OSDN Git Service
Colin Watson [Wed, 17 Nov 2010 15:14:50 +0000 (15:14 +0000)]
libparted: ignore zero-length devices
* bootstrap.conf (gnulib_modules): Add xstrtoll.
* configure.ac: Create DYNAMIC_LOADING output variable.
* libparted/arch/linux.c (_device_get_length): Return the value of
PARTED_TEST_DEVICE_LENGTH if set in the environment, strictly for use by
the test suite.
(init_generic): If geometry probing fails because the device was
zero-length, return quietly rather than throwing an exception. This has
been observed in the wild with cciss devices, and it's difficult for
partitioners to tell the difference between that and more serious
errors.
* libparted/tests/Makefile.am (TESTS): Add t2100-zerolen.sh.
(check_PROGRAMS): Add zerolen.
(zerolen_SOURCES): Add.
(TESTS_ENVIRONMENT): Add DYNAMIC_LOADING and ENABLE_DEVICE_MAPPER.
* libparted/tests/t2100-zerolen.sh: New file.
* libparted/tests/zerolen.c: New file.
* tests/test-lib.sh (wait_for_dev_to_appear_): New function, copied from
tests/t-local.sh.
* NEWS (Bug fixes): Mention it.
Jim Meyering [Tue, 9 Nov 2010 09:25:36 +0000 (10:25 +0100)]
dos: fix a bug affecting very small devices (smaller than 1 cylinder)
This bug was introduced in commit
c79d91ec, "dos: accommodate very
small devices (useful for testing)".
* libparted/labels/dos.c (_primary_constraint): The bug was to
skip setting start_geom for small devices. That led to a used-
uninitialized bug in the subsequent ped_constraint_new call.
The fix is to relax the constraint to use a starting sector of "1",
if necessary. Report and diagnosis by Jean-Christian de Rivaz in
http://thread.gmane.org/gmane.comp.gnu.parted.bugs/10178
* NEWS (Bug fixes): Mention it.
Jim Meyering [Sun, 10 Oct 2010 17:01:52 +0000 (19:01 +0200)]
maint: don't use obsolete gnulib modules
* bootstrap.conf (gnulib_modules): Use calloc-gnu, malloc-gnu
and realloc-gnu modules, rather than calloc, malloc and realloc.
The shorter-named modules are now deprecated.
Jim Meyering [Sat, 9 Oct 2010 13:24:48 +0000 (15:24 +0200)]
maint: describe policy on copyright year number ranges
* README: Mention coreutils' long-standing policy on use of M-N
ranges in copyright year lists. Requested by Richard Stallman.
Jim Meyering [Sun, 10 Oct 2010 15:59:27 +0000 (17:59 +0200)]
build: update gnulib submodule to latest, and update bootstrap
* bootstrap: update from gnulib
Petr Uzel [Mon, 23 Aug 2010 12:29:46 +0000 (14:29 +0200)]
build: suggest libuuid from util-linux-ng package
* configure.ac: libuuid was moved from e2fsprogs to
util-linux-ng-2.16. Because distributions now often
build e2fsprogs without libuuid support, suggest installing
libuuid from util-linux-ng if it is missing, not from e2fsprogs.
Colin Watson [Thu, 19 Aug 2010 10:07:20 +0000 (11:07 +0100)]
libparted: remove limits on loop labels
There's no reason to impose any particular limit on loop labels, since
they just represent a single large partition. Sector counts over 2^32
are needed for large RAID arrays. Change the limit to 2^64 since that's
the upper limit imposed by libparted and it saves us implementing the
limit functions separately. This bug appears to have been introduced
by commit
2dbc645c.
* libparted/labels/pt-limit.gperf: Change limits on "loop" to 2^64.
* tests/t9021-maxima.sh: Update for the new loop limit.
* NEWS (Bug fixes): Mention it.
Jim Meyering [Wed, 18 Aug 2010 22:52:02 +0000 (18:52 -0400)]
avoid non-srcdir "make check" failure
* tests/Makefile.am (EXTRA_DIST): Add init.cfg.
The failure was introduced by commit
6f7c0f12.
Jim Meyering [Wed, 18 Aug 2010 22:39:53 +0000 (18:39 -0400)]
build and tests: update bootstrap and init.sh from gnulib
* tests/init.sh: Update from gnulib.
* bootstrap: Likewise.
Jim Meyering [Wed, 18 Aug 2010 21:41:43 +0000 (17:41 -0400)]
build: update gnulib submodule to latest
Hans de Goede [Tue, 20 Jul 2010 10:17:53 +0000 (12:17 +0200)]
linux: recognize scsi disks with a high major number (128-135)
* libparted/arch/linux.c (SCSI_BLK_MAJOR): Recognize high major number.
