platform/upstream/btrfs-progs.git
8 years agobtrfs-progs: extend balance args to take min/max usage filter
David Sterba [Thu, 22 Oct 2015 08:19:38 +0000 (10:19 +0200)]
btrfs-progs: extend balance args to take min/max usage filter

Add the overlapping usage and [usage_min, usage_max] members to the
balance args. The min/max values are interpreted iff the corresponding
flag BTRFS_BALANCE_ARGS_USAGE_RANGE is set.

The minimum boundary is inclusive, maximum is exclusive:
* usage_min <= chunk_usage < usage_max

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: balance: add stripes filter
Gabríel Arthúr Pétursson [Mon, 28 Sep 2015 22:33:04 +0000 (22:33 +0000)]
btrfs-progs: balance: add stripes filter

Add new balance filter 'stripes=<range>' to process only chunks that are
spread accross given number of chunks.

The range minimum and maximum are inclusive.

Signed-off-by: Gabríel Arthúr Pétursson <gabriel@system.is>
[ reworked a bit to use the range helpers, dropped the single value
  for stripes ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: balance: enhance the limit fiter with range
David Sterba [Sun, 11 Oct 2015 22:52:43 +0000 (00:52 +0200)]
btrfs-progs: balance: enhance the limit fiter with range

We can do more with the balance filter. Enhance it so we can specify
also the minimum number of block groups to process. The 'limit' filter
now accepts a range (a..b, can be partial) and needs kernel support.

The 'limit=value' filter is equivalent to 'limit=..value' but works on
older kernels as well.

The min/max values are 32bit, unlike the single-value limit which is
64bit.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: extend balance args to take min/max limit filter
David Sterba [Sun, 11 Oct 2015 22:34:58 +0000 (00:34 +0200)]
btrfs-progs: extend balance args to take min/max limit filter

Add the overlapping limit and [limit_min, limit_max] members to the
balance args. The min/max values are interpreted iff the corresponding
flag BTRFS_BALANCE_ARGS_LIMIT_RANGE is set.

The minimum and maximum are inclusive.

Note that the values are only 32bit, but this should be enough for the
foreseeable future.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: test multiple-linked file corruption
Naohiro Aota [Tue, 8 Dec 2015 02:07:13 +0000 (11:07 +0900)]
btrfs-progs: tests: test multiple-linked file corruption

This commit extends the leaf corruption test to try to repair a file
linked from multiple directory. It stresses a case that some links to a
file is broken but others kept valid.

Signed-off-by: Naohiro Aota <naota@elisp.net>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Introduce new members for btrfs_convert_context
Qu Wenruo [Tue, 1 Dec 2015 07:11:23 +0000 (15:11 +0800)]
btrfs-progs: Introduce new members for btrfs_convert_context

Introduce 3 new members for btrfs_convert_context:

1) struct cache_tree used
Records accurate byte ranges which are used by old filesystem.
This will be used to create old filesystem image.

2) struct cache_tree data_chunks
Records batched ranges which must be covered by data chunks.
The bytenr range is optimized to meet all the chunk requirement.

3) u64 total_bytenr
Records how large the filesystem is in bytenr.
Yes, we can calculate it easy, but that's for old blocks based
filesystem.
This will make it more friendly for extent based filesystem.
And later cctx->block_counts and may be removed

And 2 for mkfs_config:
1) char *chunk_uuid.
Used as temporary chunk_uuid (unparsed) string for later
make_convert_btrfs()

2) u64 super_bytenr
Records the new temporary super bytenr after make_btrfs().

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: extent-tree: Add add_merge_cache_extent function
Qu Wenruo [Tue, 1 Dec 2015 07:11:22 +0000 (15:11 +0800)]
btrfs-progs: extent-tree: Add add_merge_cache_extent function

This add_merge_cache_extent() function will try to merge adjusted
cache_extent.

This is used for later btrfs-convert ext2 free space cache.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: extent-cache: Add comments for search/lookup functions
Qu Wenruo [Tue, 1 Dec 2015 07:11:21 +0000 (15:11 +0800)]
btrfs-progs: extent-cache: Add comments for search/lookup functions

There are quite a lot search/lookup functions with different behavior,
add comments for them, as it will take extra time to view source to
understand the behavior difference.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: check, properly reset nlink of multi-linked file
Naohiro Aota [Fri, 4 Dec 2015 05:37:26 +0000 (14:37 +0900)]
btrfs-progs: check, properly reset nlink of multi-linked file

If a file is linked from more than one directory and only one
of the links is corrupted, btrfs check dose not reset the nlink
properly. Actually it can go into infinite loop to link the broken file
into lost+found.

This patch fix two part of the code. The first one delay the freeing
valid (no error, found inode ref, directory index, and directory
item) backrefs. Freeing valid backrefs earier prevent reset_nlink() to
add back all valid links.

The second fix is obvious: passing `ref_type' to btrfs_add_link() is just
wrong. It should be `filetype' instead. The current code can break all valid
file links.

Signed-off-by: Naohiro Aota <naota@elisp.net>
Reviewed-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: print test type
David Sterba [Mon, 30 Nov 2015 17:13:01 +0000 (18:13 +0100)]
btrfs-progs: tests: print test type

Makes it a bit more clear:

    [TEST/conv]   ext4 32k nodesize, btrfs no-holes
    [TEST/conv]   ext2 64k nodesize, btrfs no-holes
    [TEST/conv]   ext3 64k nodesize, btrfs no-holes
    [TEST/conv]   ext4 64k nodesize, btrfs no-holes
    [TEST]        misc-tests.sh
    [TEST/misc]   001-btrfstune-features
    [TEST/misc]   002-uuid-rewrite
    [TEST/misc]   003-zero-log
    [TEST/misc]   004-shrink-fs
    [TEST/misc]   005-convert-progress-thread-crash
    [TEST/misc]   006-image-on-missing-device
    [TEST/misc]   007-subvolume-sync
    [TEST/misc]   008-leaf-crossing-stripes
    [TEST/misc]   009-subvolume-sync-must-wait
    [TEST/misc]   010-convert-delete-ext2-subvol
    [TEST/misc]   011-delete-missing-device
    [TEST]        fuzz-tests.sh
    [TEST/fuzz]   001-simple-unmounted

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: show-super: copy implementation of btrfs_read_sys_array
David Sterba [Mon, 30 Nov 2015 16:56:58 +0000 (17:56 +0100)]
btrfs-progs: show-super: copy implementation of btrfs_read_sys_array

btrfs-show-super has own implementation to read the sys array and it
lacks all the sanity checks.

