]> git.proxmox.com Git - systemd.git/log
systemd.git
11 years agotest-strv.c: added strv_merge_concat test
Daniel Buch [Wed, 13 Feb 2013 15:13:37 +0000 (16:13 +0100)]
test-strv.c: added strv_merge_concat test

11 years agotest-strv.c: added strv_merge test
Daniel Buch [Wed, 13 Feb 2013 15:13:36 +0000 (16:13 +0100)]
test-strv.c: added strv_merge test

11 years agoTODO
Lennart Poettering [Wed, 13 Feb 2013 18:35:28 +0000 (19:35 +0100)]
TODO

11 years agoselinux: use a dynamically sized string when generating audit messages
Lennart Poettering [Wed, 13 Feb 2013 18:35:08 +0000 (19:35 +0100)]
selinux: use a dynamically sized string when generating audit messages

https://bugzilla.redhat.com/show_bug.cgi?id=883043

11 years agouse streq instead of strcmp
Thomas Hindoe Paaboel Andersen [Wed, 13 Feb 2013 17:13:22 +0000 (18:13 +0100)]
use streq instead of strcmp

11 years agobootctl: always show slashes not backslashes
Kay Sievers [Wed, 13 Feb 2013 16:49:48 +0000 (17:49 +0100)]
bootctl: always show slashes not backslashes

11 years agoefi: unify BootXXXX reading
Kay Sievers [Wed, 13 Feb 2013 16:34:57 +0000 (17:34 +0100)]
efi: unify BootXXXX reading

11 years agoRevert "log_error() if inotify_add_watch() fails"
Lennart Poettering [Wed, 13 Feb 2013 15:25:33 +0000 (16:25 +0100)]
Revert "log_error() if inotify_add_watch() fails"

This reverts commit 2826d14091e43ed3397d862dee79d09d0115c84e.

We never should generate log messages from a library.

11 years agolog_error() if inotify_add_watch() fails
Harald Hoyer [Wed, 13 Feb 2013 14:09:36 +0000 (15:09 +0100)]
log_error() if inotify_add_watch() fails

[zj: Reworded message s/to watch/to add watch on/ to make it clear
     that it was the watch init action that failed, and not the
     "process of watching". I think this way it'll be clearer to
     people who don't know what inotify does.]

11 years agosysctl: fix uninitalized memory access in error path
Zbigniew Jędrzejewski-Szmek [Wed, 13 Feb 2013 14:55:32 +0000 (09:55 -0500)]
sysctl: fix uninitalized memory access in error path

src/sysctl/sysctl.c: In function ‘parse_file’:
src/sysctl/sysctl.c:172:41: warning: ‘property’ may be used uninitialized in this function [-Wmaybe-uninitialized]

11 years agoman: rename systemd.conf to systemd-system.conf
Zbigniew Jędrzejewski-Szmek [Mon, 11 Feb 2013 19:42:24 +0000 (14:42 -0500)]
man: rename systemd.conf to systemd-system.conf

Alias as systemd-user.conf is also provided. This should help
users running systemd in session mode.

https://bugzilla.redhat.com/show_bug.cgi?id=690868

11 years agoman: Make options consistent
William Giokas [Wed, 13 Feb 2013 09:39:22 +0000 (03:39 -0600)]
man: Make options consistent

Option listings seemed to be pretty much random, some were short opt,
long opt, others were long opt, short opt. This just makes every option
with a short and long opt that I could find in the order short opt, long
opt, for formatting's sake.

11 years agoman: Write man page for systemd-bootchart(1)
William Giokas [Wed, 13 Feb 2013 09:38:49 +0000 (03:38 -0600)]
man: Write man page for systemd-bootchart(1)

Mostly coppied from bootchart.conf(5)

11 years agotest-udev: include missing.h
Aleksander Morgado [Wed, 13 Feb 2013 07:50:15 +0000 (08:50 +0100)]
test-udev: include missing.h

We need MS_REC and MS_SHARED defined for a proper compilation, so get them from
'missing.h' if not defined already.

11 years agoratelimit: fix off-by-one
Michael Olbrich [Wed, 13 Feb 2013 07:00:46 +0000 (08:00 +0100)]
ratelimit: fix off-by-one

The current code might make sense during startup for service restart
limiting:
With burst=1 after starting num is 1, so the next ('first') restart is
still accepted ( 1 <= 1). However, once interval has expired, num is 1
after the first restart, so the second restart is also accepted.

