chiark / gitweb /
elogind.git
9 years agoudev-builtin-btrfs: properly initialize ioctl struct to zeroes
Lennart Poettering [Fri, 12 Dec 2014 01:34:12 +0000 (02:34 +0100)]
udev-builtin-btrfs: properly initialize ioctl struct to zeroes

9 years agoutil: document why we have alloca_align()
Lennart Poettering [Fri, 12 Dec 2014 01:33:48 +0000 (02:33 +0100)]
util: document why we have alloca_align()

9 years agoutil: when using basename() for creating temporary files, verify the resulting name...
Lennart Poettering [Fri, 12 Dec 2014 01:32:33 +0000 (02:32 +0100)]
util: when using basename() for creating temporary files, verify the resulting name is actually valid

Also, rename filename_is_safe() to filename_is_valid(), since it
actually does a full validation for what the kernel will accept as file
name, it's not just a heuristic.

9 years agojournal: fix dangling 'else' ambiguity
David Herrmann [Fri, 12 Dec 2014 08:52:06 +0000 (09:52 +0100)]
journal: fix dangling 'else' ambiguity

Rework the sd-journal iterators to avoid dangling 'else' ambiguity. For a
detailed explanation, see:

    commit bff686e2a981ccd0888cdf1981977d24320f1770
    Author: David Herrmann <dh.herrmann@gmail.com>
    Date:   Fri Dec 12 09:43:54 2014 +0100

        hwdb: fix dangling 'else' ambuguity

9 years agohwdb: fix dangling 'else' ambuguity
David Herrmann [Fri, 12 Dec 2014 08:43:54 +0000 (09:43 +0100)]
hwdb: fix dangling 'else' ambuguity

Imagine the following use of hwdb:

    if (condition_A)
        SD_HWDB_FOREACH_PROPERTY(hwdb, modalias, key, value)
            operation_A(key, value);
    else
        log_error("...");

This should work just fine, but but definitely does not what you would
expect. Due to how SD_HWDB_FOREACH_PROPERTY is defined, the dangling
'else' is linked to the hidden 'if' statement in the macro instead of the
outer 'if (condition_A)'. This is unexpected and really annoying to debug.

Fix this by never leaving un-finished if-statements in
SD_HWDB_FOREACH_PROPERTY(). We simply inverse the if() statement and
explicitly add an 'else'-branch. This way, the statement is closed and all
ambuguities are resolved.

9 years agohwdb: FOREACH_HWDB_PROPERTY -> SD_HWDB_FOREACH_PROPERTY
David Herrmann [Fri, 12 Dec 2014 08:30:50 +0000 (09:30 +0100)]
hwdb: FOREACH_HWDB_PROPERTY -> SD_HWDB_FOREACH_PROPERTY

Lets not pollute the global namespace. Prefix all our exported names and
macros with SD_HWDB_*.

9 years agohwdb: Update database of Bluetooth company identifiers
Marcel Holtmann [Thu, 11 Dec 2014 21:50:50 +0000 (22:50 +0100)]
hwdb: Update database of Bluetooth company identifiers

9 years agouse correct format types
Thomas Hindoe Paaboel Andersen [Wed, 10 Dec 2014 19:24:18 +0000 (20:24 +0100)]
use correct format types

9 years agotest-condition: add more test cases
Ronny Chevalier [Thu, 11 Dec 2014 17:30:37 +0000 (18:30 +0100)]
test-condition: add more test cases

9 years agotest-strv: add test for strv_equal
Ronny Chevalier [Thu, 11 Dec 2014 17:30:16 +0000 (18:30 +0100)]
test-strv: add test for strv_equal

9 years agotest-execute: add tests for UMask directive
Ronny Chevalier [Thu, 11 Dec 2014 16:59:10 +0000 (17:59 +0100)]
test-execute: add tests for UMask directive

9 years agotest-unit-name: add tests for %f
Ronny Chevalier [Thu, 11 Dec 2014 16:58:40 +0000 (17:58 +0100)]
test-unit-name: add tests for %f

9 years agobus: sync with kdbus.git
David Herrmann [Thu, 11 Dec 2014 16:25:09 +0000 (17:25 +0100)]
bus: sync with kdbus.git

Changes:
 * bloom parameters are returned in an offset via HELLO
 * FREE now takes items just like any other ioctl

9 years agotree-wide: use our memset() macros instead of memset() itself
Lennart Poettering [Thu, 11 Dec 2014 15:58:45 +0000 (16:58 +0100)]
tree-wide: use our memset() macros instead of memset() itself