Copy and adapt the improved implementation from btrfs_read_sys_array.
This will fix crashes on a fuzzed/corrupted images. It's duplicating
code but at the moment looks easier than refactoring and restructuring
btrfs_read_sys_array that now almost matches the kernel code, which we
want in the long term.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: add sys-array-num-stripes-0.raw.xz
David Sterba [Mon, 30 Nov 2015 16:42:15 +0000 (17:42 +0100)]
btrfs-progs: tests: add sys-array-num-stripes-0.raw.xz

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: verbose error when we find an unexpected item in sys_array
David Sterba [Mon, 30 Nov 2015 16:10:20 +0000 (17:10 +0100)]
btrfs-progs: verbose error when we find an unexpected item in sys_array

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: handle invalid num_stripes in sys_array
David Sterba [Mon, 30 Nov 2015 15:44:29 +0000 (16:44 +0100)]
btrfs-progs: handle invalid num_stripes in sys_array

We can handle the special case of num_stripes == 0 directly inside
btrfs_read_sys_array. The BUG_ON in btrfs_chunk_item_size is there to
catch other unhandled cases where we fail to validate external data,
like in btrfs-show-super.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: add more checks to btrfs_read_sys_array
David Sterba [Wed, 5 Nov 2014 14:24:51 +0000 (15:24 +0100)]
btrfs-progs: add more checks to btrfs_read_sys_array

Port of kernel commit e3540eab29e1b2260bc4b9b3979a49a00e3e3af8

Verify that the sys_array has enough bytes to read the next item.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: cleanup, rename a few variables in btrfs_read_sys_array
David Sterba [Mon, 30 Nov 2015 15:07:35 +0000 (16:07 +0100)]
btrfs-progs: cleanup, rename a few variables in btrfs_read_sys_array

Port of kernel commit 1ffb22cf8c322bbfea6b35fe23d025841b49fede

There's a pointer to buffer, integer offset and offset passed as
pointer, try to find matching names for them.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: update btrfs_read_sys_array to match kernel implementation
David Sterba [Mon, 30 Nov 2015 14:57:36 +0000 (15:57 +0100)]
btrfs-progs: update btrfs_read_sys_array to match kernel implementation

There are some sanity checks missing on both sides, kernel/userspace.
Preparation to port the missing changes.

Sync code with parent of kernel commit
1ffb22cf8c322bbfea6b35fe23d025841b49fede ("btrfs: cleanup, rename a few
variables in btrfs_read_sys_array")

This effectively reverts progs commit
be96777126d283773e4397278a5662d90676ab88 ("btrfs-progs: Cleanup unneeded
extra variant in btrfs_read_sys_array") so we can apply more of the
kernel patches.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: mkfs, check sectorsize and nodesize combinations
Qu Wenruo [Mon, 30 Nov 2015 01:39:30 +0000 (09:39 +0800)]
btrfs-progs: tests: mkfs, check sectorsize and nodesize combinations

Add mkfs selftest for invalid and valid sectorsize/nodesize
combinations.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[ switched to TEST_DEV ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: install to /usr/local by default again
David Sterba [Mon, 30 Nov 2015 09:36:28 +0000 (10:36 +0100)]
btrfs-progs: install to /usr/local by default again

The conversion to autotools changed the default prefix to /usr. There's
no reason to diverge. Distributions builds set prefix the /usr path and
local builds are supposed to go to /usr/local .

Reference: https://bugzilla.kernel.org/show_bug.cgi?id=108571
Reported-by: Karl Richter <krichter722@aol.de>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: mkfs, implications of DUP on devices
David Sterba [Thu, 26 Nov 2015 16:56:45 +0000 (17:56 +0100)]
btrfs-progs: docs: mkfs, implications of DUP on devices

We offer DUP but still depend on the hardware, to do the right thing.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: enhance 001-basic-profiles with --data DUP
David Sterba [Thu, 26 Nov 2015 15:22:16 +0000 (16:22 +0100)]
btrfs-progs: tests: enhance 001-basic-profiles with --data DUP

Add the remaining valid combinations.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: mkfs: allow --data DUP for single device
Zhao Lei [Thu, 19 Nov 2015 09:36:24 +0000 (17:36 +0800)]
btrfs-progs: mkfs: allow --data DUP for single device

Current code don't support DUP profile on single device, except it
is in mixed mode, because of following reasons:

1: Some SSD do deduplication internally, so the duplication on
   the filesystem side has no effect.
2: On a physical device, if the entire disk broken, --data DUP does not
   help.
3: Half performance compared to single profile.
4: We have a workaround: create multi-partition on a single device,
   and btffs will treat them as multi device.

Instead of refusing --data DUP, we give the user a choice and print
a wrning.