This change is also useful in combination with watchdogs:
With burst=1 and a very large interval the system can be restarted on
the first failure.

11 years agotests: add test for continuation lines with comments
Zbigniew Jędrzejewski-Szmek [Wed, 13 Feb 2013 04:28:48 +0000 (23:28 -0500)]
tests: add test for continuation lines with comments

11 years agobuild-sys: disable tests in po/, docs/ with separate build dir
Zbigniew Jędrzejewski-Szmek [Thu, 7 Feb 2013 02:11:27 +0000 (21:11 -0500)]
build-sys: disable tests in po/, docs/ with separate build dir

Both gtk-doc and intltoolize have problems with VPATH builds.
"Creatively" disable tests when configuring from outside the
source directory.

This more-or-less reverts 9795da43c.

11 years agobootchart: Fix typos and some English
William Giokas [Wed, 13 Feb 2013 01:08:26 +0000 (19:08 -0600)]
bootchart: Fix typos and some English

bootchart.conf: direct users to `bootchart.conf(5)`, not a nonexistent
man page.

svg.c: Fix some English in the svg comment header.
  * Chrome/Chromium => Chrome, Chromium
  * firefox => Firefox
  * much more slow => more slowly

11 years agoman: typo fix
Lennart Poettering [Wed, 13 Feb 2013 03:57:36 +0000 (04:57 +0100)]
man: typo fix

11 years agoupdate TODO
Lennart Poettering [Wed, 13 Feb 2013 03:57:30 +0000 (04:57 +0100)]
update TODO

11 years agoutil: *DO NOT* loop for EINTR handling with close_nointr()
Colin Walters [Fri, 25 Jan 2013 16:21:20 +0000 (11:21 -0500)]
util: *DO NOT* loop for EINTR handling with close_nointr()

See the linked references for why we should not do this.

11 years agoboot: add stub Makefile symlink
Lennart Poettering [Wed, 13 Feb 2013 00:03:26 +0000 (01:03 +0100)]
boot: add stub Makefile symlink

11 years agoTODO: bootchart items
Auke Kok [Wed, 13 Feb 2013 00:58:00 +0000 (16:58 -0800)]
TODO: bootchart items

11 years agobootchart: make bootchart work from within the initrd
Harald Hoyer [Tue, 12 Feb 2013 23:36:29 +0000 (15:36 -0800)]
bootchart: make bootchart work from within the initrd

With this patch, bootchart can be started from within the initramfs via
the kernel command line "rdinit=/usr/lib/systemd/systemd-bootchart"

see:
    http://harald.fedorapeople.org/downloads/bootchart-20130207-1652.svg

11 years agojournal: Don't use loginuid if it's not valid
Colin Walters [Tue, 12 Feb 2013 17:24:30 +0000 (12:24 -0500)]
journal: Don't use loginuid if it's not valid

Code above this attempted to load loginuid, if this failed for
whatever reason, we'd still end up using that value (0) in place of
realuid.  Fix this by setting a bool when we know the loginuid is
valid.

This fixes journal messages showing up in per-user journals in
gnome-ostree (not configured with loginuid, but I'll shortly fix
that).

11 years agoutil: introduce strcaseeq/strncaseeq
Thomas Hindoe Paaboel Andersen [Tue, 12 Feb 2013 20:47:37 +0000 (21:47 +0100)]
util: introduce strcaseeq/strncaseeq

11 years agouse strneq instead of strncmp
Thomas Hindoe Paaboel Andersen [Tue, 12 Feb 2013 20:47:36 +0000 (21:47 +0100)]
use strneq instead of strncmp

11 years agoMerge branch 'master' of ssh://git.freedesktop.org/git/systemd/systemd into work
Auke Kok [Tue, 12 Feb 2013 23:28:00 +0000 (15:28 -0800)]
Merge branch 'master' of ssh://git.freedesktop.org/git/systemd/systemd into work

11 years agoefi: add missing #include
Lennart Poettering [Tue, 12 Feb 2013 23:18:29 +0000 (00:18 +0100)]
efi: add missing #include

11 years agomanager: pass environment over daemon-reexec
Oleksii Shevchuk [Mon, 11 Feb 2013 23:14:39 +0000 (01:14 +0200)]
manager: pass environment over daemon-reexec

Fixes this bug:
alxchk > systemctl --user set-environment A=B
alxchk > systemctl --user show-environment | grep ^A=
A=B
alxchk > systemctl --user daemon-reexec
alxchk > systemctl --user show-environment | grep ^A=
alxchk >

11 years agosystemd-bootchart: relicense to LGPLv2.1+
Auke Kok [Tue, 12 Feb 2013 23:14:15 +0000 (15:14 -0800)]
systemd-bootchart: relicense to LGPLv2.1+

When we merged systemd-bootchart we omitted relicensing this
to LGPL. Now that code is being used for other purposes we
need to make sure the code is shareable within systemd
without further problems.

