chiark / gitweb /
6 years agotools: add gdb command to dump hashmap information
Michal Schmidt [Tue, 14 Oct 2014 23:31:23 +0000 (01:31 +0200)]
tools: add gdb command to dump hashmap information

$ sudo gdb -p 1
(gdb) source
(gdb) sd_dump_hashmaps
  ... lists allocated hashmaps ...
(gdb) sd_dump_hashmaps 1
  ... lists allocated hashmaps, their DIB histograms and contiguous
      blocks statistics ...

6 years agohashmap: rewrite the implementation
Michal Schmidt [Tue, 14 Oct 2014 23:27:16 +0000 (01:27 +0200)]
hashmap: rewrite the implementation

This is a rewrite of the hashmap implementation. Its advantage is lower
memory usage.

It uses open addressing (entries are stored in an array, as opposed to
linked lists). Hash collisions are resolved with linear probing and
Robin Hood displacement policy. See the references in hashmap.c.

Some fun empirical findings about hashmap usage in systemd on my laptop:
  - 98 % of allocated hashmaps are Sets.
  - Sets contain 78 % of all entries, plain Hashmaps 17 %, and
    OrderedHashmaps 5 %.
  - 60 % of allocated hashmaps contain only 1 entry.
  - 90 % of allocated hashmaps contain 5 or fewer entries.
  - 75 % of all entries are in hashmaps that use trivial_hash_ops.

Clearly it makes sense to:
  - store entries in distinct entry types. Especially for Sets - their
    entries are the most numerous and they require the least information
    to store an entry.
  - have a way to store small numbers of entries directly in the hashmap
    structs, and only allocate the usual entry arrays when the direct
    storage is full.

The implementation has an optional debugging feature (enabled by
defining the ENABLE_HASHMAP_DEBUG macro), where it:
  - tracks all allocated hashmaps in a linked list so that one can
    easily find them in gdb,
  - tracks which function/line allocated a given hashmap, and
  - checks for invalid mixing of hashmap iteration and modification.

Since entries are not allocated one-by-one anymore, mempools are not
used for entries. Originally I meant to drop mempools entirely, but it's
still worth it to use them for the hashmap structs. My testing indicates
that it makes loading of units about 5 % faster (a test with 10000 units
where more than 200000 hashmaps are allocated - pure malloc: 449±4 ms,
mempools: 427±7 ms).

Here are some memory usage numbers, taken on my laptop with a more or
less normal Fedora setup after booting with SELinux disabled (SELinux
increases systemd's memory usage significantly):

systemd (PID 1)                            Original   New    Change
dirty memory (from pmap -x 1) [KiB]            2152  1264     -41 %
total heap allocations (from gdb-heap) [KiB]   1623   756     -53 %

6 years agotest: adjust max load factor in test_hashmap_many()
Michal Schmidt [Wed, 15 Oct 2014 09:28:23 +0000 (11:28 +0200)]
test: adjust max load factor in test_hashmap_many()

A reimplementation of hashmaps will follow and it will use 0.8.

6 years agomempool: add a zeroing alloc function
Michal Schmidt [Fri, 24 Oct 2014 13:30:18 +0000 (15:30 +0200)]
mempool: add a zeroing alloc function

Add mempool_alloc0_tile(). It's like mempool_alloc_tile(), but it
initializes the allocated tile's memory to zero.

6 years agoutil: add log2u(), log2u_round_up()
Michal Schmidt [Tue, 14 Oct 2014 23:28:54 +0000 (01:28 +0200)]
util: add log2u(), log2u_round_up()

Two's logarithms for unsigned.

6 years agomemfd: rename memfd.h to memfd-util.h to avoid any confusion with any libc provided...
Lennart Poettering [Thu, 30 Oct 2014 17:32:37 +0000 (18:32 +0100)]
memfd: rename memfd.h to memfd-util.h to avoid any confusion with any libc provided headers

6 years agomemfd: always use our internal utility functions where we have them
Lennart Poettering [Thu, 30 Oct 2014 17:28:37 +0000 (18:28 +0100)]
memfd: always use our internal utility functions where we have them

6 years agomemfd: drop memfd_get_name() as it is unused
Lennart Poettering [Thu, 30 Oct 2014 17:28:01 +0000 (18:28 +0100)]
memfd: drop memfd_get_name() as it is unused

