- add rule to generated version.h when any relevant stuff changed
- add rule to clean generated files on "make clean-all"
Signed-off-by: Karel Zak <kzak@redhat.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
- define basic default CFLAGS in configure.ac, because:
* autoconf default is -g -O2, but btrfs uses -g -O1
* it's better to follow autoconf; standard way to modify
CFLAGS is to call: CFLAGS="foo bar" ./configure
- move all flags to one place in Makefile.in
- don't use AM_CFLAGS, the CFLAGS and STATIC_CFLAGS are enough
- don't mix objects and flags in $LIBS, it's more readable to
add $(libs) to make rules
Signed-off-by: Karel Zak <kzak@redhat.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
The original homemade solution is unnecessary, autotools provides better
infrastructure to generate files.
Signed-off-by: Karel Zak <kzak@redhat.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
- use standard PACKAGE_{NAME,VERSION,STRING,URL,...} autoconf macros
rather than homemade BTRFS_BUILD_VERSION
- don't #include version.h, now the file is necessary for library API only
Note that "btrfs version" returns "btrfs-progs <version>" instead of
the original confusing "btrfs <version>".
Signed-off-by: Karel Zak <kzak@redhat.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
- the header file is generated by ./configure, the standard autotools
way is to use -include config.h on compiler command line rather than
include the file directly from code
- remove _GNU_SOURCE from code, the macros is already defined in config.h
by AC_USE_SYSTEM_EXTENSIONS autoconf macro
Signed-off-by: Karel Zak <kzak@redhat.com>
[_GNU_SOURCE changes already done]
Signed-off-by: David Sterba <dsterba@suse.cz>
- add ./autogen.sh script (necessary after git clean/clone)
- add ./configure.ac
- copy autotool helper scripts from automake
- modify version.sh to be usable from the configure script
- rename Makefile to Makefile.in and use basic variables from configure.ac
Signed-off-by: Karel Zak <kzak@redhat.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Based on a user report, 'max' in help does not clearly point to the
limits that are commonly used as a quota-related term.
Signed-off-by: David Sterba <dsterba@suse.cz>
Particularly during support conversations, people get confused about
which options to use with btrfs check. Adding a flag, --readonly, which
implies the default read-only behaviour and which conflicts with the
read-write operations, should help make the behaviour of the tool clear.
Signed-off-by: Hugo Mills <hugo@carfax.org.uk>
Signed-off-by: David Sterba <dsterba@suse.cz>
The current approach to option parsing, where long-only options are
selected on the basis of their position in the long_options array is
fragile and painful to modify if options are to be inserted into the
list, rather than appended.
Instead, use the last field of struct option to return a value which
cannot be a char (and hence a short option), and simply switch on those
within the case statement.
Signed-off-by: Hugo Mills <hugo@carfax.org.uk>
Signed-off-by: David Sterba <dsterba@suse.cz>
Problem: when the size is too big, the output format will be unaligned.
The __update__columns_max_len function has been updated to fix this
problem.
Signed-off-by: Fan Chengniang <fancn.fnst@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
glibc 2.10+ (5+ years old) enables all the desired features:
_XOPEN_SOURCE 700, __XOPEN2K8, POSIX_C_SOURCE, DEFAULT_SOURCE; with a
single _GNU_SOURCE define in the makefile alone. For portability to
other libc implementations (e.g. dietlibc) _XOPEN_SOURCE=700 is also
defined.
This also resolves Debian bug report filed by Michael Tautschnig -
"Inconsistent use of _XOPEN_SOURCE results in conflicting
declarations". Whilst I was not able to reproduce the results, the
reported fact is that _XOPEN_SOURCE set to 500 in one set of files
(e.g. cmds-filesystem.c) generates/defines different struct stat from
other files (cmds-replace.c).
This patch thus cleans up all feature defines, and sets them at a
consistent level.
Bug-Debian: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=747969
Signed-off-by: Dimitri John Ledkov <dimitri.j.ledkov@intel.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Add an alias to -h to 'filesystem usage', 'filesystem df' and
'device usage' commands, same as the traditional 'df'.
Signed-off-by: David Sterba <dsterba@suse.cz>
Add explain on 'source' and 'target', which is somewhat confusing for
users who want to restore dumped image.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
[reworded texts]
Signed-off-by: David Sterba <dsterba@suse.cz>
Change the immediate number in btrfs_open_ctree_flags to bit shift.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Now we can use -f with -S option when setting seeding flags or clearing
seeding flags
Reported-by: Chen Hanxiao <chenhanxiao@cn.fujitsu.com>
Signed-off-by: Fan Chengniang <fancn.fnst@cn.fujitsu.com>
[updated docs]
Signed-off-by: David Sterba <dsterba@suse.cz>
Sudo may not be configured to run without user prompt, in that case the
tests would be stuck.
Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Scrub on multiple devices may report wrong status if scrub finishes
early on one of them.
Reported-by: Sandy McArthur Jr <sandymac@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Most of the checks run fine without root, but some of them may need to
do a mount test or access the data. Add the support to selectively run
commands under root, hardcoded to sudo for now.
Signed-off-by: David Sterba <dsterba@suse.cz>
Move leaf-corruption type no extent data case to its dir, and add
verification script in test.sh
The verification script is based on manual btrfs-debug-tree check.
The image can also be reused to other leaf-corruption type, like
corrupted leaf contains regular file extent data case.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Now 006-bad-root-items has two cases, one for default case, and one for
skinny metadata case.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Move these obvious btrfs-image to its corresponding dir to use the new
infrastructure.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Change the old btrfsck test infrastructure (btrfs-image dump or xz raw
dump) to the new test infrastructure.