The original committers and contributors of bootchart before it
was merged all have agreed with the relicensing.

11 years agoupdate TODO
Lennart Poettering [Tue, 12 Feb 2013 23:12:40 +0000 (00:12 +0100)]
update TODO

11 years agoefi: various cleanups
Lennart Poettering [Tue, 12 Feb 2013 23:07:55 +0000 (00:07 +0100)]
efi: various cleanups

11 years agomissing: define MS_REC and MS_SHARED if not defined already
Aleksander Morgado [Tue, 12 Feb 2013 13:26:17 +0000 (14:26 +0100)]
missing: define MS_REC and MS_SHARED if not defined already

11 years agoudev: use unique names for temporary files created in /dev
Kay Sievers [Tue, 12 Feb 2013 15:03:45 +0000 (16:03 +0100)]
udev: use unique names for temporary files created in /dev

On Tue, Feb 12, 2013 at 2:18 PM, Robert Milasan <rmilasan@suse.com> wrote:
> Under some circumstances udev mixed with multipath fails:
>
> udevd-work[1376]:
> symlink(../../sdk, /dev/disk/by-id/scsi-36005076305ffc0670000000000002842.udev-tmp)
> failed: File exists udevd-work[1432]:
> rename(/dev/disk/by-id/scsi-36005076305ffc0850000000000000a88.udev-tmp, /dev/disk/by-id/scsi-36005076305ffc0850000000000000a88)
> failed: No such file or directory
>
> This is non-fatal, but there is no point of created the symlink or
> renaming the symlink if it already exists.
>
> Reference: https://bugzilla.novell.com/show_bug.cgi?id=791503

It looke like this now:
  stat("/dev/disk/by-id", {st_mode=S_IFDIR|0755, st_size=80, ...}) = 0
  symlink("../../sda", "/dev/disk/by-id/ata-INTEL...N.tmp-b8:0") = 0
  rename("/dev/disk/by-id/ata-INTEL...N.tmp-b8:0", "/dev/disk/by-id/ata-INTEL...N") = 0

11 years agocdrom_id: add data track count for bad virtual drive implementations
Harald Hoyer [Tue, 12 Feb 2013 14:19:38 +0000 (15:19 +0100)]
cdrom_id: add data track count for bad virtual drive implementations

/# /lib/udev/cdrom_id --debug /dev/sr0
probing: '/dev/sr0'
INQUIRY: [AMI     ][Virtual CDROM   ][1.00]
GET CONFIGURATION failed with SK=5h/ASC=20h/ACQ=00h
drive is pre-MMC2 and does not support 46h get configuration command
trying to work around the problem
READ DISC INFORMATION failed with SK=5h/ASC=20h/ACQ=00h
no current profile, but disc is present; assuming CD-ROM
READ TOC: len: 12, start track: 1, end track: 1
last track 1 starts at block 0
READ DISC INFORMATION failed with SK=5h/ASC=20h/ACQ=00h
ID_CDROM=1
ID_CDROM_MEDIA=1
ID_CDROM_MEDIA_CD=1

What is missing here is ID_CDROM_MEDIA_TRACK_COUNT_DATA to trigger
blkid in /lib/udev/rules.d/60-persistent-storage.rules

KERNEL=="sr*", ENV{DISK_EJECT_REQUEST}!="?*",
ENV{ID_CDROM_MEDIA_TRACK_COUNT_DATA}=="?*",
ENV{ID_CDROM_MEDIA_SESSION_LAST_OFFSET}=="", \
  IMPORT{builtin}="blkid --noraid"