9 years agotreewide: correct spacing near eol in code comments
Torstein Husebø [Wed, 10 Dec 2014 19:00:09 +0000 (20:00 +0100)]
treewide: correct spacing near eol in code comments

9 years agocore: correct spacing near eol in code comments
Torstein Husebø [Wed, 10 Dec 2014 19:00:08 +0000 (20:00 +0100)]
core: correct spacing near eol in code comments

9 years agoshared: correct spacing near eol in code comments
Torstein Husebø [Wed, 10 Dec 2014 19:00:07 +0000 (20:00 +0100)]
shared: correct spacing near eol in code comments

9 years agojournald: correct spacing near eol code comments
Torstein Husebø [Wed, 10 Dec 2014 19:00:06 +0000 (20:00 +0100)]
journald: correct spacing near eol code comments

9 years agosd-bus: correct spacing near eol in code comments
Torstein Husebø [Wed, 10 Dec 2014 19:00:05 +0000 (20:00 +0100)]
sd-bus: correct spacing near eol in code comments

9 years agonetworkd/resolved: correct spacing near eol in code comments
Torstein Husebø [Wed, 10 Dec 2014 19:00:04 +0000 (20:00 +0100)]
networkd/resolved: correct spacing near eol in code comments

9 years agobuild-sys: fix user unit m4 logic
Lennart Poettering [Thu, 11 Dec 2014 13:56:33 +0000 (14:56 +0100)]
build-sys: fix user unit m4 logic

9 years agoupdate TODO
Lennart Poettering [Thu, 11 Dec 2014 00:53:43 +0000 (01:53 +0100)]
update TODO

9 years agosd-dhcp-client: log if we fail to set up lease timers
Tom Gundersen [Thu, 11 Dec 2014 13:43:09 +0000 (14:43 +0100)]
sd-dhcp-client: log if we fail to set up lease timers

9 years agotests: udev - assume /etc/machine-id rather than /etc/hosts
Tom Gundersen [Thu, 11 Dec 2014 13:17:35 +0000 (14:17 +0100)]
tests: udev - assume /etc/machine-id rather than /etc/hosts

On todays machines /etc/hosts is not mandatory. /etc/machine-id is though, so let's rely on that instead.

This makes the udev tests pass again for me.

9 years agolibudev: make libudev-hwdb a wrapper around sd-hwdb
Tom Gundersen [Wed, 3 Dec 2014 16:32:40 +0000 (17:32 +0100)]
libudev: make libudev-hwdb a wrapper around sd-hwdb

9 years agolibsystemd: add sd-hwdb library
Tom Gundersen [Wed, 3 Dec 2014 16:02:34 +0000 (17:02 +0100)]
libsystemd: add sd-hwdb library

This is libudev-hwdb, but decoupled from libudev and in the libsystemd style.

The core code is unchanged, apart from the following minor changes:

 - hwdb.bin located in /**/systemd/hwdb/ take preference over the ones located
   in /**/udev/
 - properties are stored internally in an OrderedHashmap, rather than a
   linked list.
 - a new API call allows individual properties to be queried directly, rather
   than iterating over them all
 - the iteration over properties have been moved inside the library, rather than
   exposing a list directly
 - the unused 'flags' parameter was dropped

9 years agobus: sync with kdbus.git
David Herrmann [Thu, 11 Dec 2014 12:19:41 +0000 (13:19 +0100)]
bus: sync with kdbus.git

Sync up with recent kdbus changed:
 * several ioctls gained .size and .items members (but still unused)
 * CMD_SEND gained its own ioctl structure
 * several members of kdbus_msg were dropped as they were only used during
   SEND, not during RECV etc.
 * CMD_RECV and CMD_SEND now share a kdbus_reply member which contains the
   offset and size of the returned message.

9 years agobus: zero cmd_free before passing to ioctl
David Herrmann [Thu, 11 Dec 2014 09:45:58 +0000 (10:45 +0100)]
bus: zero cmd_free before passing to ioctl

Make sure the whole cmd_free object is zeroed before passing it into the
kernel. This makes valgrind happy and makes us future proof.

9 years agobus: fix memfd-cache regarding memfd offsets
David Herrmann [Thu, 11 Dec 2014 09:44:41 +0000 (10:44 +0100)]
bus: fix memfd-cache regarding memfd offsets

We must restore part->mmap_begin when poping memfds from the memfd-cache.
We rely on the memfds to be unsealed, so we can be sure that we own the
whole FD. Therefore, simply set part->mmap_begin to the same as
part->data.

This fixes test-bus-kernel-benchmark.

9 years agoTODO: update
David Herrmann [Thu, 11 Dec 2014 08:01:34 +0000 (09:01 +0100)]
TODO: update

Drop resolved kdbus issues.