6 years agoupdate TODO
Lennart Poettering [Thu, 30 Oct 2014 16:39:29 +0000 (17:39 +0100)]
update TODO

6 years agojournal: when sending huge log messages prefer memfds over temporary files in /dev/shm
Lennart Poettering [Thu, 30 Oct 2014 16:36:02 +0000 (17:36 +0100)]
journal: when sending huge log messages prefer memfds over temporary files in /dev/shm

Previously when a log message grew beyond the maximum AF_UNIX/SOCK_DGRAM
datagram limit we'd send an fd to a deleted file in /dev/shm instead.
Because the sender could still modify the file after delivery we had to
immediately copy the data on the receiving side.

With memfds we can optimize this logic, and also remove the dependency
on /dev/shm: simply send a sealed memfd around, and if we detect the
seal memory map the fd and use it directly.

6 years agoCODING_STYLE: clarify that we really should use O_CLOEXEC everywhere
Lennart Poettering [Thu, 30 Oct 2014 16:05:25 +0000 (17:05 +0100)]
CODING_STYLE: clarify that we really should use O_CLOEXEC everywhere

6 years agomemfd: always create our memfds with CLOEXEC set
Lennart Poettering [Thu, 30 Oct 2014 15:23:34 +0000 (16:23 +0100)]
memfd: always create our memfds with CLOEXEC set

We really shouldn't create fds ever that have the flag unset.

6 years agoutil: unify how we see srand()
Lennart Poettering [Thu, 30 Oct 2014 14:35:37 +0000 (15:35 +0100)]
util: unify how we see srand()

6 years agoutil: don't block on getrandom()
Lennart Poettering [Thu, 30 Oct 2014 14:27:53 +0000 (15:27 +0100)]
util: don't block on getrandom()

6 years agosysusers: Preserve ownership and mode on /etc/passwd and friends
Colin Guthrie [Wed, 29 Oct 2014 14:03:41 +0000 (14:03 +0000)]
sysusers: Preserve ownership and mode on /etc/passwd and friends

When running sysusers we would clobber file ownership and permissions
on the files /etc/passwd, /etc/group and /etc/[g]shadow.

This simply preserves the ownership and mode if existing files are

6 years agokeymap: Ignore brightness keys on Dell Inspiron 1520 to avoid double events
Hans de Goede [Thu, 30 Oct 2014 09:15:54 +0000 (10:15 +0100)]
keymap: Ignore brightness keys on Dell Inspiron 1520 to avoid double events

On the Dell Inspiron 1520 both the atkbd and acpi-video input devices report
an event for pressing the brightness up / down key-combos, resulting in user
space seeing double events and increasing / decreasing the brightness 2 steps
for each keypress.

This hwdb snippet suppresses the atkbd events, making the Inspiron 1520 work
like most modern laptops which emit brightness up / down events through
acpi-video only.

Reported by Pavel Malyshev <>

6 years agozsh-completion: update start/restart completions
Zbigniew Jędrzejewski-Szmek [Thu, 30 Oct 2014 03:47:55 +0000 (23:47 -0400)]
zsh-completion: update start/restart completions

Now zsh should behave the same for those two subcommands as bash.

6 years agobash-completion: use improved filtering to make things faster
Zbigniew Jędrzejewski-Szmek [Thu, 30 Oct 2014 03:06:58 +0000 (23:06 -0400)]
bash-completion: use improved filtering to make things faster

6 years agosystemctl: obey --state in list-unit-files
Zbigniew Jędrzejewski-Szmek [Thu, 30 Oct 2014 02:51:00 +0000 (22:51 -0400)]
systemctl: obey --state in list-unit-files

6 years agosystemctl: let list-{units,unit-files } honour --type
Zbigniew Jędrzejewski-Szmek [Thu, 30 Oct 2014 02:46:30 +0000 (22:46 -0400)]
systemctl: let list-{units,unit-files } honour --type

The docs don't clarify what is expected, but I don't see any reason
why --type should be ignored.

Also restucture the compund conditions into separate clauses for
easier reading.