11 years agolocalectl: fix --help text
Lennart Poettering [Mon, 11 Feb 2013 23:34:39 +0000 (00:34 +0100)]
localectl: fix --help text

https://bugs.freedesktop.org/show_bug.cgi?id=60595

11 years agojournalctl: allow both "-n 55" and "-n55" on the command line, as equivalent syntaxes
Lennart Poettering [Mon, 11 Feb 2013 23:31:13 +0000 (00:31 +0100)]
journalctl: allow both "-n 55" and "-n55" on the command line, as equivalent syntaxes

https://bugs.freedesktop.org/show_bug.cgi?id=60596

11 years agomanager: clean environment before passing it on to others
Lennart Poettering [Mon, 11 Feb 2013 22:53:14 +0000 (23:53 +0100)]
manager: clean environment before passing it on to others

11 years agobinfmt,tmpfiles,modules-load,sysctl: rework the various early-boot services that...
Lennart Poettering [Mon, 11 Feb 2013 22:48:36 +0000 (23:48 +0100)]
binfmt,tmpfiles,modules-load,sysctl: rework the various early-boot services that work on .d/ directories

This unifies much of the logic behind them:

- All four will now ofllow the rule that the earlier file and earlier
  assignment in the .d/ directories wins. Before, sysctl was the only
  outlier, where the later setting always won.

- All four now support getopt() and --help on the command line.

- All four can now handle specification of configuration file names on
  the command line to apply. The tools will automatically find them, and
  apply them. Previously only tmpfiles could do that. This is useful for
  %post scripts in RPMs and suchlike.

- This fixes various error path issues in conf_files_list()

11 years agocore: properly validate environment data from Environment= lines in unit files
Lennart Poettering [Mon, 11 Feb 2013 22:45:59 +0000 (23:45 +0100)]
core: properly validate environment data from Environment= lines in unit files

11 years agomanager: validate environment parameters for SetEnvironment(), UnsetEnvironment(...
Lennart Poettering [Mon, 11 Feb 2013 22:41:15 +0000 (23:41 +0100)]
manager: validate environment parameters for SetEnvironment(), UnsetEnvironment() bus calls

11 years agobootctl: print partition uuids as /dev/disk/by-partuuid/...
Kay Sievers [Mon, 11 Feb 2013 22:19:31 +0000 (23:19 +0100)]
bootctl: print partition uuids as /dev/disk/by-partuuid/...

11 years agocore: don't accept invalid environment assignments from the kernel cmdline
Lennart Poettering [Mon, 11 Feb 2013 04:51:14 +0000 (05:51 +0100)]
core: don't accept invalid environment assignments from the kernel cmdline

11 years agotests: more tests
Thomas Hindoe Paaboel Andersen [Mon, 11 Feb 2013 20:40:59 +0000 (21:40 +0100)]
tests: more tests

tests for:
 parse_boolean
 env_name_is_valid

Also convert assert to assert_se to make sure they are run even with
optimization. (see commit 9b5d6bd909855543cba75a4469bff6f82845cf0d)

11 years agologind: Allow active sessions to always shutdown by default
Adel Gadllah [Wed, 6 Feb 2013 13:24:56 +0000 (14:24 +0100)]
logind: Allow active sessions to always shutdown by default

Currently local user are being asked for an admin  password when another
user is logged into the system. This does not make sense as the user has
the power to shut down the system anyway regardless if he/she knows the
password or not (by pulling the plug, battery or whatever).

So only require the admin authentification for remote sessions.

https://bugzilla.gnome.org/show_bug.cgi?id=693385

11 years agobootctl: add boot loader and firmware interface tool
Kay Sievers [Fri, 8 Feb 2013 16:24:43 +0000 (17:24 +0100)]
bootctl: add boot loader and firmware interface tool

11 years agobuild-sys: add -U_FORTIFY_SOURCE to $PYTHON_CFLAGS when necessary
Zbigniew Jędrzejewski-Szmek [Mon, 11 Feb 2013 06:08:30 +0000 (01:08 -0500)]
build-sys: add -U_FORTIFY_SOURCE to $PYTHON_CFLAGS when necessary

This patch only adds one line, but moves python detection
after cflags detection, so the result of the latter can
be used in the former.

$PYTHON_CFLAGS usually includes -D_FORTIFY_SOURCE, which will generate
a warning when compiling without optimization. Avoid by undefining
_FORTIFY_SOURCE.