9 years agolocalectl,man: make it obvious that set-*-keymaps sets both keymaps
Zbigniew Jędrzejewski-Szmek [Thu, 11 Dec 2014 05:02:54 +0000 (00:02 -0500)]
localectl,man: make it obvious that set-*-keymaps sets both keymaps

https://bugs.freedesktop.org/show_bug.cgi?id=85411

9 years agozsh-completion: remove duplicate function v218
Zbigniew Jędrzejewski-Szmek [Wed, 10 Dec 2014 22:34:41 +0000 (17:34 -0500)]
zsh-completion: remove duplicate function

https://bugs.archlinux.org/task/43069

9 years agobuild-sys: update Lennart's private target
Lennart Poettering [Wed, 10 Dec 2014 22:19:55 +0000 (23:19 +0100)]
build-sys: update Lennart's private target

9 years agoNEWS: minor wording improvement
Lennart Poettering [Wed, 10 Dec 2014 22:17:54 +0000 (23:17 +0100)]
NEWS: minor wording improvement

9 years agobuild-sys: add missing file to CLEANFILES
Lennart Poettering [Wed, 10 Dec 2014 22:01:13 +0000 (23:01 +0100)]
build-sys: add missing file to CLEANFILES

9 years agoNEWS: update contributors list
Lennart Poettering [Wed, 10 Dec 2014 21:33:21 +0000 (22:33 +0100)]
NEWS: update contributors list

9 years agobuild-sys: bump package and library versions in preparation for 218 release
Lennart Poettering [Wed, 10 Dec 2014 21:31:27 +0000 (22:31 +0100)]
build-sys: bump package and library versions in preparation for 218 release

9 years agobuild-sys: turn off SMACK capabilities stuff for now, since it is incompatible with...
Lennart Poettering [Wed, 10 Dec 2014 21:10:46 +0000 (22:10 +0100)]
build-sys: turn off SMACK capabilities stuff for now, since it is incompatible with nspawn

9 years agoupdate TODO
Lennart Poettering [Wed, 10 Dec 2014 21:08:49 +0000 (22:08 +0100)]
update TODO

9 years agoscope: make attachment of initial PIDs a bit more robust
Lennart Poettering [Wed, 10 Dec 2014 21:06:44 +0000 (22:06 +0100)]
scope: make attachment of initial PIDs a bit more robust

9 years agocore: don't migrate PIDs for units that may contain subcgroups, do this only for...
Lennart Poettering [Wed, 10 Dec 2014 19:38:24 +0000 (20:38 +0100)]
core: don't migrate PIDs for units that may contain subcgroups, do this only for leaf units

Otherwise a slice or delegation unit might move PIDs around ignoring the
fact that it is attached to a subcgroup.

9 years agoupdate TODO
Lennart Poettering [Wed, 10 Dec 2014 18:57:52 +0000 (19:57 +0100)]
update TODO

9 years agocore: properly pass unit file state to clients via the bus
Lennart Poettering [Wed, 10 Dec 2014 18:57:45 +0000 (19:57 +0100)]
core: properly pass unit file state to clients via the bus

9 years agobuild-sys: fix link-order
Kay Sievers [Wed, 10 Dec 2014 18:46:00 +0000 (19:46 +0100)]
build-sys: fix link-order

9 years agoupdate TODO
Lennart Poettering [Wed, 10 Dec 2014 18:07:32 +0000 (19:07 +0100)]
update TODO

9 years agosd-bus: make BUS_ERROR_MAP_ELF_USE() use a const variable
Lennart Poettering [Wed, 10 Dec 2014 18:06:43 +0000 (19:06 +0100)]
sd-bus: make BUS_ERROR_MAP_ELF_USE() use a const variable

9 years agonetworkd: rename section [BridgePort] → [Bridge]
Lennart Poettering [Wed, 10 Dec 2014 18:03:08 +0000 (19:03 +0100)]
networkd: rename section [BridgePort] → [Bridge]

Let's stick to generic sections that describe the general technology,
instead of specific per-object sections, unless we really have a reason
to do that otherwise.

9 years agosd-bus: move common errors src/shared/bus-errors.h → src/libsystemd/sd-bus/bus-common...
Lennart Poettering [Wed, 10 Dec 2014 18:00:46 +0000 (19:00 +0100)]
sd-bus: move common errors src/shared/bus-errors.h → src/libsystemd/sd-bus/bus-common-errors.h

Stuff in src/shared/ should not use stuff from src/libsystemd/ really.