6 years agobash-completion: rework startable/restartable units once more
Zbigniew Jędrzejewski-Szmek [Thu, 30 Oct 2014 02:25:33 +0000 (22:25 -0400)]
bash-completion: rework startable/restartable units once more

I tried to use 'systemctl --all list-units' to filter unit files, but
this always filters out unit files which are not loaded. We want to complete
systemctl start with those units too, so this approach is not going to work.

New version is rather slow, but hopefully correct.

6 years agomissing.h: fix wrong __NR_getrandom syscall def
Dave Reisner [Thu, 30 Oct 2014 00:30:25 +0000 (20:30 -0400)]
missing.h: fix wrong __NR_getrandom syscall def

278 is vmsplice on x86_64. 318 is what we want:

6 years agoudev: path_id - update comments
Kay Sievers [Thu, 30 Oct 2014 00:18:34 +0000 (01:18 +0100)]
udev: path_id - update comments

6 years agobusctl: add new "capture" verb to record bus messages in libpcap compatible files...
Lennart Poettering [Thu, 30 Oct 2014 00:13:11 +0000 (01:13 +0100)]
busctl: add new "capture" verb to record bus messages in libpcap compatible files, for dissection with wireshark

6 years agonspawn: ignore EEXIST when creating mount point
Dave Reisner [Wed, 29 Oct 2014 17:32:43 +0000 (13:32 -0400)]
nspawn: ignore EEXIST when creating mount point

A combination of commits f3c80515c and 79d80fc14 cause nspawn to
silently fail with a commandline such as:

  # systemd-nspawn -D /build/extra-x86_64 --bind=/usr

strace shows the culprit:

  [pid 27868] writev(2, [{"Failed to create mount point /build/extra-x86_64/usr: File exists", 82}, {"\n", 1}], 2) = 83

6 years agosd-bus: properly handle removals of non-existing matches
Lennart Poettering [Wed, 29 Oct 2014 16:58:43 +0000 (17:58 +0100)]
sd-bus: properly handle removals of non-existing matches

6 years agomissing: no tabs please, we are british
Lennart Poettering [Wed, 29 Oct 2014 16:08:18 +0000 (17:08 +0100)]
missing: no tabs please, we are british

6 years agoupdate TODO
Lennart Poettering [Wed, 29 Oct 2014 16:08:00 +0000 (17:08 +0100)]
update TODO

6 years agomissing: simplify memfd ifdeffery
Lennart Poettering [Wed, 29 Oct 2014 16:07:47 +0000 (17:07 +0100)]
missing: simplify memfd ifdeffery

6 years agoutil: make use of the new getrandom() syscall if it is available when needing entropy
Lennart Poettering [Wed, 29 Oct 2014 16:06:32 +0000 (17:06 +0100)]
util: make use of the new getrandom() syscall if it is available when needing entropy

Doesn't require an fd, and could be a bit faster, so let's make use of
it, if it is available.

6 years agosd-rtnl: add bridge port rtnl attributes.
Susant Sahani [Mon, 21 Jul 2014 08:02:09 +0000 (13:32 +0530)]
sd-rtnl: add bridge port rtnl attributes.

Add bridge port attributes to sd-rtnl to configure
via networkd.

6 years agosd-rtnl: add support to set packet family type
Susant Sahani [Mon, 21 Jul 2014 07:45:38 +0000 (13:15 +0530)]
sd-rtnl: add support to set packet family type

This patch adds functionality to set family type
in the rtnl message for example PF_BRIDGE.

6 years agosd-rtnl: mark nested attributes with the NLA_F_NESTED flag
Tom Gundersen [Wed, 29 Oct 2014 09:52:07 +0000 (10:52 +0100)]
sd-rtnl: mark nested attributes with the NLA_F_NESTED flag

The kernel mostly does not check this, but let's be consisntent and allways set it anyway. Based
on patch from Susant Sahani.

6 years agosd-rtnl: fix reading of nla type
Tom Gundersen [Wed, 29 Oct 2014 09:50:33 +0000 (10:50 +0100)]
sd-rtnl: fix reading of nla type

We must filter out the 'network-byteorder' and 'nested' flags.