Test:
1: Tested by xfstests
   Run with modified xfstests, I add test items of -d dup in single
   device into btrfs/* and common/rc, run tests of btrfs/*,
   with all mount option, no regression diffed with v4.3.
2: Tested by btrfs-progs
   Checked following commands in "-m dup -d dup" fs with memleck
   checking, all passed:
   mkfs.btrfs -f --data dup --metadata dup /dev/sda6
   btrfs filesystem show /dev/sda6
   btrfs filesystem label /dev/sda6 btrfs_label_test
   btrfs filesystem label /dev/sda6
   btrfs device scan --all-devices
   btrfs device scan /dev/sda6
   btrfs device scan /dev/sda6
   btrfs device ready /dev/sda6
   btrfs check /dev/sda6
   btrfs check -s 1 /dev/sda6
   btrfs check --repair /dev/sda6
   btrfs check --init-csum-tree /dev/sda6
   btrfs check --init-extent-tree /dev/sda6
   btrfs check --check-data-csum /dev/sda6
   btrfs check --qgroup-report /dev/sda6
   btrfs rescue super-recover -y /dev/sda6
   btrfs rescue zero-log /dev/sda6
   btrfs restore -l /dev/sda6
   btrfs restore /dev/sda6 /
   btrfs restore -s /dev/sda6 /
   btrfs restore -x /dev/sda6 /
   btrfs restore -m /dev/sda6 /
   btrfs restore -S /dev/sda6 /
   btrfs restore -v /dev/sda6 /
   btrfs restore -i /dev/sda6 /
   btrfs restore -o /dev/sda6 /
   btrfs restore -u0 /dev/sda6 /
   btrfs restore -u1 /dev/sda6 /
   btrfs restore -D /dev/sda6 /
   btrfs property list /dev/sda6
   btrfs property get /dev/sda6 label
   btrfs property set /dev/sda6 label test
   btrfs property set /dev/sda6 label btrfs_label_test
   btrfs help
   btrfs help --full
   btrfs version
   btrfsck /dev/sda6
   btrfs-find-root /dev/sda6
   btrfs-find-root -a /dev/sda6
   btrfs-map-logical -l1 /dev/sda6
   btrfs-map-logical -l1 -c1 /dev/sda6
   btrfs-map-logical -l1 -o /tmp/btrfs-map-logic-out /dev/sda6
   btrfs-map-logical -l1 -b1 /dev/sda6
   btrfs-select-super -s 0 /dev/sda6
   btrfs-select-super -s 1 /dev/sda6
   btrfstune -S 1 /dev/sda6
   btrfstune -f -S 0 /dev/sda6
   btrfstune -r /dev/sda6
   btrfstune -x /dev/sda6
   btrfstune -n /dev/sda6
   btrfstune -f -U 00000000-0000-0000-0000-000000000000 /dev/sda6
   btrfstune -f -u /dev/sda6
   btrfs-calc-size /dev/sda6
   btrfs-calc-size -v /dev/sda6
   btrfs-calc-size -b /dev/sda6
   btrfs-debug-tree /dev/sda6
   btrfs-debug-tree -e /dev/sda6
   btrfs-debug-tree -d /dev/sda6
   btrfs-debug-tree -r /dev/sda6
   btrfs-debug-tree -R /dev/sda6
   btrfs-debug-tree -u /dev/sda6
   btrfs-debug-tree -b 0 /dev/sda6
   btrfs-debug-tree -t 0 /dev/sda6
   btrfs-debug-tree -t 2 /dev/sda6
   btrfs-show-super /dev/sda6
   btrfs-show-super -i 0 /dev/sda6
   btrfs-show-super -i 1 /dev/sda6
   btrfs-show-super -i 2 /dev/sda6
   btrfs-show-super -a /dev/sda6
   btrfs-show-super -f /dev/sda6
   btrfs-show-super -F /dev/sda6
   btrfs subvolume list /mnt/btrfs-progs-tests
   btrfs subvolume create /mnt/btrfs-progs-tests/mysubvol
   btrfs subvolume list /mnt/btrfs-progs-tests
   btrfs subvolume get-default /mnt/btrfs-progs-tests
   btrfs subvolume set-default 258 /mnt/btrfs-progs-tests
   btrfs subvolume get-default /mnt/btrfs-progs-tests
   btrfs subvolume set-default  /mnt/btrfs-progs-tests
   btrfs subvolume snapshot /mnt/btrfs-progs-tests/mysubvol /mnt/btrfs-progs-tests/mysubvol_snap
   btrfs subvolume list /mnt/btrfs-progs-tests
   btrfs subvolume find-new /mnt/btrfs-progs-tests 0
   btrfs subvolume find-new /mnt/btrfs-progs-tests 0
   btrfs subvolume find-new /mnt/btrfs-progs-tests/mysubvol 0
   btrfs subvolume find-new /mnt/btrfs-progs-tests/mysubvol 0
   btrfs subvolume show /mnt/btrfs-progs-tests
   btrfs subvolume show /mnt/btrfs-progs-tests/mysubvol
   btrfs subvolume show /mnt/btrfs-progs-tests/mysubvol_snap
   btrfs subvolume sync /mnt/btrfs-progs-tests
   btrfs subvolume delete /mnt/btrfs-progs-tests/mysubvol_snap
   btrfs subvolume delete /mnt/btrfs-progs-tests/mysubvol
   btrfs subvolume sync /mnt/btrfs-progs-tests
   btrfs filesystem df /mnt/btrfs-progs-tests
   btrfs filesystem show /mnt/btrfs-progs-tests
   btrfs filesystem sync /mnt/btrfs-progs-tests
   btrfs filesystem label /mnt/btrfs-progs-tests btrfs_label_test
   btrfs filesystem label /mnt/btrfs-progs-tests
   btrfs filesystem usage /mnt/btrfs-progs-tests
   btrfs filesystem defragment -s 1024 -l 2048 /mnt/btrfs-progs-tests/filesystem_test_dir/test_dir_0/test_file_0
   btrfs filesystem defragment /mnt/btrfs-progs-tests/filesystem_test_dir/test_dir_0/test_file_1
   btrfs filesystem defragment -f /mnt/btrfs-progs-tests/filesystem_test_dir/test_dir_0/test_file_2
   btrfs filesystem defragment -czlib /mnt/btrfs-progs-tests/filesystem_test_dir/test_dir_0/test_file_3
   btrfs filesystem defragment -clzo /mnt/btrfs-progs-tests/filesystem_test_dir/test_dir_0/test_file_4
   btrfs filesystem defragment /mnt/btrfs-progs-tests/filesystem_test_dir
   btrfs filesystem defragment -r /mnt/btrfs-progs-tests/filesystem_test_dir
   btrfs filesystem defragment /mnt/btrfs-progs-tests
   btrfs filesystem resize 1:-10M /mnt/btrfs-progs-tests
   btrfs filesystem resize 1:max /mnt/btrfs-progs-tests
   btrfs balance start /mnt/btrfs-progs-tests
   btrfs balance start -v /mnt/btrfs-progs-tests
   btrfs balance start -f /mnt/btrfs-progs-tests
   btrfs balance status -v /mnt/btrfs-progs-tests
   btrfs balance pause /mnt/btrfs-progs-tests
   btrfs balance status /mnt/btrfs-progs-tests
   btrfs balance resume /mnt/btrfs-progs-tests
   btrfs balance status -v /mnt/btrfs-progs-tests
   btrfs balance cancel /mnt/btrfs-progs-tests
   btrfs balance start -dprofiles=single /mnt/btrfs-progs-tests
   btrfs balance start -dconvert=single /mnt/btrfs-progs-tests
   btrfs balance start -ddevid=1 /mnt/btrfs-progs-tests
   btrfs balance start -f -mprofiles=single /mnt/btrfs-progs-tests
   btrfs balance start -f -mconvert=single /mnt/btrfs-progs-tests
   btrfs balance start -f -mdevid=1 /mnt/btrfs-progs-tests
   btrfs balance start -f -sprofiles=single /mnt/btrfs-progs-tests
   btrfs balance start -f -sconvert=single /mnt/btrfs-progs-tests
   btrfs balance start -f -sdevid=1 /mnt/btrfs-progs-tests
   btrfs device add -f /dev/sda10 /mnt/btrfs-progs-tests
   btrfs device del /dev/sda10 /mnt/btrfs-progs-tests
   btrfs device stats /dev/sda6
   btrfs device stats -z /dev/sda6
   btrfs device stats /mnt/btrfs-progs-tests
   btrfs device stats -z /mnt/btrfs-progs-tests
   btrfs device usage /mnt/btrfs-progs-tests
   btrfs scrub status /mnt/btrfs-progs-tests
   btrfs scrub start -B /mnt/btrfs-progs-tests
   btrfs scrub start -B -d /mnt/btrfs-progs-tests
   btrfs scrub start -B -r /mnt/btrfs-progs-tests
   btrfs scrub status /mnt/btrfs-progs-tests
   btrfs scrub start /mnt/btrfs-progs-tests
   btrfs scrub status /mnt/btrfs-progs-tests
   btrfs scrub status /mnt/btrfs-progs-tests
   btrfs scrub status -d /mnt/btrfs-progs-tests
   btrfs scrub status -R /mnt/btrfs-progs-tests
   btrfs scrub status /mnt/btrfs-progs-tests
   btrfs scrub start /dev/sda6
   btrfs scrub status /dev/sda6
   btrfs scrub status /dev/sda6
   btrfs scrub status -d /dev/sda6
   btrfs scrub status -R /dev/sda6
   btrfs scrub status /dev/sda6
   btrfs subvolume snapshot -r /mnt/btrfs-progs-tests /mnt/btrfs-progs-tests/snap1
   btrfs send -f /tmp/btrfs_snapshot_test /mnt/btrfs-progs-tests/snap1
   btrfs send -e -f /tmp/btrfs_snapshot_test /mnt/btrfs-progs-tests/snap1
   btrfs send --no-data -f /tmp/btrfs_snapshot_test /mnt/btrfs-progs-tests/snap1
   btrfs quota enable /mnt/btrfs-progs-tests
   btrfs quota rescan /mnt/btrfs-progs-tests
   btrfs quota rescan -s /mnt/btrfs-progs-tests
   btrfs quota rescan -w /mnt/btrfs-progs-tests
   btrfs quota disable /mnt/btrfs-progs-tests
   btrfs quota enable /mnt/btrfs-progs-tests
   btrfs qgroup create 1/5 /mnt/btrfs-progs-tests
   btrfs qgroup create 2/5 /mnt/btrfs-progs-tests
   btrfs qgroup assign 1/5 2/5 /mnt/btrfs-progs-tests
   btrfs qgroup limit 1G 1/5 /mnt/btrfs-progs-tests
   btrfs qgroup show /mnt/btrfs-progs-tests
   btrfs qgroup show -p -c -r -e -F -f /mnt/btrfs-progs-tests
   btrfs qgroup remove 1/5 2/5 /mnt/btrfs-progs-tests
   btrfs qgroup destroy 2/5 /mnt/btrfs-progs-tests
   btrfs qgroup destroy 1/5 /mnt/btrfs-progs-tests
   btrfs quota disable /mnt/btrfs-progs-tests
   btrfs replace start -f -B /dev/sda6 /dev/sda10 /mnt/btrfs-progs-tests
   btrfs replace status /mnt/btrfs-progs-tests
   btrfs replace start -f -B /dev/sda10 /dev/sda6 /mnt/btrfs-progs-tests
   btrfs-convert /dev/sda6
   btrfs-convert -r /dev/sda6
   btrfs-convert -d /dev/sda6
   btrfs-convert -i /dev/sda6
   btrfs-convert -n /dev/sda6
   btrfs-convert -N 4096 /dev/sda6
   btrfs-convert -l test /dev/sda6
   btrfs-convert -L /dev/sda6
   btrfs-convert --no-progress /dev/sda6
   mkfs.btrfs -f /dev/sda6
   btrfs-image /dev/sda6 /tmp/btrfs_image.img
   btrfs-image -r /tmp/btrfs_image.img /dev/sda6
   mkfs.btrfs -f /dev/sda6
   btrfs-image -c 0 /dev/sda6 /tmp/btrfs_image.img
   btrfs-image -r /tmp/btrfs_image.img /dev/sda6
   mkfs.btrfs -f /dev/sda6
   btrfs-image -c 9 /dev/sda6 /tmp/btrfs_image.img
   btrfs-image -r /tmp/btrfs_image.img /dev/sda6
   mkfs.btrfs -f /dev/sda6
   btrfs-image -t 0 /dev/sda6 /tmp/btrfs_image.img
   btrfs-image -r /tmp/btrfs_image.img /dev/sda6
   mkfs.btrfs -f /dev/sda6
   btrfs-image -t 1 /dev/sda6 /tmp/btrfs_image.img
   btrfs-image -r /tmp/btrfs_image.img /dev/sda6
   mkfs.btrfs -f /dev/sda6
   btrfs-image -t 32 /dev/sda6 /tmp/btrfs_image.img
   btrfs-image -r /tmp/btrfs_image.img /dev/sda6
   mkfs.btrfs -f /dev/sda6
   btrfs-image -w /dev/sda6 /tmp/btrfs_image.img
   btrfs-image -r /tmp/btrfs_image.img /dev/sda6
   mkfs.btrfs -f /dev/sda6
   btrfs-image -w /dev/sda6 /tmp/btrfs_image.img
   btrfs-image -r /tmp/btrfs_image.img /dev/sda6
   btrfs-image -r -t 0 /tmp/btrfs_image.img /dev/sda6
   btrfs-image -r -t 1 /tmp/btrfs_image.img /dev/sda6
   btrfs-image -r -t 32 /tmp/btrfs_image.img /dev/sda6
   btrfs-image -r -o /tmp/btrfs_image.img /dev/sda6
3: Manual check relation source by:
   grep DUP *.c
   Confirmed that all source are modified.
4: Use this raid type manually, do some operations in fs,
   no error found in command and dmesg.
5: Combination of dup conversion with fsck
   Confirmed OK with relative kernel patch titled:
   [PATCH] btrfs: Support convert to -d dup for btrfs-convert