9 years agonetworkd-dhcp6: Support ICMPv6 Other information
Patrik Flykt [Wed, 10 Dec 2014 14:17:35 +0000 (16:17 +0200)]
networkd-dhcp6: Support ICMPv6 Other information

When ICMPv6 Other information is received, enable Information request
in DHCPv6. If the DHCPv6 client already exists, only update the client
if there is a transition from Other to Managed state.

9 years agonetworkd-dhcp6: Move ICMPv6 and DHCPv6 configuration to new file
Patrik Flykt [Wed, 10 Dec 2014 14:17:34 +0000 (16:17 +0200)]
networkd-dhcp6: Move ICMPv6 and DHCPv6 configuration to new file

Handle all aspects of ICMPv6 and DHCPv6 in a file of its own as is done
with DHCPv4 and IPv4LL.

9 years agotest-dhcp6-client: Add a simple Information Request test case
Patrik Flykt [Wed, 10 Dec 2014 14:17:33 +0000 (16:17 +0200)]
test-dhcp6-client: Add a simple Information Request test case

Start the DHCPv6 test case by sending an Information Request, verifying
the reply and continuing at once with the normal address acquisition
procedure. Reuse the DHCPv6 Solicit Reply so that the client code is
verified to ignore any erroneously added IPv6 address information.

9 years agosd-dhcp6-client: Implement Information Request message
Patrik Flykt [Wed, 10 Dec 2014 14:17:32 +0000 (16:17 +0200)]
sd-dhcp6-client: Implement Information Request message

Implement Information Request message according to RFC 3315, section
18.1.5. with the excepion that the first message is not delayed by a
random amount. Instead systemd-networkd is supposed to take care of
desynchronizing between other clients.

Initialize the DHCPv6 client structure in sd_dhcp6_client_start()
as this allows toggling between information request and normal
DHCPv6 address aquisition modes.

9 years agosd-dhcp6-lease: Return only -EINVAL when a NULL lease is supplied
Patrik Flykt [Wed, 10 Dec 2014 14:17:31 +0000 (16:17 +0200)]
sd-dhcp6-lease: Return only -EINVAL when a NULL lease is supplied

Suppyling a NULL lease is not a condition that needs to be asserted,
returning -EINVAL is informative enough to the caller. This simplifies
calling code and doesn't falsely indicate that something erroneous was
attempted.

9 years agosd-dhcp6-client: Make end of successfull option parsing explicit
Patrik Flykt [Wed, 10 Dec 2014 14:17:30 +0000 (16:17 +0200)]
sd-dhcp6-client: Make end of successfull option parsing explicit

When all DHCPv6 options have been parsed, dhcp6_option_parse() returns
-ENOMSG. Explicitely set the return value to indicate success so that
later code does not need to take this special value into account.

9 years agohwdb: fix typo
Torstein Husebø [Wed, 10 Dec 2014 16:41:54 +0000 (17:41 +0100)]
hwdb: fix typo

9 years agohwdb: add two more mice to the DPI database
Lennart Poettering [Wed, 10 Dec 2014 16:38:47 +0000 (17:38 +0100)]
hwdb: add two more mice to the DPI database

9 years agohwdb: add another Logitech mouse (M-BZ96C)
Lennart Poettering [Wed, 10 Dec 2014 14:59:38 +0000 (15:59 +0100)]
hwdb: add another Logitech mouse (M-BZ96C)

9 years agoTODO
Lennart Poettering [Wed, 10 Dec 2014 14:46:28 +0000 (15:46 +0100)]
TODO

9 years agoman: improve docs for systemd-modules-load.service
Lennart Poettering [Wed, 10 Dec 2014 14:46:22 +0000 (15:46 +0100)]
man: improve docs for systemd-modules-load.service

9 years agohwdb: add DPI data for M325 mouse
Lennart Poettering [Wed, 10 Dec 2014 14:46:08 +0000 (15:46 +0100)]
hwdb: add DPI data for M325 mouse

9 years agotest: fix some tests when running inside a container
Jan Synacek [Wed, 10 Dec 2014 09:20:11 +0000 (10:20 +0100)]
test: fix some tests when running inside a container

9 years agovirt: when detecting containers and /run/systemd/container cannot be read, check...
Lennart Poettering [Wed, 10 Dec 2014 12:23:49 +0000 (13:23 +0100)]
virt: when detecting containers and /run/systemd/container cannot be read, check /proc/1/environ

This way, we should be in a slightly better situation if a container is
booted up with only a shell as PID 1. In that case
/run/systemd/container will not be populated, and a check for it hence
be ineffective.