6 years agonetworkd: add Patch of MTU discovery for IPIP and GRE
Susant Sahani [Tue, 22 Jul 2014 11:21:53 +0000 (16:51 +0530)]
networkd: add Patch of MTU discovery for IPIP and GRE

Add path of MTU disovery for IPIP and GRE Kind of tunnels

6 years agonetworkd: add back path of MTU discovery for SIT
Susant Sahani [Tue, 22 Jul 2014 11:16:39 +0000 (16:46 +0530)]
networkd: add back path of MTU discovery for SIT

6 years agoNEWS: well, it's Options= now, not Discard=
Lennart Poettering [Tue, 28 Oct 2014 19:36:32 +0000 (20:36 +0100)]
NEWS: well, it's Options= now, not Discard=

6 years agobuild-sys: do not use "label" functions in libsystemd_shared v217
Kay Sievers [Tue, 28 Oct 2014 17:46:53 +0000 (18:46 +0100)]
build-sys: do not use "label" functions in libsystemd_shared

6 years agoupdate NEWS
Lennart Poettering [Tue, 28 Oct 2014 17:10:48 +0000 (18:10 +0100)]
update NEWS

6 years agotest: use assert_se() when testing tables so that we get a useful error when somethin...
Lennart Poettering [Tue, 28 Oct 2014 17:07:20 +0000 (18:07 +0100)]
test: use assert_se() when testing tables so that we get a useful error when something fails

6 years agoservice: add missing state table entry
Lennart Poettering [Tue, 28 Oct 2014 17:06:58 +0000 (18:06 +0100)]
service: add missing state table entry

6 years agocore: send sigabrt on watchdog timeout to get the stacktrace
Umut Tezduyar Lindskog [Tue, 28 Oct 2014 15:35:40 +0000 (16:35 +0100)]
core: send sigabrt on watchdog timeout to get the stacktrace

if sigabrt doesn't do the job, follow regular shutdown
routine, sigterm > sigkill.

6 years agoswap: don't add too many deps for swap files form /proc/swap
Lennart Poettering [Tue, 28 Oct 2014 16:27:38 +0000 (17:27 +0100)]
swap: don't add too many deps for swap files form /proc/swap

6 years agoudev: path_id - set supported_parent for well-known SCSI setups
Kay Sievers [Tue, 28 Oct 2014 15:50:24 +0000 (16:50 +0100)]
udev: path_id - set supported_parent for well-known SCSI setups

6 years agoNEWS: fix typos
Ronny Chevalier [Tue, 28 Oct 2014 15:04:21 +0000 (16:04 +0100)]
NEWS: fix typos

6 years agoNEWS: update
Kay Sievers [Tue, 28 Oct 2014 14:53:44 +0000 (15:53 +0100)]
NEWS: update

6 years agoNEWS
Tom Gundersen [Tue, 28 Oct 2014 14:42:57 +0000 (15:42 +0100)]

6 years agoupdate NEWS
Lennart Poettering [Tue, 28 Oct 2014 14:44:00 +0000 (15:44 +0100)]
update NEWS

6 years agoupdate NEWS
Lennart Poettering [Tue, 28 Oct 2014 14:35:35 +0000 (15:35 +0100)]
update NEWS

6 years agobuild-sys: bump version number in preparation for release
Lennart Poettering [Tue, 28 Oct 2014 14:29:49 +0000 (15:29 +0100)]
build-sys: bump version number in preparation for release

6 years agobuild-sys: bump library versions in preparation for 217 release
Lennart Poettering [Tue, 28 Oct 2014 14:29:36 +0000 (15:29 +0100)]
build-sys: bump library versions in preparation for 217 release

6 years agoNEWS: add contributor list in preparation for 217 release
Lennart Poettering [Tue, 28 Oct 2014 14:20:16 +0000 (15:20 +0100)]
NEWS: add contributor list in preparation for 217 release

6 years agoupdate NEWS
Lennart Poettering [Tue, 28 Oct 2014 14:17:53 +0000 (15:17 +0100)]
update NEWS

6 years agopolkit: actually generate new InteractiveAuthorizationRequired error on the right...
Lennart Poettering [Tue, 28 Oct 2014 14:11:39 +0000 (15:11 +0100)]
polkit: actually generate new InteractiveAuthorizationRequired error on the right occasions

6 years agoupdate TODO
Lennart Poettering [Tue, 28 Oct 2014 14:06:30 +0000 (15:06 +0100)]
update TODO

6 years agosd-bus: add support for new InteractiveAuthorizationRequired error from dbus spec
Lennart Poettering [Tue, 28 Oct 2014 13:59:08 +0000 (14:59 +0100)]
sd-bus: add support for new InteractiveAuthorizationRequired error from dbus spec

