tests: cleanup ts/cramfs/mkfs-endianness

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: cleanup ts/cramfs/fsck-endianness

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: cleanup ts/bitops

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: add fsck:ismounted reg.test

Signed-off-by: Karel Zak <kzak@redhat.com>

lib: add test_ismounted for regression test

Signed-off-by: Karel Zak <kzak@redhat.com>

po: update list of .c files

Signed-off-by: Karel Zak <kzak@redhat.com>

fsck: cosmetic changes (NLS, paths, ...)

Signed-off-by: Karel Zak <kzak@redhat.com>

fsck: link with generic fsprobe wrapper

This is a way how link fsck with libblkid (e2sprogs or util-linxu-ng
version) and libvolume_id.

Signed-off-by: Karel Zak <kzak@redhat.com>

lib: make open_device() optional in fsprobe.c

Signed-off-by: Karel Zak <kzak@redhat.com>

fsck: move fsck from e2fsprogs to util-linux-ng

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: fix low-probe mode return codes

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add TODO hint about blkid_parse_tag_string()

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add cmdline interface for blkid_probe_filter_usage()

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: check calloc() return value

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add TODO hint about DM devnames in sysfs

Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: use pkg-config for blkid and volume_id

It seems that blkid.pc from e2fsprogs has been fixed:

   $ pkg-config --libs blkid
   -lblkid

and the pkg-config does not return any other extra libraries (such
-luuid or -ldevmapper).

Signed-off-by: Karel Zak <kzak@redhat.com>

mount: generic blkid/volume_id wrapper, use blkid_evaluate_*

 * moves the generic libblkid/libvolume_id wrapper (fsprobe.c) from
   mount/ to lib/. We'll use the wrapper in cfdisk and fsck.

   The wrapper supports:

    - obsolete volume_id (udev)
    - obsolete libblkid (e2fsprogs)
    - libblkid (util-linux-ng)

 * mount, umount and swapon when linked against the new libblkid use

    - low-level probing code to read LABEL, UUID or FSTYPE from a device
    - high-level blkid_evaluate_spec() to convert LABEL/UUID to devname

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add Christoph's note about libdisk to TODO

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: refresh TODO file

Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: add temporary libtool *.m4 stuff

We require libtool-2 (because old 1.5.x is crap). Unfortunately,
libtool-2 is still not available in many Linux distributions  -- now I
see it in unstable distributions only. The painless way how to resolve
this problem is to __temporary__ commit generated libtool-2 stuff to
our repository.

Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: libtoolize mount/Makefile.am

 * cleanup mount/Makefile.am
 * add {VOLUMEID,BLKID}_CFLAGS -- necessary for pkg-config
 * add support for linking with in-tree libblkid

Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: libtoolize by libtool-2

We need an infrastructure for stared libraries. The latest libtool-2
seems useful and it's definitely better than the old 1.5 crap.

You need to install libtool-2 when you want to run ./autogen.sh script
after checkout from git repository.

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: start to use ABI versioning

The library ABI and API is backwardly compatible, so it does not
make sense to change the library SONAME. This patch adds a symbol
versioning, naming paradigm is:

BLKID_<maj>.<min>

The original libblkid from e2fsprogs uses "1.0" as a .so version and
"libblkid.1" as a SONAME for all time (at least according to stuff in
/lib/libblkid*)

And the original library is without symbols versioning. It means that
many private functions are exported to applications ;-(

Note that the original blkid_get_library_version() returns
E2FSPROGS_VERSION. The version in util-linux-ng returns BLKID_VERSION
which is <maj>.<min>.<rel>. The <maj>.<min> is the same version as we
use for ABI. This concept seems less confusing than mix a library
version and package version.

Summary:

OLD (e2fsprogs):
  ABI versioning:              -none-
  SONAME:                      libblkid.1
  .so version:                 libblkid.so.1.0
  blkid_get_library_version(): @E2FSPROGS_VERSION@ (e.g. 1.41.1)

NEW (util-linux-ng):
  ABI versioning:              BLKID_<maj>.<min>
  SONAME:                      libblkid.1
  .so version:                 libblkid.so.<maj.<min> (e.g. 1.41)
  blkid_get_library_version(): @BLKID_VERSION@ (e.g. 1.41.1)

  (BLKID_VERSION = <maj>.<min>.<rel>)

Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: add --with=fsprobe=builtin

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add findfs(8)

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: refresh TODO file

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: clean up man pages

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add -L -U options (evaluation API)

Signed-off-by: root <root@nb.net.home>

blkid: add blkid_evaluate_spec()

Signed-off-by: Karel Zak <kzak@redhat.com>

mount: move realpath.c code to lib/

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: fix typo (syntax error)

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: compile TEST_PROGRAMs

Well, the src/Makefile.am is not prefect. Maybe it's time to switch to
libtools ....