Jim Meyering [Thu, 15 Jul 2010 00:16:14 +0000 (19:16 -0500)]
libparted: avoid regression when processing a whole-disk FAT partition
Without this change, we would improperly classify a whole-disk partition
containing a FAT file system as a DOS partition table with no partitions.
Introduced by commit
d732a2b7 on 2008-05-28.
* libparted/labels/dos.c (maybe_FAT): New function.
(msdos_probe): Use it.
Reported by ChenMin in
http://thread.gmane.org/gmane.comp.gnu.parted.bugs/10115
* NEWS (Bug fixes): Mention it.
* tests/t3400-whole-disk-FAT-partition.sh: New file. Test for the bug.
* tests/Makefile.am (TESTS): Add it.
Jim Meyering [Sat, 17 Jul 2010 12:20:54 +0000 (07:20 -0500)]
tests: use init.cfg; required for init.sh-using tests
* tests/init.cfg: New file.
(require_512_byte_sector_size_): Slightly different function
than the one in test-lib.sh.
* tests/init.cfg: New file.
Jim Meyering [Wed, 14 Jul 2010 21:39:05 +0000 (16:39 -0500)]
build: update gnulib submodule to latest
Jim Meyering [Sat, 17 Jul 2010 00:21:44 +0000 (19:21 -0500)]
maint: remove trailing blank
* tests/t1101-busy-partition.sh: Remove trailing blank.
Otavio Salvador [Tue, 29 Jun 2010 17:22:34 +0000 (14:22 -0300)]
tests: update resize-fs to use wait_for_dev_to_appear_
* t3000-resize-fs.sh: rely on wait_for_dev_to_appear_ to wait for the
device to be crated.
Otavio Salvador [Tue, 29 Jun 2010 12:51:03 +0000 (09:51 -0300)]
tests: rely on require_scsi_debug_module_ to modprobe scsi_debug module
* t2310-dos-extended-2-sector-min-offset.sh: remove explicit modprobe
since it is done by require_scsi_debug_module_.
* t3200-type-change.sh: likewise.
* t9020-alignment.sh: likewise.
* t9030-align-check.sh: likewise.
* t9040-many-partitions.sh: likewise.
Otavio Salvador [Mon, 28 Jun 2010 22:45:51 +0000 (19:45 -0300)]
tests: add one: try to remove a busy partition
This test tries to reproduce the issue reported in Debian bug #582818[1].
1. http://bugs.debian.org/582818
Basically it does:
* create two primary partitions
* mount the second one
* remove the first (must work)
* try to remove the second (must fail)
* tests/t1101-busy-partition.sh: new file.
* tests/Makefile.am (TESTS): add the new test in check target.
Otavio Salvador [Sat, 26 Jun 2010 22:23:16 +0000 (19:23 -0300)]
tests: skip loop-clobber-infloop if mkswap cannot be found
* tests/t0400-loop-clobber-infloop.sh: Skip the test is case of mkswap
cannot be found.
Jim Meyering [Sat, 26 Jun 2010 06:59:18 +0000 (08:59 +0200)]
build: update gnulib submodule to latest
* bootstrap: Update from gnulib.
* tests/init.sh: Likewise.
Colin Watson [Tue, 15 Jun 2010 18:49:40 +0000 (19:49 +0100)]
sun: revert "implement disk flag operations"
This reverts the libparted/labels/sun.c part of
723ea23c5df68cbe67d1f518ef484f4c77f516fa. Sun disk labels do
not appear to be able to handle non-cylinder alignment
(http://bugs.debian.org/579948).
* libparted/labels/sun.c: Revert.
Jim Meyering [Sat, 26 Jun 2010 07:22:59 +0000 (09:22 +0200)]
tests: adjust sun-partition-creating test to conform
* tests/t4000-sun-raid-type.sh: Adjust partition size so the
end falls on a cylinder boundary.
Jim Meyering [Fri, 28 May 2010 19:02:45 +0000 (21:02 +0200)]
maint: update README-release
* README-release: Update to match the template used for diffutils,
coreutils, etc.
Jim Meyering [Fri, 28 May 2010 18:49:04 +0000 (20:49 +0200)]
maint: add gettext to list of bootstrapped-with tools in announcement
* cfg.mk (bootstrap-tools): Add gettext to the list.
Jim Meyering [Fri, 28 May 2010 18:33:10 +0000 (20:33 +0200)]
post-release administrivia
* NEWS: Add header line for next release.
* .prev-version: Record previous version.
* cfg.mk (old_NEWS_hash): Auto-update.
Jim Meyering [Fri, 28 May 2010 18:26:06 +0000 (20:26 +0200)]
version 2.3
* NEWS: Record release date.