   export TEST_DEV='/dev/vdc'
   export TEST_DIR='/var/ltf/tester/mnt'

   do_dup_test()
   {
       local m_from="$1"
       local d_from="$2"
       local m_to="$3"
       local d_to="$4"

       echo "Convert from -m $m_from -d $d_from to -m $m_to -d $d_to"

       umount "$TEST_DIR" &>/dev/null
       ./mkfs.btrfs -f -m "$m_from" -d "$d_from" "$TEST_DEV" >/dev/null || return 1
       mount "$TEST_DEV" "$TEST_DIR" || return 1

       cp -a /sbin/* "$TEST_DIR"

       [[ "$m_from" != "$m_to" ]] && {
           ./btrfs balance start -f -mconvert="$m_to" "$TEST_DIR" || return 1
       }

       [[ "$d_from" != "$d_to" ]] && {
       local opt=()
           [[ "$d_to" == single ]] && opt+=("-f")
           ./btrfs balance start "${opt[@]}" -dconvert="$d_to" "$TEST_DIR" || return 1
       }

       umount "$TEST_DIR" || return 1
       ./btrfsck "$TEST_DEV" || return 1
       echo

       return 0
   }

   test_all()
   {
       for m_from in single dup; do
       for d_from in single dup; do
       for m_to in single dup; do
       for d_to in single dup; do
       do_dup_test "$m_from" "$d_from" "$m_to" "$d_to" || return 1
       done
       done
       done
       done
   }

   test_all

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Tested-by: Austin S. Hemmelgarn <ahferroin7@gmail.com>
[ minor updates in the changelog ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: convert: add context and operations struct to allow different file systems
Jeff Mahoney [Wed, 14 Aug 2013 18:39:21 +0000 (14:39 -0400)]
btrfs-progs: convert: add context and operations struct to allow different file systems

Push out common convert operations into function pointers that we can
ultimately allow other file systems to use to provide their own
conversions.

Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
8 years agobtrfs-progs: tests: add 019-non-skinny-false-alert
David Sterba [Thu, 26 Nov 2015 14:20:05 +0000 (15:20 +0100)]
btrfs-progs: tests: add 019-non-skinny-false-alert

Catch a buggy condition fixed by "btrfs-progs: fsck: Fix a false alert
where extent record has wrong metadata flag"

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: fsck: Fix a false alert where extent record has wrong metadata flag
Qu Wenruo [Wed, 25 Nov 2015 06:19:06 +0000 (14:19 +0800)]
btrfs-progs: fsck: Fix a false alert where extent record has wrong metadata flag

In process_extent_item(), it gives 'metadata' initial value 0, but for
non-skinny-metadata case, metadata extent can't be judged just from key
type and it forgot that case.

This causes a lot of false alert in non-skinny-metadata filesystem.

Fix it by set correct metadata value before calling add_extent_rec().

Reported-by: Christoph Anton Mitterer <calestyo@scientia.net>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: update raid table in mkfs manpage
David Sterba [Tue, 24 Nov 2015 17:14:39 +0000 (18:14 +0100)]
btrfs-progs: docs: update raid table in mkfs manpage

* split copies to copies and parity and add a common header for all the
  raid options
* add missing RAID1
* n/a were dropped

Based on feedback from Duncan <1i5t5.duncan@cox.net>.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: chunk-recovery: Fix a float point error
Qu Wenruo [Thu, 19 Nov 2015 06:42:39 +0000 (14:42 +0800)]
btrfs-progs: chunk-recovery: Fix a float point error

Fix a zero division causing chunk-recovery fail.

Also fix a typo "strpie_length" -> "stripe_length".

Reported-by: Scotty Edmonds <scotty@scottyedmonds.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: add 007-mix-nodesize-sectorsize
David Sterba [Thu, 19 Nov 2015 15:37:15 +0000 (16:37 +0100)]
btrfs-progs: tests: add 007-mix-nodesize-sectorsize

Test combinations of sectorsize and nodesize on a single device.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: mkfs: use correct size for superblock csum and writeout
David Sterba [Thu, 19 Nov 2015 14:55:05 +0000 (15:55 +0100)]
btrfs-progs: mkfs: use correct size for superblock csum and writeout

If sectorsize is not BTRFS_SUPER_INFO_SIZE (4k), the superblock checksum
is wrong and mkfs fails. This has been reported on ppc64 where we pick
sectorisize from page size (64k).  This has been broken since ages
(2008) and discovered by the recently added superblock checks.

Reported-by: Dinar Valeev <dvaleev@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: docs: fix typo in balance man page
Jeffrey Schiller [Tue, 17 Nov 2015 05:01:46 +0000 (00:01 -0500)]
btrfs-progs: docs: fix typo in balance man page

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agoBtrfs progs v4.3.1
David Sterba [Mon, 16 Nov 2015 13:29:25 +0000 (14:29 +0100)]
Btrfs progs v4.3.1

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: fix fd leak in resolve_loop_device_with_loopdev
David Sterba [Fri, 13 Nov 2015 18:15:27 +0000 (19:15 +0100)]
btrfs-progs: fix fd leak in resolve_loop_device_with_loopdev

Resolves-coverity-id: 1339300
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: add 006-partitioned-loopdev
David Sterba [Fri, 13 Nov 2015 17:43:04 +0000 (18:43 +0100)]
btrfs-progs: tests: add 006-partitioned-loopdev

Create filesystem on a partitioned loop device, test for "btrfs-progs:
Fix partitioned loop devices resolving".

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: utils: do not expect lo_file_name to be null terminated
David Sterba [Fri, 13 Nov 2015 17:20:38 +0000 (18:20 +0100)]
btrfs-progs: utils: do not expect lo_file_name to be null terminated

The loop_info64::lo_file_name might not be null terminated. Avoid strlen
and trim the length to whatever size of the loop_info buffer.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Fix partitioned loop devices resolving
Florian Margaine [Fri, 13 Nov 2015 17:09:05 +0000 (18:09 +0100)]
btrfs-progs: Fix partitioned loop devices resolving

When using partitions on a loop device, the device's name can be
e.g. /dev/loop0p1 or similar, and no relevant entry exists in the /sys
filesystem, so the current resolve_loop_device function fails.

Instead of using string functions to extract the device name and reading
this file, this patch uses the loop device API through ioctl to get the
correct backing file.

Signed-off-by: Florian Margaine <florian@platform.sh>
[ changed checks of error values from open and ioctl ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: find-root: Add support to search chunk root
Qu Wenruo [Fri, 13 Nov 2015 02:53:41 +0000 (10:53 +0800)]
btrfs-progs: find-root: Add support to search chunk root

Add support to search chunk root, as we only need to search tree roots
in system chunk, which should be very easy to add, just iterate in
system chunks.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[ renamed to btrfs_next_bg_* ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: cleanup cmd_device_usage, simplify loop checks
Zhao Lei [Mon, 9 Nov 2015 09:59:38 +0000 (17:59 +0800)]
btrfs-progs: cleanup cmd_device_usage, simplify loop checks

1: Remove more_than_one variable, use iterators value instead
2: Remove "out" label, we use break instead.

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
[ changelog update ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: device: use warning/error for error message
Zhao Lei [Mon, 9 Nov 2015 09:59:37 +0000 (17:59 +0800)]
btrfs-progs: device: use warning/error for error message

Switch to common warning()/error() for cmds-device.c.

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
[ minor tweaks ]
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Check periodic.timer_fd's value before use
Zhao Lei [Mon, 9 Nov 2015 09:06:53 +0000 (17:06 +0800)]
btrfs-progs: Check periodic.timer_fd's value before use

periodic.timer_fd's value is 0 on inititlize-failed case,
if no value-checking before read(), the code will run as
read(STDIN).

This patch fixed above case.

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: add 005-long-device-name-for-ssd
David Sterba [Fri, 13 Nov 2015 16:29:11 +0000 (17:29 +0100)]
btrfs-progs: tests: add 005-long-device-name-for-ssd

A test for "btrfs-progs: mkfs: increase buffer size in is_ssd". Create a
device with a long name through loop device wrapped to a device mapper
linear device, switch it to the "ssd" mode status.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: mkfs: increase buffer size in is_ssd
Michael Lass [Sun, 8 Nov 2015 15:33:03 +0000 (16:33 +0100)]
btrfs-progs: mkfs: increase buffer size in is_ssd

In current versions of util-linux the buffer passed to blkid_devno_to_wholedisk
has to be sufficiently large to not only hold the device name but the complete
target of the /sys/dev/block/<maj:min> symlink. This was changed only recently
in 4419ffb9eff5801fdbd385a4a6199b3877f802ad.

The small buffer size currently can lead to failure of is_ssd due to truncated
device names:

readlink("/sys/dev/block/254:7", "../../devices/virtual/block/dm-", 31) = 31
open("/sys/block/dm-/queue/rotational", O_RDONLY) = -1 ENOENT (No such file or directory)

Signed-off-by: Michael Lass <bevan@bi-co.net>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: tests: add 011-delete-missing-device
David Sterba [Fri, 13 Nov 2015 15:49:55 +0000 (16:49 +0100)]
btrfs-progs: tests: add 011-delete-missing-device

Test for "btrfs-progs: allow device deletion using 'missing' keyword
again".