11 years agoefi: fix returned length of efi_get_variable()
Lennart Poettering [Mon, 11 Feb 2013 05:20:55 +0000 (06:20 +0100)]
efi: fix returned length of efi_get_variable()

11 years agoutil: rework load_env_file()
Lennart Poettering [Mon, 11 Feb 2013 04:36:54 +0000 (05:36 +0100)]
util: rework load_env_file()

Inner library calls should not invoke log_oom(), that's something for
main programs, not library calls.

Don't read through uninitialized memory if a file ends in a continuation
line.

Add comments for the non-obvious bits.

Don't choke on comment lines that are continuation lines.

Simplify some things.

11 years agopath-util: fix memory leak
Lennart Poettering [Mon, 11 Feb 2013 04:10:23 +0000 (05:10 +0100)]
path-util: fix memory leak

11 years agoutil: introduce FOREACH_LINE for iterating through files
Lennart Poettering [Mon, 11 Feb 2013 04:09:29 +0000 (05:09 +0100)]
util: introduce FOREACH_LINE for iterating through files

11 years agotest-env-replace: better user assert_se() everywhere
Lennart Poettering [Mon, 11 Feb 2013 02:52:32 +0000 (03:52 +0100)]
test-env-replace: better user assert_se() everywhere

For test tools it's better to use assert_se() rather than assert(),
since the former is not optimized away, even if -DNDEBUG is used. That
means the test program now retains its usefulness even if -DNDEBUG is
used.

This also adds in some OOM checks, to be extra anal...

11 years agoenv: considerably beef up environment cleaning logic
Lennart Poettering [Mon, 11 Feb 2013 02:46:08 +0000 (03:46 +0100)]
env: considerably beef up environment cleaning logic

Now, actually check if the environment variable names and values used
are valid, before accepting them. With this in place are at some places
more rigid than POSIX, and less rigid at others. For example, this code
allows lower-case environment variables (which POSIX suggests not to
use), but it will not allow non-UTF8 variable values.

All in all this should be a good middle ground of what to allow and what
not to allow as environment variables.

(This also splits out all environment related calls into env-util.[ch])

11 years agotests: add tests of strxcpyx
Thomas Hindoe Paaboel Andersen [Sun, 10 Feb 2013 20:56:03 +0000 (21:56 +0100)]
tests: add tests of strxcpyx

also fix a bad indentation in test-strbug.c

11 years agoman: typo fix in systemd.socket
Michael Biebl [Sat, 9 Feb 2013 13:09:05 +0000 (14:09 +0100)]
man: typo fix in systemd.socket

11 years agoupdate TODO
Lennart Poettering [Sat, 9 Feb 2013 01:47:42 +0000 (02:47 +0100)]
update TODO

11 years agovconsole-setup: don't set the kbd mode to unicode if is is currently in raw/off mode
Lennart Poettering [Sat, 9 Feb 2013 01:40:57 +0000 (02:40 +0100)]
vconsole-setup: don't set the kbd mode to unicode if is is currently in raw/off mode

Let's not confuse X11

11 years agoshutdown: issue a sync() as soon as shutdown.target is queued
Lennart Poettering [Fri, 25 Jan 2013 21:33:33 +0000 (22:33 +0100)]
shutdown: issue a sync() as soon as shutdown.target is queued

11 years agotests: add test of strbuf
Thomas Hindoe Paaboel Andersen [Sat, 9 Feb 2013 00:39:44 +0000 (01:39 +0100)]
tests: add test of strbuf

11 years agologind: rename "seat-master" tag to "master-of-seat"
Lennart Poettering [Sat, 9 Feb 2013 00:25:55 +0000 (01:25 +0100)]
logind: rename "seat-master" tag to "master-of-seat"

We currently enforce that seats are to be named in the form of
"seatXXX", i.e. need to begin with the 4 characters "seat". Thus,
"seat-master" would qualify as a seat name. As seat names are frequently
used as tags on devices, the "seat-master" tag might hence confuse
logind if the user decides to name a seat "seat-master".

Hence, avoid any ambuigity: let's rename the "seat-master" tag to
"master-of-seat".