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add support for /etc/blkid.conf file

We need a config file for the library. Now the library supports
config options:

EVALUATION=<method>[,<method>]

SEND_UEVENT=<yes|not>

CACHE_FILE=<path>

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add new options to blkid.8 and help output

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: support via raid version 2

[kzak@redhat.com: patch from udev volume_id]

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: refresh TODO file

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: fix typo in low-probe test

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: cleanup _LOGPROBE debug messages

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add blkid_do_safeprobe()

The function blkid_do_probe() is able to detect more filesystems on
the device

while(blkid_do_probe(pr) == 0)
...

but in many cases we need only one exact answer, and we also need to
be sure that there is not any other FS on the device.

For example it's possible to create valid LUKS (or vfat, ...) header
and valid linux swap header on the same device -- in such case the
device can be interpreted (by mount/swapon) in two completely
different ways. An ambivalent result is always error -- the library
never returns such result.

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add lvm2 reg.test

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add version support to LVM2

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add GFS2 reg. test

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add uuid and version support to gfs2

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add reg.tests for HFS and HFS+

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add vol_id call to blkid regression test

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add  to reiser

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add ocfs2 version

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: fix ..._strncpy_uuid

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add ddf raid regression test

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: fix blkid_do_probe()

Fix small non-sense from from a recent commit.

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: remove unnecessary debug message

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: use blkid_probe_strcpy_uuid() for luks

LUKS stores UUID as a string...

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: improve ddf detection

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: set size for non-blkdevs, add blkid_probe_strcpy_uuid()

- fix blkid_probe_set_device() to work for non-blkdevs (e.g. file images),
  in such case we need to set the size of data from stat->st_size

- add blkid_probe_strcpy_uuid() for filesystems where UUID is stored as a string
  (e.g. DDF raid, luks, ...)

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add netware regression test

Thanks to Kay Sievers for the image and information.

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: netware SB has to be packed

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add hpfs regression test

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: fix udev output

- uses the "safe" output for non-encoded values
- makes the print_udev_format() more readable

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: remove unexpected exit from *_subtest functions

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: fix blkid_safe_string()

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add a note to TODO list

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: add libblkid regression tests (images from e2fsprogs)

- adds basic infrastructure for blkid regression tests
- imports test images from e2fsprogs (all tests PASS!:-)

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: fix the final message for subtests

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: add support for subtests

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: move some generic stuff from ts_init() to a new ts_init_env()

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: add "byte-order" to helpers/test_sysinfo

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: fix output string

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: cleanup ts/mount

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: cleanup ts/swapon

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: cleanup ts/script

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: cleanup ts/paths

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: cleanup ts/namei

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: cleanup ts/look

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: cleanup ts/login

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: cleanup ts/ipcs

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: cleanup ts/hwclock

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: refresh and cleanup cramfs/mkfs

Signed-off-by: root <root@nb.net.home>

tests: cleanup ts/col scripts

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: cleanup ts/cal scripts

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: fix TS_* paths

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: add ./run.sh <component>

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: add support for subdirs to basic test functions

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: create expected/$(component)/$(testname)

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: remove input directory

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: create subdirs for test scripts

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: refresh TODO file

... some nice ideas from brainstorming with Ted, Kay and Scott.

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: use sizeof() for hfs uuid

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: hfs - do not set UUID for emtpy finder info

Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>

blkid: refresh TODO file

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add udev ID_FS_* output to blkid binary

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add udev string encoding routines

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add -p and low-probe mode to blkid binary

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add DEBUG_LOWPROBE, cleanup a little debug stuff

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add btrfs support

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: remove unnecessary ifdef __cplusplus

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: recognize ext3 with test_fs set as ext3

It seems that if we have the test_filesystem flag set on an ext3
filesystem(!) on a system which provides ext4, blkid gets confused.

According to the current logic:

* It's not an ext4dev filesystem, because the system provides ext4.
* It's not an ext4 filesystem, because it has no ext4 features.
* It's not an ext3 filesystem, because the test flag is set.

In the end, it's nothing.

blkid should return *something* that is mountable... I'm inclined to
think that ext3 should be the right answer, if no ext4-specific features
are set.

This would mean just dropping the EXT2_FLAGS_TEST_FILESYS test in
probe_ext3(), because ext4 & ext4dev probes have come first already.

[kzak@redhat.com: port from e2fsprogs to util-linux-ng tree]

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: Refuse to create a device structure for a non-existent device

Fix blkid_get_dev() so it will never return a device structure if the
device file doesn't exist.

[kzak@redhat.com: port from e2fsprogs to util-linux-ng tree]

Addresses-Debian-Bug: #502541
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: fix file descriptor leak when checking for a module

[kzak@redhat.com: port from e2fsprogs to util-linux-ng tree]

Signed-off-by: Karel Zak <kzak@redhat.com>

blkis: fix detection of ext4dev as ext4

If only ext4 is available (as a module or in /proc/filesystems)
blkid wasn't properly testing for it, because the time checks
were backwards and always failed.  This caused old ext4dev
filesystems to fail to mount as ext4.  With this patch it works
fine.

Also, don't try to check for modules on a non-Linux system.

[kzak@redhat.com: port from e2fsprogs to util-linux-ng tree]

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: use Requires.private and fix the include directory

Signed-off-by: Karel Zak <kzak@redhat.com>