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: allow device deletion using 'missing' keyword again
Alexander Fougner [Sat, 7 Nov 2015 00:26:24 +0000 (01:26 +0100)]
btrfs-progs: allow device deletion using 'missing' keyword again

Device deletion procedures ensures the device is a block device.
This patch introduces 'missing' as keyword again, correctly
passing it on to the kernel instead of complaining about
'missing' not being a block device.

Signed-off-by: Alexander Fougner <fougner89@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: mkfs: handler memory allocation error in make_btrfs
David Sterba [Fri, 6 Nov 2015 17:52:09 +0000 (18:52 +0100)]
btrfs-progs: mkfs: handler memory allocation error in make_btrfs

Do the allocation early, no need to cleanup.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: image: reorder initialization in metadump_init
David Sterba [Fri, 6 Nov 2015 17:31:27 +0000 (18:31 +0100)]
btrfs-progs: image: reorder initialization in metadump_init

Put the allocations first, move pthread cond and mutex last so we don't
have to do cleanup.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: corrupt-block: use on-stack path buffer in corrupt_dir_item
David Sterba [Fri, 6 Nov 2015 17:27:55 +0000 (18:27 +0100)]
btrfs-progs: corrupt-block: use on-stack path buffer in corrupt_dir_item

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: calc-size: kill fs_roots structure and its user
David Sterba [Fri, 6 Nov 2015 17:25:04 +0000 (18:25 +0100)]
btrfs-progs: calc-size: kill fs_roots structure and its user

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: calc-size: use on-stack buffer for a helper structure
David Sterba [Fri, 6 Nov 2015 17:20:52 +0000 (18:20 +0100)]
btrfs-progs: calc-size: use on-stack buffer for a helper structure

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: drop unused argument from zero_output_file
David Sterba [Fri, 6 Nov 2015 17:12:44 +0000 (18:12 +0100)]
btrfs-progs: drop unused argument from zero_output_file

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: use fixed size buffer in zero_output_file
David Sterba [Fri, 6 Nov 2015 17:10:29 +0000 (18:10 +0100)]
btrfs-progs: use fixed size buffer in zero_output_file

Rewrite the loop so we don't need to allocate sectorsize and write in 4k
steps instead. We know that sectorsize is divisible by 4096.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: handle memory allocation failures in traverse_directory
David Sterba [Fri, 6 Nov 2015 17:03:24 +0000 (18:03 +0100)]
btrfs-progs: handle memory allocation failures in traverse_directory

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: handle memory allocation failure in add_file_items
David Sterba [Fri, 6 Nov 2015 16:57:41 +0000 (17:57 +0100)]
btrfs-progs: handle memory allocation failure in add_file_items

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: use on-stack buffer in __csum_tree_block_size
David Sterba [Fri, 6 Nov 2015 16:50:48 +0000 (17:50 +0100)]
btrfs-progs: use on-stack buffer in __csum_tree_block_size

We know the maximum size of a checksum, calling malloc for 4 bytes is
weird.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: use on-stack buffer in __ino_to_path_fd
David Sterba [Fri, 6 Nov 2015 16:47:17 +0000 (17:47 +0100)]
btrfs-progs: use on-stack buffer in __ino_to_path_fd

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: use on-stack buffer in recover_prepare
David Sterba [Fri, 6 Nov 2015 16:38:31 +0000 (17:38 +0100)]
btrfs-progs: use on-stack buffer in recover_prepare

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: use on-stack variable in __rebuild_device_items
David Sterba [Fri, 6 Nov 2015 16:34:45 +0000 (17:34 +0100)]
btrfs-progs: use on-stack variable in __rebuild_device_items

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: use on-stack buffer for btrfs_scan_one_device
David Sterba [Fri, 6 Nov 2015 16:09:17 +0000 (17:09 +0100)]
btrfs-progs: use on-stack buffer for btrfs_scan_one_device

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: use on-stack buffer for dev_to_fsid
David Sterba [Fri, 6 Nov 2015 16:09:17 +0000 (17:09 +0100)]
btrfs-progs: use on-stack buffer for dev_to_fsid

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: remove unused parameter from print_one_fs
David Sterba [Fri, 6 Nov 2015 16:05:36 +0000 (17:05 +0100)]
btrfs-progs: remove unused parameter from print_one_fs

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: don't print version info from embedded subcommands
David Sterba [Fri, 6 Nov 2015 16:00:36 +0000 (17:00 +0100)]
btrfs-progs: don't print version info from embedded subcommands

The version is provided by 'btrfs --version'.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: simplify empty stirngs check
David Sterba [Fri, 6 Nov 2015 15:53:55 +0000 (16:53 +0100)]
btrfs-progs: simplify empty stirngs check

We can do a strlen(str) == 0 in a simpler way.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: utils: rename helpinfo unit vairables
David Sterba [Fri, 6 Nov 2015 15:39:49 +0000 (16:39 +0100)]
btrfs-progs: utils: rename helpinfo unit vairables

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agoBtrfs progs v4.3
David Sterba [Fri, 6 Nov 2015 11:07:54 +0000 (12:07 +0100)]
Btrfs progs v4.3

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: mkfs: do not truncate the image when --rootdir is set
David Sterba [Thu, 5 Nov 2015 13:22:10 +0000 (14:22 +0100)]
btrfs-progs: mkfs: do not truncate the image when --rootdir is set

With the rootdir option we try to guess the final size of the image and
fill it with zeros, preceded by truncation. After patch

"Btrfs-progs: Do not force mixed block group creation unless '-M' option
 is specified"

the misc test 002 will fail, because of the non-mixed mode. I think we
should not touch the image size (no change for block devices) and try to
fit into whatever is provided by user.

Signed-off-by: David Sterba <dsterba@suse.com>
8 years agobtrfs-progs: Allow btrfs_leaf_free_space to accept NULL root
Qu Wenruo [Thu, 5 Nov 2015 08:32:59 +0000 (16:32 +0800)]
btrfs-progs: Allow btrfs_leaf_free_space to accept NULL root

Btrfs_leaf_free_space() function is used to determine the leaf/node
size.
It's OK to use root->nodesize to determine nodesize, but in fact,
extent_buffer->len can also be used to determine the nodesize if caller
can ensure it's a tree block.

So this patch will add support for NULL root for btrfs_leaf_free_space()
function, to allow btrfs_print_leaf() functions to be called in gdb or
to debug temporary btrfs in make_btrfs() without a valid root.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: mkfs: Round device size down to sectorsize
Zhao Lei [Tue, 3 Nov 2015 03:23:36 +0000 (11:23 +0800)]
btrfs-progs: mkfs: Round device size down to sectorsize

When do following command in a vm, whose disks are created by
qemu-img create -f raw 11 2.6G:
 # mkfs.btrfs -f /dev/vdd /dev/vde /dev/vdf
 # btrfs-show-super /dev/vdd /dev/vde /dev/vdf | grep dev_item.total_bytes
 dev_item.total_bytes    2791727104
 dev_item.total_bytes    2791729152
 dev_item.total_bytes    2791729152
We can see that the first device's size is little smaller.