11 years agoupdate TODO
Lennart Poettering [Sat, 9 Feb 2013 00:17:36 +0000 (01:17 +0100)]
update TODO

11 years agoman: fix prototype of sd_get_uids()
Lennart Poettering [Fri, 8 Feb 2013 22:31:13 +0000 (23:31 +0100)]
man: fix prototype of sd_get_uids()

11 years agoupdate TODO
Lennart Poettering [Fri, 8 Feb 2013 22:25:55 +0000 (23:25 +0100)]
update TODO

11 years agobuild-sys: link libudev against the internal sd-daemon lib, not the source file
Kay Sievers [Fri, 8 Feb 2013 12:25:26 +0000 (13:25 +0100)]
build-sys: link libudev against the internal sd-daemon lib, not the source file

11 years agoudev: udevadm hwdb - add --root=
Kay Sievers [Fri, 8 Feb 2013 11:00:05 +0000 (12:00 +0100)]
udev: udevadm hwdb - add --root=

https://bugs.freedesktop.org/show_bug.cgi?id=60398

11 years agoshared: conf-files - use root parameter and convert to auto-cleanup
Kay Sievers [Fri, 8 Feb 2013 09:45:30 +0000 (10:45 +0100)]
shared: conf-files - use root parameter and convert to auto-cleanup

11 years agoshared: conf-files - add root parameter
Kay Sievers [Fri, 8 Feb 2013 09:22:02 +0000 (10:22 +0100)]
shared: conf-files - add root parameter

11 years agosystemd: fix uninitialized memory access in selinux
Zbigniew Jędrzejewski-Szmek [Fri, 8 Feb 2013 01:58:02 +0000 (20:58 -0500)]
systemd: fix uninitialized memory access in selinux

==6129== Syscall param socketcall.getsockopt(optlen) points to uninitialised byte(s)
==6129== at 0x312D0F325A: getsockopt (syscall-template.S:81)
==6129== by 0x4615E6: get_audit_data (selinux-access.c:268)
==6129== by 0x461A81: selinux_access_check (selinux-access.c:394)
==6129== by 0x453B23: bus_manager_message_handler (dbus-manager.c:1395)
==6129== by 0x313581D9C4: _dbus_object_tree_dispatch_and_unlock (dbus-object-tree.c:862)
==6129== by 0x313580FC1F: dbus_connection_dispatch (dbus-connection.c:4672)
==6129== by 0x44CD4C: bus_dispatch (dbus.c:528)
==6129== by 0x41292F: manager_loop (manager.c:1573)
==6129== by 0x40CD8E: main (main.c:1735)
==6129== Address 0x7fefff2c0 is on thread 1's stack

11 years agotests: add more tests
Thomas Hindoe Paaboel Andersen [Thu, 7 Feb 2013 23:22:26 +0000 (00:22 +0100)]
tests: add more tests

Adds tests of:
 strv_env_delete
 strv_env_unset
 strv_env_set

11 years agobuild-sys: fix build without -Wl,--gc-sections
Zbigniew Jędrzejewski-Szmek [Thu, 7 Feb 2013 21:22:29 +0000 (16:22 -0500)]
build-sys: fix build without -Wl,--gc-sections

In 8d81eb8e the new library was added for libudev, use it for
libsystemd-journal too.

11 years agoudev: remove HWDB_BIN
Kay Sievers [Thu, 7 Feb 2013 20:53:10 +0000 (21:53 +0100)]
udev: remove HWDB_BIN

11 years agotest-env-replace: assert results instead of printing them
Thomas Hindoe Paaboel Andersen [Thu, 7 Feb 2013 20:10:31 +0000 (21:10 +0100)]
test-env-replace: assert results instead of printing them

Also code cleanup and split into functions in preparation for more tests
to come.

11 years agobuild-sys: at least for now, never link libudev against systemd's shared libraries
Kay Sievers [Thu, 7 Feb 2013 12:47:46 +0000 (13:47 +0100)]
build-sys: at least for now, never link libudev against systemd's shared libraries

11 years agobuild-sys: add libsystemd-id128 as libudev dependency
David Strauss [Thu, 7 Feb 2013 06:23:59 +0000 (01:23 -0500)]
build-sys: add libsystemd-id128 as libudev dependency

After d848b9cbfa0 'Move generic specifier functions to shared' libudev
depends (through) libsystemd-shared.la on libsystemd-id128.so.  The
problem only appears when the linker does not support --gc-sections
and manifests itself as the inability to resolve sd_id128_get_machine
and other libsystemd-id128 functions, which aren't really used.

11 years agobuild-sys: restore man/index.html as link to systemd.index.html
Zbigniew Jędrzejewski-Szmek [Thu, 7 Feb 2013 06:10:21 +0000 (01:10 -0500)]
build-sys: restore man/index.html as link to systemd.index.html

It got lost in the transformation to XML generation.

