfind.net/linux/rpm2html/search.php?query=pkgconfig&submit=Search+...&system=centos&arch=

I updated my pkgconfig to 0.30, but still failed at above error.
(Maybe it is my setting problem)

To make user in centos 6 and 7 building btrfs-progs without
more changes, we can avoid using PKG_CHECK_VAR in following
way found in:
https://github.com/audacious-media-player/audacious-plugins/commit/f95ab6f939ecf0d9232b3165f9241d2ea9676b9e

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: autogen: Don't show success message on fail
Zhao Lei [Fri, 13 May 2016 03:44:53 +0000 (11:44 +0800)]
btrfs-progs: autogen: Don't show success message on fail

When autogen.sh failed, the success message is still in output:
 # ./autogen.sh
 ...
 configure.ac:131: error: possibly undefined macro: PKG_CHECK_VAR
      If this token and others are legitimate, please use m4_pattern_allow.
      See the Autoconf documentation.

 Now type './configure' and 'make' to compile.
 #

Fixed by check return value of autoconf.

After patch:
 # ./autogen.sh
 ...
 configure.ac:132: error: possibly undefined macro: PKG_CHECK_VAR
      If this token and others are legitimate, please use m4_pattern_allow.
      See the Autoconf documentation.
 #

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: autogen: Avoid chdir fail on dirname with blank
Zhao Lei [Fri, 13 May 2016 03:44:51 +0000 (11:44 +0800)]
btrfs-progs: autogen: Avoid chdir fail on dirname with blank

If source put in dir with blanks, as:
  /var/lib/jenkins/workspace/btrfs progs

autogen will failed:
./autogen.sh: line 95: cd: /var/lib/jenkins/workspace/btrfs: No such file or directory

Can be fixed by adding quotes into cd command.

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: send: add quiet option
M G Berberich [Thu, 12 May 2016 19:49:25 +0000 (21:49 +0200)]
btrfs-progs: send: add quiet option

Add new options -q,--quiet to prevent printing messages on stderr, added
--verbose as alternative for -v.  Moved 'Mode NO_FILE_DATA enabled'
message to stderr. The default verboisty level is 1 to keep some
backward compatibility.

Signed-off-by: M G Berberich <btrfs@oss.m-berberich.de>
[ minor adjustments in the options, help text and changelog, added
  manual page text ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: build: add support for debugging builds
David Sterba [Thu, 12 May 2016 12:13:30 +0000 (14:13 +0200)]
btrfs-progs: build: add support for debugging builds

Add an easy way to build a debugging version, without optimizations and
with debugging info by default. Any overrides can be done via
DEBUG_CFLAGS that are added on top of the defaults. Can be combined with
V, C or W options.

Usage:
  $ make D=1

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: typo review of strings and comments
Nicholas D Steeves [Wed, 11 May 2016 23:50:36 +0000 (19:50 -0400)]
btrfs-progs: typo review of strings and comments

Signed-off-by: Nicholas D Steeves <nsteeves@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: udev: add rules for dm devices
Jeff Mahoney [Mon, 9 May 2016 15:44:26 +0000 (11:44 -0400)]
btrfs-progs: udev: add rules for dm devices

Systemd's btrfs rule runs btrfs dev ready on each device
as it's discovered.  The btrfs command is executed as a builtin
command via an IMPORT{builtin} rule, which means it gets
executed at rule evaluation time, not rule execution time.  That
means that the device mapper links haven't been setup yet and the only
nodes that can be depended upon are /dev/dm-#.  That we see
/dev/mapper/name names in /proc/mounts is only because we replace the
device name we have cached with the one passed in via mount.  If
we have a multi-device file system and the primary device is removed,
the remaining devices will show /dev/dm-#.  In addition, if the
udev rule is executed again by someone generating a change event (e.g.
partprobe), the names are also replaced by the /dev/dm-# names.

This patch adds a new rule that adds a run rule that calls btrfs dev
ready again using the device mapper links once they're created.

Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: prop: remove conditions which never be satisfied
Satoru Takeuchi [Mon, 9 May 2016 07:52:11 +0000 (16:52 +0900)]
btrfs-progs: prop: remove conditions which never be satisfied

parse_args() always set at least one parameter, 'object', for
{get,list} subcommands. In addition, it always set all three
parameters, 'object', 'name', and 'value' for set subcommand.
So the following conditions can be removed.

Signed-off-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: prop: simplify parse_args()
Satoru Takeuchi [Mon, 9 May 2016 07:49:19 +0000 (16:49 +0900)]
btrfs-progs: prop: simplify parse_args()

Since <object> parameter is mandatory for all subcommands,
'object' is always set by parse_args()'s callers.
In addition, on setting '*name' and '*value', if 'optind < argc'
is satisfied here, they are always set by parse_args()'s callers.

Signed-off-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agoBtrfs progs v4.5.3
David Sterba [Wed, 11 May 2016 14:40:33 +0000 (16:40 +0200)]
Btrfs progs v4.5.3

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: update CHANGES for 4.5.3
David Sterba [Mon, 2 May 2016 13:32:02 +0000 (15:32 +0200)]
btrfs-progs: update CHANGES for 4.5.3

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: add fuzzed testing images, superblock and chunks
Liu Bo [Mon, 2 May 2016 18:18:55 +0000 (11:18 -0700)]
btrfs-progs: add fuzzed testing images, superblock and chunks

This adds 4 fuzz testing images, btrfsck either doesn't detect errors
in them or crashes immediately.

Reported-by: Vegard Nossum <vegard.nossum@oracle.com>
Reported-by: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>