And it fails xfstests btrfs/011.

Reason:
 First device's size is rounded down to sectorsize in make_btrfs(),
 but other devices are not.

Fix:
 Round down remain devices' size in btrfs_add_to_fsid().

Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: Rename variables in btrfs_add_to_fsid
Zhao Lei [Tue, 3 Nov 2015 03:23:37 +0000 (11:23 +0800)]
btrfs-progs: Rename variables in btrfs_add_to_fsid

There are two total_bytes in btrfs_add_to_fsid(), local variable
of total_bytes means fs_total_bytes, and device->total_bytes means
device's total_bytes.
And device's total_bytes in argument is named block_count in current
code.

This patch rename:
 total_bytes -> fs_total_bytes
 block_count -> device_total_bytes

To make code more readable.

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: print-tree: Output stripe dev uuid
Qu Wenruo [Tue, 3 Nov 2015 06:48:09 +0000 (14:48 +0800)]
btrfs-progs: print-tree: Output stripe dev uuid

Add output for dev uuid for print_chunk().

Quite useful to debug temporary btrfs in btrfs-convert.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: mkfs: remove unused code of format uuid string
Zhao Lei [Tue, 3 Nov 2015 11:03:01 +0000 (19:03 +0800)]
btrfs-progs: mkfs: remove unused code of format uuid string

Variant named dev_uuid and uuid_unparse() for set its value are
not used, remove it.

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: mkfs: output device list in sorted order
Zhao Lei [Tue, 3 Nov 2015 11:03:00 +0000 (19:03 +0800)]
btrfs-progs: mkfs: output device list in sorted order

list_for_each_entry_reverse() in current code can not output
devices in sorted order, because the sequence are broken in
btrfs_alloc_chunk().

We can use list_sort() instead.

Before patch:
 # mkfs.btrfs -f /dev/vdd /dev/vde /dev/vdf
 ...
 Number of devices:  3
 Devices:
    ID        SIZE  PATH
     3     2.60GiB  /dev/vdf
     1     2.60GiB  /dev/vdd
     2     2.60GiB  /dev/vde

After patch:
 # mkfs.btrfs -f /dev/vdd /dev/vde /dev/vdf
 ...
 Number of devices:  3
 Devices:
    ID        SIZE  PATH
     1     2.60GiB  /dev/vdd
     2     2.60GiB  /dev/vde
     3     2.60GiB  /dev/vdf

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: fi usage: print header and footer separation line
David Sterba [Tue, 3 Nov 2015 00:09:50 +0000 (01:09 +0100)]
btrfs-progs: fi usage: print header and footer separation line

Print the full row width and change to '-' as we're not using '='
anywhere.

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: string table: add specifier for fillup char
David Sterba [Tue, 3 Nov 2015 00:00:23 +0000 (01:00 +0100)]
btrfs-progs: string table: add specifier for fillup char

Currently it's one by a single "=", but we might want to use
a different filler, let's make it explicit by "*".

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: string table: cleanup, rename single letter variables
David Sterba [Mon, 2 Nov 2015 23:51:29 +0000 (00:51 +0100)]
btrfs-progs: string table: cleanup, rename single letter variables

Exception for indexing variables.

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: string table: fix whitespace damage
David Sterba [Mon, 2 Nov 2015 23:46:52 +0000 (00:46 +0100)]
btrfs-progs: string table: fix whitespace damage

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: fi usage: print device id column in the tabular output<F2>
David Sterba [Mon, 2 Nov 2015 23:19:41 +0000 (00:19 +0100)]
btrfs-progs: fi usage: print device id column in the tabular output<F2>

Example output:

             Data     Metadata System
Id Path      single   RAID1    RAID1    Unallocated

 1 /dev/sdc2 44.94GiB  7.93GiB 32.00MiB     1.00GiB
 2 /dev/sde1 44.94GiB  7.93GiB 32.00MiB     1.00GiB
             ======== ======== ======== ===========
   Total     89.88GiB  7.93GiB 32.00MiB     2.00GiB
   Used      74.28GiB  4.44GiB 20.00KiB

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: fi usage: cleanup, replace space info starting column constant
David Sterba [Mon, 2 Nov 2015 23:08:32 +0000 (00:08 +0100)]
btrfs-progs: fi usage: cleanup, replace space info starting column constant

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: fi usage: cleanup, replace header constant
David Sterba [Mon, 2 Nov 2015 17:54:48 +0000 (18:54 +0100)]
btrfs-progs: fi usage: cleanup, replace header constant

The size of the header is not obvious, let's make it more visible by
replacing it with a varaible.

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: fi usage: properly count real space infos
David Sterba [Mon, 2 Nov 2015 17:44:45 +0000 (18:44 +0100)]
btrfs-progs: fi usage: properly count real space infos

We did not account the column for path but abused the skipped global
block reserve colum instead. Properly count the real infos and manually
added headers.

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: fi usage: print path header in the tabular mode
David Sterba [Mon, 2 Nov 2015 17:40:57 +0000 (18:40 +0100)]
btrfs-progs: fi usage: print path header in the tabular mode

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: fi usage: cleanup, print header in one go
David Sterba [Mon, 2 Nov 2015 17:32:04 +0000 (18:32 +0100)]
btrfs-progs: fi usage: cleanup, print header in one go

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: fi usage: do not print global block reserve
David Sterba [Mon, 2 Nov 2015 17:24:28 +0000 (18:24 +0100)]
btrfs-progs: fi usage: do not print global block reserve

Global block reserve is inherently part of metadata and should not be
listed separately in the output of 'fi usage' in the tabular output.

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agoBtrfs progs v4.3-rc1
David Sterba [Mon, 2 Nov 2015 14:46:50 +0000 (15:46 +0100)]
Btrfs progs v4.3-rc1

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: show-super: Add option to print superblock at given bytenr
Qu Wenruo [Mon, 2 Nov 2015 08:34:19 +0000 (16:34 +0800)]
btrfs-progs: show-super: Add option to print superblock at given bytenr

Add '-s <sb_bytenr>' option to show superblock at given bytenr.