11 years agoman: fix return value for sd_journal_flush_matches
Zbigniew Jędrzejewski-Szmek [Thu, 7 Feb 2013 05:53:11 +0000 (00:53 -0500)]
man: fix return value for sd_journal_flush_matches

https://bugs.freedesktop.org/show_bug.cgi?id=60133

11 years agoman: document COREDUMP_UNIT and COREDUMP_USER_UNIT
Zbigniew Jędrzejewski-Szmek [Thu, 7 Feb 2013 05:50:37 +0000 (00:50 -0500)]
man: document COREDUMP_UNIT and COREDUMP_USER_UNIT

Also fix formatting for kernel related fields, enabling them
to show up in the directives index.

11 years agocoredump: make use of the cleanup macros
Mirco Tischler [Tue, 5 Feb 2013 00:43:10 +0000 (01:43 +0100)]
coredump: make use of the cleanup macros

11 years agologs-show: show messages for all unit types in systemctl status
Mirco Tischler [Mon, 4 Feb 2013 14:13:24 +0000 (15:13 +0100)]
logs-show: show messages for all unit types in systemctl status

I can't find a reason why we shouldn't try to output messages for other
unit types than .service, .socket, .mount and .swap as well. It's probably
a leftover from before we started logging UNIT= from inside PID 1.

11 years agojournal: log user units for coredumps and show them in systemctl status
Mirco Tischler [Mon, 4 Feb 2013 14:13:23 +0000 (15:13 +0100)]
journal: log user units for coredumps and show them in systemctl status

11 years agonss-myhostname: copy first result to preallocated buffer
Eelco Dolstra [Mon, 4 Feb 2013 14:56:26 +0000 (15:56 +0100)]
nss-myhostname: copy first result to preallocated buffer

Fixes a segfault in nscd when using nss-myhostname.

Nscd expects that an NSS module's gethostbyname4_r function returns
its first result in the pre-allocated gaih_addrtuple denoted by **pat.
(See nscd/aicache.c in the Glibc sources.)  However, nss-myhostname
doesn't fill in **pat but allocates the first result in ‘buffer’, then
sets *pat.  So nscd crashes (e.g. when running ‘getent ahosts
my-machine’).

Hard to tell if this is a bug in nscd, since there doesn't seem to be
a proper API spec for gethostbyname4_r.  But in any case, this patch
fixes the crash by copying the first result to **pat.

11 years agoman: describe unit load path in systemd.unit(5)
Zbigniew Jędrzejewski-Szmek [Mon, 4 Feb 2013 03:23:53 +0000 (22:23 -0500)]
man: describe unit load path in systemd.unit(5)

In other cases where multiple directories are searched for unit files,
the list of directories is described in the man page describing the
format. I think this makes sense too in case of systemd directories,
since the systemd(1) manpage already has an overview of many different
topics.

11 years agobuild-sys: substitute strings in systemd.unit(5)
Zbigniew Jędrzejewski-Szmek [Tue, 5 Feb 2013 15:31:20 +0000 (10:31 -0500)]
build-sys: substitute strings in systemd.unit(5)

Makefile.am is updated to deal with .xml.in sources.  Nothing in the
output is really changed yet, this is just preparation.

11 years agosystemd: raise level of message about lookup paths
Zbigniew Jędrzejewski-Szmek [Sat, 2 Feb 2013 15:36:11 +0000 (10:36 -0500)]
systemd: raise level of message about lookup paths

This can be pretty important for the user, and is not trivial
to figure out in all cases.

Also show failing path in error messages.

11 years agoshared: remove now unused function
Zbigniew Jędrzejewski-Szmek [Sat, 2 Feb 2013 21:12:33 +0000 (16:12 -0500)]
shared: remove now unused function

11 years agosystemd: do not remove empty paths from unit lookup path
Zbigniew Jędrzejewski-Szmek [Thu, 7 Feb 2013 03:49:19 +0000 (22:49 -0500)]
systemd: do not remove empty paths from unit lookup path

The ability to start a new unit with 'systemctl start ...' should not
depend on whether there are other units in the directory. Previously,
an additional 'systemctl daemon-reload' would be necessary to tell
systemd to update the list of unit lookup paths.

