The 'Current fsid:' value does not match the real fsid.
Reported-by: Mike Fleetwood <mike.fleetwood@googlemail.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
[BUG]
The original map-logical has the following problems:
1) Assert if we pass any tree root bytenr.
The problem is easy to trigger, here the number 29622272 is the bytenr of tree root:
# btrfs-map-logical -l 29622272 /dev/sda6
mirror 1 logical 29622272 physical 38010880 device /dev/sda6
mirror 2 logical 29622272 physical 1111752704 device /dev/sda6
extent_io.c:582: free_extent_buffer: Assertion `eb->refs < 0` failed.
btrfs-map-logical[0x41c464]
btrfs-map-logical(free_extent_buffer+0xc0)[0x41cf10]
btrfs-map-logical(btrfs_release_all_roots+0x59)[0x40e649]
btrfs-map-logical(close_ctree+0x1aa)[0x40f51a]
btrfs-map-logical(main+0x387)[0x4077c7]
/usr/lib/libc.so.6(__libc_start_main+0xf0)[0x7f80a5562790]
btrfs-map-logical(_start+0x29)[0x4078f9]
The problem is that, btrfs-map-logical always use sectorsize as default
block size to call alloc_extent_buffer.
And when it failes to find the block with the same size, it will free
the extent buffer in a incorrect method(Free and create a new one with
refs == 1).
2) Will return map result for non-exist extent.
# btrfs-map-logical -l 1 -b 123456 /dev/sda6
mirror 1 logical 1 physical 1 device /dev/sda6
mirror 1 logical 4097 physical 4097 device /dev/sda6
mirror 1 logical 8193 physical 8193 device /dev/sda6
...
Normally, before bytenr 12582912, there should be no extent as that's
the mkfs time temp metadata/system chunk.
But map-logical will still map them out.
Not to mention the 1 offset among all results.
[FIX]
This patch will rework the whole map logical by the following methods:
1) Always do things inside a extent
Even under the following case, map logical will only return covered
range in existing extents.
|<------ range given ------->|
|<-Extent A->| |<-Extent B->| |<---Extent C->|
Result:
|<-->| |<---------->| |<-->|
So with this patch, we will search extent tree to ensure all operation
are inside a extent before we do some stupid things.
2) No direct call on alloc_extent_buffer function.
That low-level function shouldn't be called at such high level.
It's only designed for low-level tree operation.
So in this patch we will only use safe high level functions avoid such
problem.
[RESULT]
With this patch, no assert will be triggered and better handle on
non-exist extents.
# btrfs-map-logical -l 29622272 /dev/sda6
mirror 1 logical 29622272 physical 38010880 device /dev/sda6
mirror 2 logical 29622272 physical 1111752704 device /dev/sda6
# btrfs-map-logical -l 1 -b 123456 /dev/sda6
No extent found at range [1,123457)
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
This function will write extent content info desired file.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
The new function will print the mapping info of given range
[logical, logical+len).
Note, caller must ensure the ranges are completely inside an extent.
Or btrfs_map_block can return -ENOENT.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Introduce the function to get accurate extent length based on extent
tree search.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
The embedded 'path' is a pointera and we can't make it a path buffer due
to API constraints. Use a separate buffer and sto using the unsafe
path_cat interface.
Signed-off-by: David Sterba <dsterba@suse.cz>
Add versions of path_cat and path_cat3 that do not allocate the memory.
The unhandled memory allocations are still there.
Signed-off-by: David Sterba <dsterba@suse.cz>
Get rid of the allocation. The logic is changed so that a NULL
cur_subvol::path means that no subvolume/snapshot has been found so far.
Signed-off-by: David Sterba <dsterba@suse.cz>
At some places we do not clear the whole ioctl structure and could
pass garbage to kernel. Zero the ioctl vol_args and use a helper for
copying the path.
Signed-off-by: David Sterba <dsterba@suse.cz>
The path bufferes should be PATH_MAX but BTRFS_PATH_NAME_MAX is shorter
due to embedding in 4k aligned structures.
The only reason to use BTRFS_PATH_NAME_MAX is for the respective
structures btrfs_ioctl_vol_args::name.
Signed-off-by: David Sterba <dsterba@suse.cz>
Kind of a big feature of btrfs is being able to have a default subvol. However
the receive code generates the paths to the subvols from the root of the fs,
even in the case of a default subvol. So instead figure out if we're inside of
a subvol, either because we have a different default or we've chroot'ed and are
using -m. Then strip this extra path off of the subvol we find so we can look
up our parent properly. Thanks
Reported-by: Neil Horman <nhorman@redhat.com>
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Anand reports that the static buffers used for pertty size strings cause
a stack overflow on SPARC. Zach proposed to change the printf format to
wrap the number and the suffix into a macro. This would require to
change all callsites of pretty_size* and is not very convienient to
write.
This patch replaces the per-call-site static buffers with a limited
number for slots that would be used on each invokation of pretty_size
and wrap around. The number of array slots shall be 10 for now, in
current codebase there are no more than 2 calls to pretty_size in a
single argument list.
Reported-by: Anand Jain <Anand.Jain@oracle.com>
CC: Zach Brown <zab@zabbo.net>
Signed-off-by: David Sterba <dsterba@suse.cz>
The error string buffer passed as an argument is of a fixed size, though
we could print up to PATH_MAX + something bytes. Print the error message
directly.