Jim Meyering [Fri, 28 May 2010 18:24:45 +0000 (20:24 +0200)]
doc: regenerate two files using gettext-0.18
* doc/pt_BR/partprobe.8.pt_BR.po: Regenerate with gettext-0.18.
* doc/C/po/partprobe.8.pot: Likewise.
Jim Meyering [Fri, 28 May 2010 17:55:34 +0000 (19:55 +0200)]
build: require gettext-0.18 after all, because
since gnulib is pulling in m4/gettext.m4 from 0.18, and there's
a cross-check between that file and po/Makefile.in.in, it's easiest
simply to require 0.18 of those who run build tools. Otherwise,
requiring 0.17, build-from-git would fail due to the mismatch:
0.17 in Makefile.in.in and 0.18 in m4/gettext.m4.
* configure.ac (AM_GNU_GETTEXT_VERSION): Require gettext 0.18.
Jim Meyering [Fri, 28 May 2010 15:27:52 +0000 (17:27 +0200)]
build: update gnulib submodule to latest
Jim Meyering [Fri, 28 May 2010 15:26:45 +0000 (17:26 +0200)]
maint: use the do-release-commit-and-tag module from gnulib
* bootstrap.conf (gnulib_modules): Add do-release-commit-and-tag.
Petr Uzel [Fri, 28 May 2010 15:11:05 +0000 (17:11 +0200)]
doc: fix a typo
* doc/parted.texi: s/onstraints/constraints/
Petr Uzel [Fri, 28 May 2010 14:42:20 +0000 (16:42 +0200)]
parted: do_align_check returns always 1 in interactive mode
* parted/parted.c (do_align_check): Always return 1 in interactive mode.
Jim Meyering [Thu, 27 May 2010 14:32:34 +0000 (16:32 +0200)]
build: really use gettext-0.17
* configure.ac: Correct previous change to use 0.17, not 0.18,
since gettext-0.18 is new enough that it makes building from git
a little too inconvenient. Spotted by Petr Uzel.
Petr Uzel [Wed, 26 May 2010 12:29:01 +0000 (14:29 +0200)]
parted: make align-check work in interactive mode
* parted/parted.c (do_align_check): rework the function so that
it reports partition alignment in interactive mode as described in
documentation.
Petr Uzel [Wed, 26 May 2010 12:27:07 +0000 (14:27 +0200)]
parted: initialize align_type to default value
* parted/parted.c (do_align_check): Initialize align_type to
default alignment; this prevents reading of uninitialized memory
later in command_line_get_align_type()
Petr Uzel [Wed, 26 May 2010 12:10:04 +0000 (14:10 +0200)]
parted: plug memory leak
* parted/parted.c (_done_messages): Free unit_msg.
Jim Meyering [Wed, 26 May 2010 10:19:07 +0000 (12:19 +0200)]
tests: avoid spurious failure of the root-only many-partitions test
* tests/t9040-many-partitions.sh: When creating a scsi_debug
device, honor the block size suggested by the PARTED_SECTOR_SIZE
environment variable, and thus required in expected output.
Jim Meyering [Wed, 26 May 2010 09:30:02 +0000 (11:30 +0200)]
build: update gnulib submodule to latest
Jim Meyering [Wed, 26 May 2010 09:43:09 +0000 (11:43 +0200)]
maint: remove useless definitions
* libparted/fs/xfs/platform_defs.h [__sparc__] (O_DIRECT):
Remove now-unneeded definition.
* parted/table.c (wchar_t): Remove definition.
* parted/strlist.c (wchar_t): Likewise.
* .x-sc_prohibit_always-defined_macros: New file.
Exempt two files from this syntax-check rule.
* Makefile.am (EXTRA_DIST): Add it here.
Jim Meyering [Wed, 26 May 2010 09:32:56 +0000 (11:32 +0200)]
maint: update helper scripts
* tests/help-version: Update from coreutils.
* bootstrap: Update from gnulib.
* tests/init.sh: Likewise.
Jim Meyering [Wed, 26 May 2010 09:29:56 +0000 (11:29 +0200)]
build: use gettext-0.17
* configure.ac: Use gettext-0.17.
* bootstrap.conf (gnulib_modules): Use gettext-h, not gettext.
since the latter drags in a depedency on gettext 0.18.
Jim Meyering [Tue, 25 May 2010 13:42:35 +0000 (15:42 +0200)]
fix typo in comment
Jim Meyering [Mon, 10 May 2010 11:00:20 +0000 (13:00 +0200)]
libparted: fail (don't just warn) upon failure to inform kernel of...
partition change(s).
* libparted/arch/linux.c (_disk_sync_part_table): Throw an error, not
merely a warning. When this fails, it indicates that the kernel has
not been informed, and that must translate to an error, since doing
otherwise would violate the contract of ped_disk_commit_to_os.