6 years agoupdate TODO
Lennart Poettering [Tue, 28 Oct 2014 13:47:35 +0000 (14:47 +0100)]
update TODO

6 years agoman: minor corrections on how the overall system states are actually defined
Lennart Poettering [Tue, 28 Oct 2014 13:38:47 +0000 (14:38 +0100)]
man: minor corrections on how the overall system states are actually defined

6 years agomac: add mac_ prefix to distinguish origin security apis
WaLyong Cho [Fri, 24 Oct 2014 12:15:25 +0000 (21:15 +0900)]
mac: add mac_ prefix to distinguish origin security apis

6 years agoupdate TODO
Lennart Poettering [Tue, 28 Oct 2014 13:31:17 +0000 (14:31 +0100)]
update TODO

6 years agoswap: drop noauto/nofail bools from Swap structure
Lennart Poettering [Tue, 28 Oct 2014 13:30:30 +0000 (14:30 +0100)]
swap: drop noauto/nofail bools from Swap structure

We nowadays always set them to "false" anyway, hence let's get rid of
them entirely.

6 years agoswap: replace Discard= setting by a more generic Options= setting
Lennart Poettering [Tue, 28 Oct 2014 13:24:46 +0000 (14:24 +0100)]
swap: replace Discard= setting by a more generic Options= setting

For now, it's systemd itself that parses the options string, but as soon
as util-linux' swapon can take the option string directly with -o we
should pass it on unmodified.

6 years agoudev hwdb: Support shipping pre-compiled database in system images
Martin Pitt [Fri, 17 Oct 2014 13:01:01 +0000 (15:01 +0200)]
udev hwdb: Support shipping pre-compiled database in system images

In some cases it is preferable to ship system images with a pre-generated
binary hwdb database, to avoid having to build it at runtime, avoid shipping
the source hwdb files, or avoid storing large binary files in /etc.

So if hwdb.bin does not exist in /etc/udev/, fall back to looking for it in
UDEVLIBEXECDIR. This keeps the possibility to add files to /etc/udev/hwdb.d/
and re-generating the database which trumps the one in /usr/lib.

Add a new --usr flag to "udevadm hwdb --update" which puts the database

Adjust systemd-udev-hwdb-update.service to not generate the file in /etc if we
already have it in /usr.

6 years agoupdate TODO
Lennart Poettering [Tue, 28 Oct 2014 11:33:25 +0000 (12:33 +0100)]
update TODO

6 years agoman: document that $XDG_SESSION_DESKTOP only takes a single item, not a list, unlike...
Lennart Poettering [Tue, 28 Oct 2014 11:32:25 +0000 (12:32 +0100)]
man: document that $XDG_SESSION_DESKTOP only takes a single item, not a list, unlike $XDG_CURRENT_DESKTOP

6 years agosd-login: let's not needlessly yell at users
Lennart Poettering [Tue, 28 Oct 2014 11:31:11 +0000 (12:31 +0100)]
sd-login: let's not needlessly yell at users

While GNOME/KDE are generally capitalized, systemd tools generally are
not, hence let's not start doing so in the XDG_CURRENT_SESSION
environment variable.

6 years agomanager: print warning on console before reboot
Zbigniew Jędrzejewski-Szmek [Tue, 28 Oct 2014 03:14:23 +0000 (23:14 -0400)]
manager: print warning on console before reboot

It will be printed even if a prompt is blocking other messages.

6 years agomanager: convert ephemeral to enum
Zbigniew Jędrzejewski-Szmek [Tue, 28 Oct 2014 03:02:54 +0000 (23:02 -0400)]
manager: convert ephemeral to enum

In preparation for subsequent changes.