Signed-off-by: David Sterba <dsterba@suse.cz>
The error string buffer passed as an argument is of a fixed size, though
we could print up to PATH_MAX + something bytes. Print the error message
directly.
Signed-off-by: David Sterba <dsterba@suse.cz>
Add includes that let the header files compile or add explicit include
of kerncompat if the uXX types are used.
Signed-off-by: David Sterba <dsterba@suse.cz>
So far the subvol sync command takes a shortcut and looks if there are
any deleted subvols at all. It does not print the deleted subvolumes as
they get cleaned. Arguably this is what the user would like to see and
has to do
$ btrfs subvol sync /path $(btrfs subvol list -d /path | "extract the ids")
to see the progress.
Make it look for all currently deleted subvolumes automatically and
print the progress as if the ids were listed manually.
This is a slight change in the semantics of the command. Previously, any
new subvol deletion would prevent subvol sync to return. To simulate the
old behaviour, run 'subvol sync' in a loop until it returns 0.
Signed-off-by: David Sterba <dsterba@suse.cz>
We should zero the dirstream in case we close it ourselves, so the
caller does not do it again. Most current callers do not do that and
exit immediatelly.
Signed-off-by: David Sterba <dsterba@suse.cz>
Running 'btrfs' without arguments will print complete help that spans
a lot of lines and is really helpful. Print only subcommand group
names with short descriptions, similar to what 'git' does.
Signed-off-by: David Sterba <dsterba@suse.cz>
The variable argv0_buf was duplicated and the changes done in utils.c
were not propagated to help.c. So if an unknown commandline token was
found, the error message did not contain the known part:
$ btrfs scrub test
: unknown token 'test'
instead of
$ btrfs scrub test
btrfs scrub: uknown token 'test'
Signed-off-by: David Sterba <dsterba@suse.cz>
The 1st level balance subcommand is the right one but it accidentally
inherited the notice about 'filesystem balance' deprecation.
Signed-off-by: David Sterba <dsterba@suse.cz>
This patch prints the summary of the filesystem after the creation.
The main fileds printed are:
- devices list with their uuid, devid, path and size
- raid profile (dup,single,raid0...)
- leafsize/nodesize/sectorsize
- filesystem features (raid56, extref, mixed-bg)
- chunk size and type
If the '-v' switched is passed, the output is more verbose; if the '-q'
switched is passed, only the errors are printed.
Below an example:
BTRFS filesystem summary:
Label: btrfs-test
UUID: 14ae8a88-98ac-4f22-8441-79f76ec622f7
Node size: 4096
Leaf size: 4096
Sector size: 4096
Initial chunks:
Data+Metadata: 9.01GiB
System: 18.06MiB
Metadata profile: RAID5
Data profile: RAID5
Mixed mode: YES
SSD detected: NO
Incompat features: mixed-bg, extref, raid56
Number of devices: 10
UUID ID SIZE PATH
------------------------------------ -- --------- -----------
df1c7f50-1980-4da2-8bc9-7ee6ffb0b554 1 50.00GiB /dev/vdb
32c808a0-cd7b-4497-a2c0-1d77a9854af9 2 50.00GiB /dev/vdc
3159782e-d108-40bc-9e15-090ecac160b4 3 50.00GiB /dev/vdd
db7eaf0c-beb8-4093-a9d0-b9c25c146305 4 50.00GiB /dev/vde
c367ca04-1f71-49c0-a331-11fc0b87e9fc 5 50.00GiB /dev/vdf
e9b73c86-4058-4b3a-90ac-18741a276e70 6 50.00GiB /dev/vdg
c4298b7a-ad41-4690-bf10-bf748b319413 7 50.00GiB /dev/vdh
1cf048c8-af8a-4225-b09a-5d12e9b217fa 8 2.00GiB /dev/vdi
7e157869-768a-4725-bad5-82e6bd05fd17 9 2.00GiB /dev/vdj
2c9431ac-c7f0-45a5-8529-cef8cf6e4033 10 2.00GiB /dev/vdk
Total devices size: 356.01GiB
Signed-off-by: Goffredo Baroncelli <kreijack@inwind.it>
Signed-off-by: David Sterba <dsterba@suse.cz>
The function make_btrfs() has as argument the fsid of the filesystem.
If this fsid is empty or null make_btrfs() generates a new fsid. However
If the buffer is valid (but the string is empty) the generated fsid is
copied back to the caller.
Signed-off-by: Goffredo Baroncelli <kreijack@inwind.it>
Reviewed-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
When creating a new btrfs_device, copy the path to track it. This path
is then used by mkfs.btrfs to list all devices.
Signed-off-by: Goffredo Baroncelli <kreijack@inwind.it>
Reviewed-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
A mkfs on a regular file will try to call the DEV_SCAN ioctl that will
take the loop through the kernel and fail:
ERROR: device scan failed './test.img' - Block device required
If a user without permission to open the control device tries to mkfs a
regular file, an error message is printed:
failed to open /dev/btrfs-control skipping device registration: Permission denied
So we should not try to pass a non-block-device besides that this makes
the --quiet option more quiet.
Signed-off-by: David Sterba <dsterba@suse.cz>
David Sterba