Suggested by Hans De Goede.
Jim Meyering [Thu, 6 May 2010 16:37:21 +0000 (18:37 +0200)]
tests: make the new many-partitions test fail if it takes "too long"
* tests/t9040-many-partitions.sh: Fail if creating 60 partitions
takes more than a minute.
Jim Meyering [Thu, 6 May 2010 13:20:00 +0000 (15:20 +0200)]
maint: use latest bootstrap script from gnulib
* bootstrap: Sync from gnulib.
Hans de Goede [Thu, 6 May 2010 13:09:35 +0000 (15:09 +0200)]
gpt: recognize microsoft recovery GUID
Microsoft advices creating recovery partitions with a GUID of
de94bba4-06d1-4d40-a16a-
bfd50179d6ac, see:
http://technet.microsoft.com/en-us/library/
dd744301%28WS.10%29.aspx
This patch makes parted recognize this guid (using the DIAG flag).
* libparted/labels/gpt.c (PARTITION_MSFT_RECOVERY, GPTPartitionData,
_parse_part_entry, gpt_partition_new, gpt_partition_set_system,
gpt_partition_set_flag, gpt_partition_get_flag,
gpt_partition_is_flag_available): Support microsoft recovery GUID.
* NEWS (New features): Add it.
Hans de Goede [Thu, 6 May 2010 08:25:35 +0000 (10:25 +0200)]
msdos: recognize id 27 as a diagnostics (recovery) partition type
msdos partitiontable partition id 27 is recommended by microsoft for
use as id for recovery partitions, see:
http://technet.microsoft.com/en-us/library/
dd744364%28WS.10%29.aspx
http://bugzilla.redhat.com/show_bug.cgi?id=534066
* libparted/labels/dos.c (PARTITION_MSFT_RECOVERY): Define.
(raw_part_parse): Recognize id 27
(msdos_partition_set_system): Recognize id 27
Jim Meyering [Thu, 6 May 2010 12:11:33 +0000 (14:11 +0200)]
tests: add one: create 60 partitions on a scsi device
* tests/t9040-many-partitions.sh: New test.
* tests/Makefile.am (TESTS): Add it.
(TESTS_ENVIRONMENT): Export AWK to test scripts.
Inspired by the report of a many-partition test
taking far longer on RHEL5 than on RHEL4 in
http://bugzilla.redhat.com/589117
Jim Meyering [Wed, 5 May 2010 16:42:26 +0000 (18:42 +0200)]
doc: update NEWS
* NEWS (Bug fixes): Mention the ext2 2TiB FS-creation bug.
Jim Meyering [Wed, 5 May 2010 16:25:38 +0000 (18:25 +0200)]
build: update gnulib submodule to latest
Martin Poole [Wed, 5 May 2010 16:24:22 +0000 (18:24 +0200)]
ext2: don't overflow when creating a partition of size 2TiB or larger
Without this patch, mkpartfs would succeed in creating an ext2
partition, but the resulting partition would fail a mkfs.ext2 -f check.
* libparted/fs/ext2/parted_io.c (do_read): Avoid overflow 32-bit
sector count overflow for a partition of size 2TiB or larger.
(do_write): Likewise.
See http://bugzilla.redhat.com/584057#c16 a reproducer.
Hans de Goede [Wed, 28 Apr 2010 17:14:36 +0000 (19:14 +0200)]
_device_unregister: check for unregistering the same device twice
_device_unregister() can get called 2 times for the same device if
a libparted user explictly removes the device from the cache using
ped_device_cache_remove(), it gets called and it then becomes the
user's responsibility to free the PedDevice by calling
ped_device_destroy(). ped_device_destroy() will then call
_device_unregister() a second time with the same device, so if the
device is not found in the list, do nothing.
* libparted/device.c(_device_unregister): Check for unregistering
the same device twice.
Reported by Jan Stodola in <http://bugzilla.redhat.com/568346>.
Jim Meyering [Fri, 30 Apr 2010 13:59:14 +0000 (15:59 +0200)]
tests: avoid root-only/XFS limit test failure on 32-bit system
* tests/t4100-msdos-partition-limits.sh: Due to an inherent XFS
limitation on 32-bit systems, this test would fail a set-up
file creation step for simulated sector sizes of 4K and up.
Skip this test in those cases.
Upon dd failure, propagate its diagnostic to output, to make this
sort of problem easier to diagnose.
* tests/t4100-dvh-partition-limits.sh: Likewise.