6 years agoman: add table of manager states
Zbigniew Jędrzejewski-Szmek [Tue, 28 Oct 2014 02:16:11 +0000 (22:16 -0400)]
man: add table of manager states

6 years agoshell-completion: systemctl set-default,get-default,is-system-running
Zbigniew Jędrzejewski-Szmek [Tue, 28 Oct 2014 02:34:28 +0000 (21:34 -0500)]
shell-completion: systemctl set-default,get-default,is-system-running

6 years agorpm: add user macros
Zbigniew Jędrzejewski-Szmek [Tue, 28 Oct 2014 02:34:09 +0000 (21:34 -0500)]
rpm: add user macros

6 years agomanager: do not print anything while passwords are being queried
Zbigniew Jędrzejewski-Szmek [Sun, 26 Oct 2014 00:30:51 +0000 (20:30 -0400)]
manager: do not print anything while passwords are being queried

6 years agocdrom_id: do not attempt to read past end of buffer
Zbigniew Jędrzejewski-Szmek [Tue, 28 Oct 2014 01:52:38 +0000 (21:52 -0400)]
cdrom_id: do not attempt to read past end of buffer

CID #1238437

6 years agoselinux: make sure we do not try to print missing fields
Zbigniew Jędrzejewski-Szmek [Tue, 28 Oct 2014 01:31:29 +0000 (21:31 -0400)]
selinux: make sure we do not try to print missing fields

UID or GID of 0 is valid, so we cannot use that to distinguish whether
calls to sd_bus_creds_get_* succeeded, and the return value from the
function is the only way to know about missing fields. Print "n/a" if
the fields are missing.

CID #1238779

6 years agologin: remove multi-seat-x
Timofey Titovets [Sat, 25 Oct 2014 21:17:24 +0000 (00:17 +0300)]
login: remove multi-seat-x

6 years agoupdate TODO
Lennart Poettering [Tue, 28 Oct 2014 01:19:37 +0000 (02:19 +0100)]
update TODO

6 years agoUpdate NEWS
Lennart Poettering [Tue, 28 Oct 2014 01:17:12 +0000 (02:17 +0100)]
Update NEWS

6 years agounits: define appropriate job timeout actions when boot or shutdown timeouts are hit
Lennart Poettering [Tue, 28 Oct 2014 00:49:39 +0000 (01:49 +0100)]
units: define appropriate job timeout actions when boot or shutdown timeouts are hit

Using the new JobTimeoutAction= setting make sure we power off the
machine after is queued for longer than 15min but not
executed. Similar, if is queued for longer than 30min
but does not complete, forcibly turn off the system. Similar, if is queued for longer than 30min but does not complete,
forcibly reboot the system.

6 years agojob: optionally, when a job timeout is hit, also execute a failure action
Lennart Poettering [Tue, 28 Oct 2014 00:49:07 +0000 (01:49 +0100)]
job: optionally, when a job timeout is hit, also execute a failure action

6 years agocore: remove system start timeout logic again
Lennart Poettering [Tue, 28 Oct 2014 00:42:13 +0000 (01:42 +0100)]
core: remove system start timeout logic again

The system start timeout as previously implemented would get confused by
long-running services that are included in the initial system startup
transaction for example by being cron-job-like long-running services
triggered immediately at boot. Such long-running jobs would be subject
to the default 15min timeout, esily triggering it.

Hence, remove this again. In a subsequent commit, introduce per-target
job timeouts instead, that allow us to control these timeouts more

6 years agoupdate TODO
Lennart Poettering [Mon, 27 Oct 2014 23:52:48 +0000 (00:52 +0100)]
update TODO

6 years agojournald: be nice to coverity, add an extra assert
Lennart Poettering [Mon, 27 Oct 2014 22:50:51 +0000 (23:50 +0100)]
journald: be nice to coverity, add an extra assert

coverity otherwise assumes that the chain object might be NULL.

6 years agoCODING_STYLE: don't clobber arguments on failure
Lennart Poettering [Mon, 27 Oct 2014 17:09:07 +0000 (18:09 +0100)]
CODING_STYLE: don't clobber arguments on failure

6 years agocalendarspec: parse 'quarterly' and 'semi-annually' as shortcuts
Lennart Poettering [Mon, 27 Oct 2014 17:08:46 +0000 (18:08 +0100)]
calendarspec: parse 'quarterly' and 'semi-annually' as shortcuts

6 years agonetworkd: network - if no prefixlength is given, try to deduce one from the address...
Tom Gundersen [Mon, 27 Oct 2014 16:39:18 +0000 (17:39 +0100)]
networkd: network - if no prefixlength is given, try to deduce one from the address class

In case of a class E or F address, ignore the address.