This is very useful to debug non-standard btrfs, like debuging the
1st stage btrfs of btrfs-convert.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[ minor updates in docs ]
Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: docs: enhance manual page for inspect-internal
David Sterba [Mon, 2 Nov 2015 13:41:32 +0000 (14:41 +0100)]
btrfs-progs: docs: enhance manual page for inspect-internal

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: docs: enhance the manual page for convert
David Sterba [Mon, 2 Nov 2015 13:19:58 +0000 (14:19 +0100)]
btrfs-progs: docs: enhance the manual page for convert

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: docs: enhance manual page for balance
David Sterba [Mon, 2 Nov 2015 12:44:05 +0000 (13:44 +0100)]
btrfs-progs: docs: enhance manual page for balance

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: docs: enhance manual page for btrfstune
David Sterba [Mon, 2 Nov 2015 10:08:57 +0000 (11:08 +0100)]
btrfs-progs: docs: enhance manual page for btrfstune

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: docs: enhance manual page for mkfs
David Sterba [Fri, 30 Oct 2015 18:16:41 +0000 (19:16 +0100)]
btrfs-progs: docs: enhance manual page for mkfs

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: mkfs: print version info first
David Sterba [Fri, 30 Oct 2015 15:58:52 +0000 (16:58 +0100)]
btrfs-progs: mkfs: print version info first

The version info should not be preceded by any messages.

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: image: fix bogus check after cpu on-line detection
David Sterba [Fri, 30 Oct 2015 14:34:55 +0000 (15:34 +0100)]
btrfs-progs: image: fix bogus check after cpu on-line detection

Comparing unsigned type for <= 0 does not make much sense, we should
really check the signed value returned by sysconf.

Resolves-coverity-id: 1324536
Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: Avoid use pointer in handle_options
Zhao Lei [Thu, 29 Oct 2015 09:31:48 +0000 (17:31 +0800)]
btrfs-progs: Avoid use pointer in handle_options

We use pointer of argc and argv in handle_options() because they
are necessary in very old code which are not exist now.

This patch move to use argc and argv directly in handle_options(),
alone with following update:
1: rename handle_options() to check_options()
   to fit its function.
2: cleanup for condition in handle_options() to make line short.

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: Fix uninitialized key.type for btrfs_find_free_objectid
Zhao Lei [Thu, 29 Oct 2015 09:31:46 +0000 (17:31 +0800)]
btrfs-progs: Fix uninitialized key.type for btrfs_find_free_objectid

To avoid using uninitialized value in btrfs_search_slot().

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: Fix negative eb's ref_cnt in btrfs-calc-size
Zhao Lei [Thu, 29 Oct 2015 09:31:44 +0000 (17:31 +0800)]
btrfs-progs: Fix negative eb's ref_cnt in btrfs-calc-size

btrfs-calc-size show following warning:
 # btrfs-calc-size /dev/sda6
 Calculating size of root tree
 ...
 extent_io.c:582: free_extent_buffer: Assertion `eb->refs < 0` failed.
 ./btrfs-calc-size[0x41d642]
 ./btrfs-calc-size(free_extent_buffer+0x70)[0x41e1c1]
 ./btrfs-calc-size(btrfs_free_fs_root+0x11)[0x40e1e8]
 ./btrfs-calc-size[0x40e215]
 ./btrfs-calc-size(rb_free_nodes+0x1d)[0x4326fe]
 ./btrfs-calc-size(close_ctree+0x3f3)[0x40f9ea]
 ./btrfs-calc-size(main+0x200)[0x431b4e]
 /lib64/libc.so.6(__libc_start_main+0xf5)[0x3858621d65]
 ./btrfs-calc-size[0x407009]

Reason:
 path in calc_root_size() is only used to save node data,
 it don't hold ref_cnt for each eb in.
 Using btrfs_free_path() to free path will reduce these eb
 again, and cause many problems, as negative ref_cnt or
 invalid memory access.

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: fix floating point exception for btrfs-calc-size
Zhao Lei [Thu, 29 Oct 2015 09:31:43 +0000 (17:31 +0800)]
btrfs-progs: fix floating point exception for btrfs-calc-size

Current code exit with floating point exception on a blank fs:
 # btrfs-calc-size -b /dev/sda6
 Calculating size of root tree
         Total size: 16384
                 Inline data: 0
         Total seeks: 0
                 Forward seeks: 0
                 Backward seeks: 0
 Floating point exception

This patch add a condition check for above case.

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: add initial tests/README
David Sterba [Tue, 27 Oct 2015 16:17:47 +0000 (17:17 +0100)]
btrfs-progs: add initial tests/README

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: add an initial README
David Sterba [Tue, 27 Oct 2015 15:30:22 +0000 (16:30 +0100)]
btrfs-progs: add an initial README

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: mkfs: remove stray message about forced mixed-bg
David Sterba [Mon, 26 Oct 2015 19:02:22 +0000 (20:02 +0100)]
btrfs-progs: mkfs: remove stray message about forced mixed-bg

We no longer force mixed-bg mode since "Btrfs-progs: Do not force mixed
block group creation unless '-M' option is specified", the message is
not relevant anymore.

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: tests: add 003-mixed-with-wrong-nodesize
David Sterba [Mon, 26 Oct 2015 18:54:57 +0000 (19:54 +0100)]
btrfs-progs: tests: add 003-mixed-with-wrong-nodesize

Mixed mode needs equal sectorsize and nodesize. This was fixed by

 "Btrfs-progs: Prevent creation of filesystem with 'mixed bgs' and
  having differing sectorsize and nodesize"

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agobtrfs-progs: tests: do not log output of run_mayfail to terminal
David Sterba [Mon, 26 Oct 2015 18:51:10 +0000 (19:51 +0100)]
btrfs-progs: tests: do not log output of run_mayfail to terminal

No need to log expected failures to the terminal, the results file is
fine; pass the return value of the command.

Signed-off-by: David Sterba <dsterba@suse.com>
9 years agoBtrfs-progs: Prevent creation of filesystem with 'mixed bgs' and having differing...
Chandan Rajendra [Wed, 14 Oct 2015 17:40:38 +0000 (23:10 +0530)]
Btrfs-progs: Prevent creation of filesystem with 'mixed bgs' and having differing sectorsize and nodesize.

mkfs.btrfs allows creation of Btrfs filesystem instances with mixed block
group feature enabled and having a sectorsize different from nodesize.
For e.g:

[root@localhost btrfs-progs]# mkfs.btrfs -f -M -s 4096 -n 16384  /dev/loop0
Forcing mixed metadata/data groups
btrfs-progs v3.19-rc2-404-gbbbd18e-dirty
See http://btrfs.wiki.kernel.org for more information.

Performing full device TRIM (4.00GiB) ...
Label:              (null)
UUID:               c82b5720-6d88-4fa1-ac05-d0d4cb797fd5
Node size:          16384
Sector size:        4096
Filesystem size:    4.00GiB
Block group profiles:
  Data+Metadata:    single            8.00MiB
  System:           single            4.00MiB
SSD detected:       no
Incompat features:  mixed-bg, extref, skinny-metadata
Number of devices:  1
Devices:
  ID        SIZE  PATH
   1     4.00GiB  /dev/loop6

This commit fixes the issue by setting BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS
feature bit before checking the validity of nodesize that was specified on the
command line.

Signed-off-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Signed-off-by: David Sterba <dsterba@suse.com>