Jim Meyering [Fri, 30 Apr 2010 10:28:16 +0000 (12:28 +0200)]
libparted: avoid race in informing the kernel of partition table changes
When sync'ing a partition table change using the latest
code, sometimes we'd get an unwarranted failure like this:
Warning: Partition(s) 1 on /dev/sdd have been written, but we
have been unable to inform the kernel of the change, probably because
it/they are in use. As a result, the old partition(s) will remain in
use. You should reboot now before making further changes.
To be precise, when running the partition-resizing root-only test
in a loop:
for i in $(seq 240); do make -C tests check VERBOSE=yes \
TESTS=t3000-resize-fs.sh >& log.$i && printf . || echo $i $?; done
I would typically see about 50% of them fail on a Fedora 13 system.
It was obvious that this was due to a race condition when I found that
modifying that tests' parted...resize invocation to go via strace changed
the timing enough to make the test pass every time.
The fix is to retry the partition-removal step upon any EBUSY failure,
currently for up to 1 second (retrying up to 100 times, sleeping 10ms
after each failure).
* libparted/arch/linux.c (_disk_sync_part_table): Allocate "ok" using
calloc, now that its initial values matter.
Retry each removal upon EBUSY-failure.
* bootstrap.conf (gnulib_modules): Use gnulib's usleep module.
Jim Meyering [Fri, 30 Apr 2010 10:01:08 +0000 (12:01 +0200)]
libparted: variable renaming, minor "goto" reorg
* libparted/arch/linux.c (_disk_sync_part_table): Rename local array:
s/rets/ok/, for readability.
Use only a single label, "cleanup:", rather than two: free_rets
and free_errnums.
Jim Meyering [Fri, 30 Apr 2010 09:45:51 +0000 (11:45 +0200)]
libparted: remove now-worse-than-useless _kernel_reread_part_table
Now that we're using BLKPG properly, there's no point in using the
less-functional BLKRRPART ioctl to make the kernel reread the partition
table.
More importantly, this function would fail when any partition is in
use, in spite of our having carefully vetted them via BLKPG ioctls.
* libparted/arch/linux.c (_kernel_reread_part_table): Remove function.
(linux_disk_commit): Don't call it.
Jim Meyering [Thu, 29 Apr 2010 17:02:07 +0000 (19:02 +0200)]
tests: correct a root-only test
* tests/t2310-dos-extended-2-sector-min-offset.sh: Don't hard-code
"/dev/sdd5" in expected output.
Jim Meyering [Thu, 29 Apr 2010 14:02:40 +0000 (16:02 +0200)]
maint: avoid code duplication
* libparted/labels/dos.c (clear_flags): New function.
Factored (5 times) out of ...
(msdos_partition_set_flag): ...here. Use the new function.
Hans de Goede [Mon, 19 Apr 2010 11:55:32 +0000 (13:55 +0200)]
dos: add a partition flag for diagnostics / recovery partitions
msdos partition tables can contain diagnostics partitions (often used
nowadays as system recovery partitions). For some users of libparted
(i.e., anaconda) it is useful to know if a dos partition is a regular
partition or a diagnostic partition. For example, see
<http://bugzilla.redhat.com/534066>.
* include/parted/disk.h(_PedPartitionFlag): Add PED_PARTITION_DIAG
* libparted/disk.c(ped_partition_flag_get_name): Add PED_PARTITION_DIAG
* libparted/labels/dos.c: Add support for PED_PARTITION_DIAG flag
* doc/parted.texi (set): Document it.
Jim Meyering [Wed, 28 Apr 2010 20:50:35 +0000 (22:50 +0200)]
tests: distribute new file, init.sh
* tests/Makefile.am (EXTRA_DIST): Add init.sh.
Jim Meyering [Wed, 28 Apr 2010 17:27:16 +0000 (19:27 +0200)]
tests: add help-version test (from coreutils)
* tests/help-version: New file
* tests/Makefile.am (TESTS_ENVIRONMENT): Pass built_programs and VERSION
to test scripts -- both are required by help-version.
(TESTS): Add help-version.
Jim Meyering [Wed, 28 Apr 2010 20:26:41 +0000 (22:26 +0200)]
ui: make --help tell where to report bugs
* parted/ui.c (help_msg): Tell where to report bugs.
Jim Meyering [Wed, 28 Apr 2010 20:18:51 +0000 (22:18 +0200)]
tests: use gnulib's init.sh
* tests/init.sh: New file, from gnulib.
Jim Meyering [Wed, 28 Apr 2010 17:42:44 +0000 (19:42 +0200)]
tests: rename generated init.sh to old-init.sh
* tests/Makefile.am (old-init.sh): Rename from init.sh.
* libparted/tests/Makefile.am: Likewise.
* tests/test-lib.sh: Rename sole use.
* .gitignore: Rename here, too.