6 years agoshared: in-addr-utils - add default_subnet_mask and default_prefixlen methods
Tom Gundersen [Mon, 27 Oct 2014 16:38:03 +0000 (17:38 +0100)]
shared: in-addr-utils - add default_subnet_mask and default_prefixlen methods

These use the (deprecated) IPv4 address classes to deduce the corresponding subnet masks. This is useful when addresses
without subnet masks and prefix lengths are given.

Make use of these new functions from sd-dhcp-lease.

6 years agoman: tmpfiles.d - recommend using b! and c!
Tom Gundersen [Mon, 27 Oct 2014 16:28:29 +0000 (17:28 +0100)]
man: tmpfiles.d - recommend using b! and c!

We should avoid creating static device nodes at runtime.

6 years agounits: tmpfiles-setup-dev - allow unsafe file creation to happen in /dev at boot
Tom Gundersen [Mon, 27 Oct 2014 16:15:42 +0000 (17:15 +0100)]
units: tmpfiles-setup-dev - allow unsafe file creation to happen in /dev at boot

This will allow us to mark static device nodes with '!' to indicate that they should only be created at early boot.

6 years agosd-bus: sync kdbus.h (ABI break)
Daniel Mack [Mon, 27 Oct 2014 16:02:31 +0000 (17:02 +0100)]
sd-bus: sync kdbus.h (ABI break)

Some comment fixes and header cleanups in kdbus.h, and the task capability
meta information has been factored out to its own struct.

6 years agoman: document what "minutely" now means
Lennart Poettering [Mon, 27 Oct 2014 12:54:19 +0000 (13:54 +0100)]
man: document what "minutely" now means

6 years agocalendar: new case 'minutely'
Daniele Medri [Mon, 27 Oct 2014 07:42:42 +0000 (08:42 +0100)]
calendar: new case 'minutely'

6 years agoupdate TODO
Lennart Poettering [Mon, 27 Oct 2014 12:32:04 +0000 (13:32 +0100)]
update TODO

6 years agoupdate NEWS
Lennart Poettering [Mon, 27 Oct 2014 12:31:56 +0000 (13:31 +0100)]
update NEWS

6 years agosd-bus: explicitly cast asprintf() return value away we are not interested in
Lennart Poettering [Mon, 27 Oct 2014 12:06:11 +0000 (13:06 +0100)]
sd-bus: explicitly cast asprintf() return value away we are not interested in

Let's give coverity a hint what's going on here.

6 years agoRevert "sd-bus: check return value of asprintf()"
Lennart Poettering [Mon, 27 Oct 2014 12:04:12 +0000 (13:04 +0100)]
Revert "sd-bus: check return value of asprintf()"

This reverts commit b1543c4c93855b61b40118e9f14a0423dac2e078.

We check b->address anyway, no need to check the return value,
especially given that the other #ifdef branch don't get the same

6 years agoshutdown: do final unmounting only if not running inside the container
Michal Sekletar [Tue, 21 Oct 2014 16:38:42 +0000 (18:38 +0200)]
shutdown: do final unmounting only if not running inside the container

If we run in the container then we run in a mount namespace. If namespace dies
then kernel should do unmounting for us, hence we skip unmounting in containers.

Also, it may be the case that we no longer have capability to do umount,
because we are running in the unprivileged container.


6 years agoman: fix typos in description of SELinuxContextFromNet=
Ivan Shapovalov [Mon, 27 Oct 2014 09:07:36 +0000 (12:07 +0300)]
man: fix typos in description of SELinuxContextFromNet=

6 years agonetworkd: don't stop the dhcp server when link goes down
Tom Gundersen [Sun, 26 Oct 2014 13:01:27 +0000 (14:01 +0100)]
networkd: don't stop the dhcp server when link goes down

6 years agoutil: fix copy-paste error and actually set the new hostname
Michal Sekletar [Mon, 27 Oct 2014 10:08:26 +0000 (11:08 +0100)]
util: fix copy-paste error and actually set the new hostname

Reported-by: sztanpet on irc
6 years agoutil: introduce sethostname_idempotent
Michal Sekletar [Tue, 21 Oct 2014 16:17:54 +0000 (18:17 +0200)]
util: introduce sethostname_idempotent

Function queries system hostname and applies changes only when necessary. Also,
migrate all client of sethostname to sethostname_idempotent while at it.