1) Test case layout
The new infrastructure is dir based, each dir is one test type, and can
contain multiple images/scripts for different corner cases.
So layout will be the following:
btrfs-progs
|-tests
|-fsck-tests
|-001-SOME-CORRUPT-TYPE
|-IMAGE-FOR-CASE1
|-IMAGE-FOR-CASE2
2) Test case image types
Only 2 types for test case images.
a) btrfs-image dump
This one is the simplest case, one only needs to add the image to
corresponding dir.
b) custom script
This one is for all the resting cases which can't fit btrfs-image,
like csum error or script can generate the image (this reduces the
size obviously and good for review)
The old binary dump also belongs to this type, so need to add script
to extract them.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
When we want to delete a subvol, we first check to see whether it is
a subvolume or not. After the check, we are sure that it is a valid
subvol, don't have to check its name.
Signed-off-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com>
[removed unused variable 'len']
Signed-off-by: David Sterba <dsterba@suse.cz>
when we use multiple options, error return status will be override by the
last option status.
example: btrfstune -S 1 -r /dev/loop0
when -S option fails and -r option succeeds, return value is 0, rather than
1, where 1 is the right return status.
Reported-by: Chen Hanxiao <chenhanxiao@cn.fujitsu.com>
Signed-off-by: Fan Chengniang <fancn.fnst@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
btrfs qgroup limit has two options -c and -e,. They were forgotten to add
to manpage.
Signed-off-by: Fan Chengniang <fancn.fnst@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
When executing 'btrfs fi show' on unmounted device, even no problem
happens, the return value is still 1 not 0.
The problem lies in search_umounted_fs_uuids(), where when it finds the
given uuid, it should return 1, but later uuid copy overwrites the return
value, causing it always return 0 under that case.
Fix it by pass found as pointer, and return value only indicates
whether anything wrong happens, whether found or not is stored in the
new parameter.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Originally broke in commit c2691f807d
__glibc__ should have been __GLIBC__
We also include features.h ; although most includes (at least stdlib.h)
typically already include it -- at least on glibc, where it matters.
Signed-off-by: Merlijn Wajer <merlijn@wizzup.org>
Reported-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
cmd_snapshot_usage in cmds-subvolume.c contains the short description
twice. Remove the first one.
Signed-off-by: Martin Volf <martin.volf.42@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
The original check_inode_recs() will return -1 if found any error in a
inode_record. This is OK for original design since there is almost
nothing can repair at that time.
However more and more error from nlink mismatch to missing inode item
can be repaired in try_repair_inode(), check_inode_recs() should not
increase the error count if the inode can be repair.
With this patch, repair function for leaf-corruption will not return
error if all corruption inode can be recovered.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Current 'man btrfs-replace' is as follows:
========================================================================
...
...
-f
force using and overwriting <targetdev> even if it looks like
containing a valid btrfs filesystem.
A valid filesystem is assumed if a btrfs superblock is found
which contains a correct checksum. Devices which are currently
mounted are never allowed to be used as the <targetdev>. -B::::
no background replace.
...
...
========================================================================
The format of 'B' option is wrong. So, fix it.
Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
The newly introduced search_chunk_tree_for_fs_info() won't count devid 0
in fi_arg->num_devices, which will cause buffer overflow since later
get_device_info() will fill di_args with devid.
This can be trigger by fstests/btrfs/069 and any operations needs to
iterate over all the devices like 'fi show' or 'dev stat' while
replacing.
The fix is do an extra probe specifically for devid 0 after
search_chunk_tree_for_fs_info() and change num_devices if needed.
Reported-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Signed-off-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Fixes various formatting issues:
Fixes quoting issues in various manpages. Before/After excerpts for
mkfs.btrfs(8), btrfs-qgroup(8):
mkfs.btrfs [-A|--alloc-start '<alloc-start>']
mkfs.btrfs [-A|--alloc-start <alloc-start>]
id 0/<subvolume id>''
id 0/<subvolume id>
If multiple '<attr>'s is given, use comma to separate.
If multiple <attr>s is given, use comma to separate.
The one that is most apparent is the removal of text in
btrfs-subvolume(8) and others. Before/After:
list [options] [-G <value>] [-C <value>] [--sort=rootid,gen,ogen,path] <path>
list [options] [-G [+|-]<value>] [-C [+|-]<value>] [--sort=rootid,gen,ogen,path] <path>
create [-i <qgroupid>] <name>
create [-i <qgroupid>] [<dest>]<name>
This happens because `<foo>` was replaced by `'<foo>'` in the sed
expression, which is then treated as a constrained quote. As a result,
the `[...]` before the string gets interpreted as "quoted text
attributes".
In this patch, the sed expression is dropped and asciidoc is configured
to recognize `<...>` as a unconstrained quoted string (such that
`<attr>s` is correctly emphasized) and to avoid adding quotes in
contexts where it is not needed.
A remaining problem is that some texts (the one between brackets) do not
add the emphasis for `<...>`. While this could be solved by replacing
`[` and `]` with `[` and `]` using sed, it introduces formatting
problems in btrfs-zero-log(8) because the context is ignored.
Signed-off-by: Peter Wu <peter@lekensteyn.nl>
Signed-off-by: David Sterba <dsterba@suse.cz>
The commit f495a2ac66 ("btrfs-progs: fsck: remove unfriendly BUG_ON()
for searching tree failure") is causing tons of extent buffer leak if some
csum mismatches in btrfsck.
This is caused by a misplaced btrfs_release_path(), fix it.
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.cz>
Karel Zak
David Sterba
Xing Gu
Hugo Mills
Fan Chengniang
Dimitri John Ledkov
Qu Wenruo
Gui Hecheng
Merlijn Wajer
Martin Volf
Tsutomu Itoh
Peter Wu