Jim Meyering [Wed, 28 Apr 2010 16:35:27 +0000 (18:35 +0200)]
build: update gnulib submodule to latest
Jim Meyering [Wed, 28 Apr 2010 17:20:21 +0000 (19:20 +0200)]
tests: improved portability
* tests/test-lib.sh: Don't use test's -a option.
Jim Meyering [Wed, 28 Apr 2010 17:16:49 +0000 (19:16 +0200)]
maint: remove empty lines at end of file
This lets one more "syntax-check" test pass.
Run this command:
git ls-files|grep -v gnulib|xargs perl -pi -0777 -e 's/\n\n+$/\n/'
Jim Meyering [Wed, 28 Apr 2010 17:14:17 +0000 (19:14 +0200)]
maint: pass newer "make syntax-check" tests from gnulib
* cfg.mk (local-checks-to-skip): Add these, to pass "make syntax-check"
sc_error_message_uppercase, sc_error_message_period.
* libparted/fs/ext2/ext2.h: Remove "#if HAVE_INTTYPES_H" guard.
Jim Meyering [Wed, 21 Apr 2010 12:00:26 +0000 (14:00 +0200)]
tests: test all boolean flags
* tests/t3310-flags.sh: New test, suggested by Hans De Goede.
* tests/Makefile.am (TESTS): Add it.
Jim Meyering [Tue, 27 Apr 2010 10:19:45 +0000 (12:19 +0200)]
doc: use a valid http:.../AUTHORS URL
* debug/clearfat/clearfat.c (AUTHORS): Point to the git repo.
* parted/parted.c (AUTHORS): Likewise.
* partprobe/partprobe.c (AUTHORS): Likewise.
* parted/parted.c: Include "version.h".
(_version): Use Version, not VERSION.
Jim Meyering [Mon, 26 Apr 2010 16:44:40 +0000 (18:44 +0200)]
ui: get version number from a separately-compiled file
* parted/Makefile.am (version.c, version.h): Generate.
(parted_LDADD): Add libver.a.
(noinst_LIBRARIES, nodist_libver_a_SOURCES): Define.
(BUILT_SOURCES): Initialize and append.
(MAINTAINERCLEANFILES): Update.
* parted/parted.c: Include "version.h".
(_version): Use the new global variable, Version, not VERSION.
Jim Meyering [Fri, 23 Apr 2010 18:11:23 +0000 (20:11 +0200)]
libparted: _disk_sync_part_table: allow interactive ignore
* libparted/arch/linux.c (_disk_sync_part_table): Change diagnostic
so that it makes it clear what failed. Let an interactive "ignore"
cause the function to succeed.
Jim Meyering [Sun, 18 Apr 2010 09:15:45 +0000 (11:15 +0200)]
tests: remove partition explicitly before mklabel, to avoid new failure
* tests/t3000-resize-fs.sh: Before, mklabel was sufficient to clear
the partition table in preparation for the next iteration in this test.
However, with recent BLKRRPART/BLKPG changes that "mklabel" would evoke
a warning. Remove the sole partition to avoid that.
Jim Meyering [Mon, 12 Apr 2010 10:08:16 +0000 (12:08 +0200)]
libparted: linux_disk_commit: don't ignore _disk_sync_part_table failure
* libparted/arch/linux.c (linux_disk_commit):
When calling _disk_sync_part_table, always return its result.
Jim Meyering [Mon, 12 Apr 2010 10:06:30 +0000 (12:06 +0200)]
libparted: _disk_sync_part_table: always return 0 upon failure
* libparted/arch/linux.c (_disk_sync_part_table):
Return 0 (not 1) upon failure.
Colin Watson [Mon, 29 Mar 2010 13:40:54 +0000 (14:40 +0100)]
dos: always allow at least two sectors for extended boot record
Unless specifically told otherwise, the Linux kernel considers extended
boot records to be two sectors long, in order to "leave room for LILO".
When using anything other than cylinder alignment, libparted was only
allowing one sector in the minimum extended partition geometry, which in
some situations (e.g. following Phillip Susi's patch to reintroduce
BLKPG) could confuse the kernel into thinking that the EBR and the first
logical partition overlapped.
* libparted/labels/dos.c (_get_min_extended_part_geom): Allow at least
two sectors for the extended boot record.
Jim Meyering [Tue, 30 Mar 2010 09:57:35 +0000 (11:57 +0200)]
tests: test for new 2-sector minimum ext-logical-separation
* tests/t2310-dos-extended-2-sector-min-offset.sh: New file.
* tests/Makefile.am (TESTS): Add it.
Jim Meyering [Tue, 30 Mar 2010 14:50:59 +0000 (16:50 +0200)]
dos: accommodate very small devices (useful for testing)
* libparted/labels/dos.c (_primary_constraint): Don't pass a negative
"device_length" to ped_geometry_init when the underlying device
has fewer sectors than a "cylinder".