11 years agobuild-sys: create Makefile-man.am automatically
Zbigniew Jędrzejewski-Szmek [Sun, 3 Feb 2013 03:47:47 +0000 (22:47 -0500)]
build-sys: create Makefile-man.am automatically

man rules were repeating the same information in too many places,
which was error prone. Those rules can be easily generated from .xml
files. For efficiency and because python is not a required dependency,
Makefile-man.am is only regenerated when requested with

  make update-man-list

If no metadata in man/*.xml changed, this file should not change. So
only when a new man page or a new alias is added, this file should
show up in 'git diff'. The change should then be committed.

If the support for building from git without python was dropped, we
could drop Makefile-man.am from version control. This would also
increase the partial build time (since more stuff would be rebuild
whenever sources in man/*.xml would be modified), so it would probably
wouldn't be worth it.

11 years agobuild-sys: split man alias rules into separate file
Zbigniew Jędrzejewski-Szmek [Sun, 3 Feb 2013 03:38:33 +0000 (22:38 -0500)]
build-sys: split man alias rules into separate file

There shouldn't be any functional change. This is an 'automake include',
so the generated Makefile is the same (apart from ordering).

11 years agobuild-sys: replace one last mkdir hook with list item
Zbigniew Jędrzejewski-Szmek [Sun, 3 Feb 2013 03:48:38 +0000 (22:48 -0500)]
build-sys: replace one last mkdir hook with list item

Previously, errors would be ignored. Now they are not. But I don't see
how we could fail to create the directory, so I don't think that it
matters.

11 years agoman: see also dracut(7) in bootup(7)
Zbigniew Jędrzejewski-Szmek [Thu, 31 Jan 2013 20:52:13 +0000 (15:52 -0500)]
man: see also dracut(7) in bootup(7)

11 years agotests: skip tests if manager cannot be created
Zbigniew Jędrzejewski-Szmek [Wed, 6 Feb 2013 22:35:53 +0000 (17:35 -0500)]
tests: skip tests if manager cannot be created

When running without a user session, tests fail.

11 years agotests: move strv_parse_nulstr to test-strv, and actually test it
Thomas Hindoe Paaboel Andersen [Wed, 6 Feb 2013 23:33:58 +0000 (00:33 +0100)]
tests: move strv_parse_nulstr to test-strv, and actually test it

The test was originally added with this commit message:
"ask-password: supported plymouth cached passwords"

11 years agostrv: add strv_print
Thomas Hindoe Paaboel Andersen [Wed, 6 Feb 2013 23:15:27 +0000 (00:15 +0100)]
strv: add strv_print

Clearer, and spares the temp variable.

11 years agotests: run manager in session mode
Zbigniew Jędrzejewski-Szmek [Wed, 6 Feb 2013 22:03:12 +0000 (17:03 -0500)]
tests: run manager in session mode

False positives pop up otherwise.

FAIL: test-unit-name (exit: 134)
================================

Failed to open /dev/tty0: Permission denied
Failed to create root cgroup hierarchy: Permission denied
Assertion 'manager_new(SYSTEMD_SYSTEM, &m) == 0' failed at src/test/test-unit-name.c:125, function test_unit_printf(). Aborting.

11 years agotest: add a few tests and tidy up
Thomas Hindoe Paaboel Andersen [Wed, 6 Feb 2013 20:15:23 +0000 (21:15 +0100)]
test: add a few tests and tidy up

adds test of:
strv_find
strv_find_prefix
strv_overlap
strv_sort
streq_ptr
first_word

Splits tests of util.c into own file to avoid clutter as we add more.

Removed a few prints and uses _cleanup_free_ to make the tests more focused.

11 years agoman: remove boot up runlevel support
Umut Tezduyar [Fri, 1 Feb 2013 12:20:49 +0000 (13:20 +0100)]
man: remove boot up runlevel support

As of 3cdebc217c42c8529086f2965319b6a48eaaeabe support for
distro specific boot up runlevels has been dropped.

11 years agoupdate TODO
Lennart Poettering [Mon, 4 Feb 2013 21:27:32 +0000 (22:27 +0100)]
update TODO

11 years agoupdate TODO
Lennart Poettering [Wed, 30 Jan 2013 01:33:16 +0000 (02:33 +0100)]
update TODO

11 years agohwdb: update
Kay Sievers [Mon, 4 Feb 2013 17:55:10 +0000 (18:55 +0100)]
hwdb: update