Jim Meyering [Mon, 12 Apr 2010 08:13:18 +0000 (10:13 +0200)]
libparted: allocate storage for diagnostic only when needed
* libparted/arch/linux.c (_disk_sync_part_table): Rename local:
s/parts/bad_part_list/
Allocate space for it only if needed.
Use malloc rather than ped_malloc.
Move decl of FD down to first use.
Jim Meyering [Mon, 12 Apr 2010 07:55:38 +0000 (09:55 +0200)]
libparted: adjust code formatting
Jim Meyering [Thu, 15 Apr 2010 11:45:44 +0000 (13:45 +0200)]
libparted: reword a diagnostic
Phillip Susi [Fri, 9 Apr 2010 19:22:20 +0000 (15:22 -0400)]
libparted: improve BLKPG error checking
This patch cleans up the BLKPG code that the previous patch put back
to perform proper error checking and in the event that some partitions
are in use, they can not be modified in the running kernel using BLKPG.
Warn the user that this is the case and advise them to reboot, just like
we do when BLKRRPART fails for the same reason, unless the partition in
question is unchanged.
Phillip Susi [Tue, 30 Mar 2010 19:04:43 +0000 (15:04 -0400)]
libparted: reenable use of BLKPG ioctls
This patch effectively reverses commit
1d8f9bec which removed the code
using the new BLKPG ioctls instead of the old BLKRRPART ioctl to update
the in-kernel partition table. The reason for this is because BLKRRPART
fails if any partition on the disk is in use, but the BLKPG ioctls allow
you to manipulate the other partitions on the disk without requiring a
reboot. Also BLKRRPART requires that the kernel understand the
partition table on the disk, which may not always be the case.
Hans de Goede [Fri, 23 Apr 2010 11:08:43 +0000 (13:08 +0200)]
linux: detect dm_task_run failure
We were checking for a return value of < 0 for dm_task_run errors, but
dm_task_run returns 0 on error (and 1 on success). Thanks to Joe Jin
for spotting this, see Red Hat bug 582907.
* libparted/arch/linux.c(_dm_remove_map_name, _dm_is_part,
_dm_remove_parts, _dm_add_partition): dm_task_run returns 0 on error.
Jim Meyering [Tue, 20 Apr 2010 16:30:20 +0000 (18:30 +0200)]
tests: test for palo and prep interaction
* tests/t3300-palo-prep.sh: New file.
* tests/Makefile.am (TESTS): Add it.
Hans de Goede [Mon, 19 Apr 2010 11:55:33 +0000 (13:55 +0200)]
dos: fix prep / palo flag handling
Setting the prep flag was not clearing the palo flag and vice versa
* libparted/labels/dos.c (msdos_partition_set_flag): Fix prep / palo
flag handling.
Jim Meyering [Tue, 20 Apr 2010 16:21:40 +0000 (18:21 +0200)]
tests: remove unnecessary (duplicate) local var assignment
* tests/t9021-maxima.sh: Remove duplicate ss=$sector_size_.
Colin Watson [Tue, 20 Apr 2010 12:01:33 +0000 (13:01 +0100)]
maint: make PED_ASSERT and PED_DEBUG if-safe
Constructions such as:
if (foo)
PED_DEBUG (0, "foo");
else
PED_DEBUG (0, "not foo");
did not compile reliably, depending on the configuration. Make this
work.
* include/parted/debug.h [DEBUG && __GNUC__ && !__JSFTRACE__]
(PED_DEBUG): Remove trailing semicolon.
[!DEBUG] (PED_ASSERT): Use do-while rather than while.
[!DEBUG] (PED_DEBUG): Likewise.
Jim Meyering [Sat, 17 Apr 2010 15:39:26 +0000 (17:39 +0200)]
doc: add HACKING
* HACKING: New file. Copied from coreutils, with s/coreutils/parted/
and a few minor edits.
Petr Uzel [Fri, 16 Apr 2010 08:01:08 +0000 (10:01 +0200)]
.gitignore: ignore index files generated by ctags and cscope
Petr Uzel [Fri, 16 Apr 2010 08:01:29 +0000 (10:01 +0200)]
libparted: add support for Linux software RAID arrays (mdraid)
Recognize MD devices, so "parted -s /dev/md0 print" now prints
"Model: Linux Software RAID Array (md)" instead of
"Model: Unknown (unknown)".
* include/parted/device.h (PedDeviceType): Add PED_DEVICE_MD.
* libparted/arch/linux.c (MD_MAJOR): New define.
* libparted/arch/linux.c (_device_probe_type): Recognize MD array.
* libparted/arch/linux.c (linux_new): Handle MD array.
* parted/parted.c(do_print): Add "md" to the list of available
transports.
* NEWS (New features): Mention it.
Jim Meyering [Tue, 13 Apr 2010 19:17:50 +0000 (21:17 +0200)]
tests: propagate recent improvements to copied-from code
* tests/t1700-ext-probe.sh: Fix typo s/2>1/2>&1/ and use
a tighter regexp also in the preceding loop.
Noted by Colin Watson.
Colin Watson [Mon, 12 Apr 2010 09:16:21 +0000 (10:16 +0100)]
maint: constify ped_device_get_constraint interface
* include/parted/device.h (ped_device_get_constraint): Constify "dev"
parameter.
* libparted/device.c (ped_device_get_constraint): Likewise.
Colin Watson [Mon, 12 Apr 2010 23:11:13 +0000 (00:11 +0100)]
libparted: uninit_bg and flex_bg features should indicate ext4
According to tune2fs(8), the uninit_bg and flex_bg features are only
supported by ext4, so libparted should treat their presence as
indicating ext4. Reported by C de-Avillez in
https://bugs.launchpad.net/ubuntu/+source/parted/+bug/561599.
* libparted/fs/ext2/ext2_fs.h: Define EXT4_FEATURE_RO_COMPAT_GDT_CSUM
and EXT4_FEATURE_INCOMPAT_FLEX_BG.
* libparted/fs/ext2/interface.c (_ext2_generic_probe): Test for
EXT4_FEATURE_RO_COMPAT_GDT_CSUM and EXT4_FEATURE_INCOMPAT_FLEX_BG when
probing for ext4.
* tests/t1700-ext-probe.sh: Test for this.
* NEWS (Bug fixes): Mention this.
Colin Watson [Fri, 9 Apr 2010 12:22:27 +0000 (13:22 +0100)]
libparted: fix diagnostics about exceeding partition limits
* libparted/labels/pt-tools.c (ptt_partition_max_start_len): Use the
correct value in the diagnostics that print the upper bound on
partition size and largest starting sector number. Before, each
would use UINT32_MAX. Now each prints the partition-table-specific
limit.
Hans de Goede [Tue, 6 Apr 2010 13:57:17 +0000 (15:57 +0200)]
libparted: add test for /dev/mapper symlink issue
Sometimes, libparted operates on device mapper files with a path of
/dev/mapper/foo. With newer lvm versions /dev/mapper/foo is a symlink
to /dev/dm-#. However some storage administration programs (anaconda,
for example) may do the following:
1) Create a ped_device for /dev/mapper/foo
2) ped_get_device resolves the symlink to /dev/dm-#, and the path
in the PedDevice struct points to /dev/dm-#
3) The program does some things to lvm, causing the symlink to
point to a different /dev/dm-# node
4) The program does something with the PedDevice, which results
in an operation on the wrong device
Newer libparted versions do not suffer from this problem, as they
do not canonicalize device names under /dev/mapper. This test checks
for this bug.
* libparted/tests/symlink.c: New test which tests for this issue.
* libparted/tests/t3000-symlink.sh: New file.
* libparted/tests/Makefile.am: Include the new files. Run the new test.
Hans de Goede [Tue, 6 Apr 2010 13:57:18 +0000 (15:57 +0200)]
libparted: don't canonicalize /dev/mapper paths
Besides fixing the issue displayed by libparted/tests/symlink.c,
this has the added advantage that the output of parted p on one of these
devices now says:
"Disk /dev/mapper/BigVol2-lv_iscsi_disk2: 34.4GB"
Which is a lot more user friendly then the output before this patch:
"Disk /dev/dm-6: 34.4GB"
* libparted/device.c (ped_device_get): Don't canonicalize names
that start with "/dev/mapper/".
* NEWS (Bug fixes): Mention it.
Thanks to Ales Kozumplik for the analysis.
Details in <http://bugzilla.redhat.com/577824>.
Colin Watson [Tue, 30 Mar 2010 09:55:42 +0000 (10:55 +0100)]
doc: add @dircategory to Info documentation
The Texinfo documentation says that Info files should use @dircategory,
and suggests consulting the Free Software Directory to select a category
name. The directory places Parted under "System administration".
* doc/parted.texi: Add `@dircategory System administration'.
Jim Meyering [Tue, 30 Mar 2010 10:13:11 +0000 (12:13 +0200)]
build: update gnulib submodule to latest
Jim Meyering [Tue, 30 Mar 2010 10:23:16 +0000 (12:23 +0200)]
tests: disable new texinfo-acronym syntax-check from gnulib
* cfg.mk (local-checks-to-skip): Add new sc_texinfo_acronym, to skip it.