Checking /proc/1/environ doesn't fully fix the problem though, as the
file is only accessible with privileges. This means if PID 1 is not
systemd, and if privileges have been dropped the container detection
will continue to fail.

9 years agosystemctl: fix invalid free when enabling sysv services fails
Zbigniew Jędrzejewski-Szmek [Tue, 9 Dec 2014 19:41:24 +0000 (14:41 -0500)]
systemctl: fix invalid free when enabling sysv services fails

The error was introduced in v215-343-g60731f32f1 'systemctl: do not
bother to mutate state on error', by causing strv_free to attempt to
free a static string. Simplify the whole thing by always keeping the
array in valid state.

9 years agoima-setup: simplify
Zbigniew Jędrzejewski-Szmek [Tue, 2 Dec 2014 01:47:37 +0000 (20:47 -0500)]
ima-setup: simplify

9 years agotreewide: sanitize loop_write
Zbigniew Jędrzejewski-Szmek [Tue, 2 Dec 2014 01:43:19 +0000 (20:43 -0500)]
treewide: sanitize loop_write

loop_write() didn't follow the usual systemd rules and returned status
partially in errno and required extensive checks from callers. Some of
the callers dealt with this properly, but many did not, treating
partial writes as successful. Simplify things by conforming to usual rules.

9 years agokdbus: downgrade warning if we cannot patch kdbus attach mask to DEBUG if kdbus is...
Lennart Poettering [Wed, 10 Dec 2014 02:25:48 +0000 (03:25 +0100)]
kdbus: downgrade warning if we cannot patch kdbus attach mask to DEBUG if kdbus is not available

9 years agoutil: introduce our own gperf based capability list
Lennart Poettering [Wed, 10 Dec 2014 02:16:14 +0000 (03:16 +0100)]
util: introduce our own gperf based capability list

This way, we can ensure we have a more complete, up-to-date list of
capabilities around, always.

9 years agoupdate TODO
Lennart Poettering [Wed, 10 Dec 2014 01:36:42 +0000 (02:36 +0100)]
update TODO

9 years agobuild-sys: fix CLEANFILES= for dns_type list
Lennart Poettering [Wed, 10 Dec 2014 01:36:35 +0000 (02:36 +0100)]
build-sys: fix CLEANFILES= for dns_type list

9 years agomount: clarify that we really need to replace the utab inotify code with the native...
Lennart Poettering [Wed, 10 Dec 2014 00:47:59 +0000 (01:47 +0100)]
mount: clarify that we really need to replace the utab inotify code with the native API for this in libmount, as soon as that's stable

9 years agocore: unify how we iterate over inotify events
Lennart Poettering [Wed, 10 Dec 2014 00:45:43 +0000 (01:45 +0100)]
core: unify how we iterate over inotify events

Let's add some syntactic sugar for iterating through inotify events, and
use it everywhere.

9 years agomount: use bools where appropriate
Lennart Poettering [Tue, 9 Dec 2014 23:52:56 +0000 (00:52 +0100)]
mount: use bools where appropriate

9 years agocatalog: fix typos in the French translation
Ronny Chevalier [Wed, 10 Dec 2014 00:33:08 +0000 (01:33 +0100)]
catalog: fix typos in the French translation

9 years agoshell-completion: systemctl edit
Ronny Chevalier [Wed, 10 Dec 2014 00:10:01 +0000 (01:10 +0100)]
shell-completion: systemctl edit

9 years agoNEWS: fix typos
Ronny Chevalier [Tue, 9 Dec 2014 23:50:24 +0000 (00:50 +0100)]
NEWS: fix typos

9 years agogitignore: hide some more files
Lennart Poettering [Tue, 9 Dec 2014 23:49:35 +0000 (00:49 +0100)]
gitignore: hide some more files

9 years agohwdb: update PCI, USB, OUI databases
Lennart Poettering [Tue, 9 Dec 2014 23:46:33 +0000 (00:46 +0100)]
hwdb: update PCI, USB, OUI databases

9 years agobuild-sys: don't fail if CAP_SETFCAP is missing
Lennart Poettering [Tue, 9 Dec 2014 23:40:23 +0000 (00:40 +0100)]
build-sys: don't fail if CAP_SETFCAP is missing

9 years agohwdb: add more mouse DPI entries
Peter Hutterer [Sun, 7 Dec 2014 23:17:26 +0000 (09:17 +1000)]
hwdb: add more mouse DPI entries

Plus a note for Logitech devices using the unified receiver: these devices
include their wireless PID in the name, the usb PID/VID is the same for all.
In kernel 3.19 the actual model number will be the name, but the patches are
still a bit in flux at this point. In the future each device will need two
entries for pre+3.19 and 3.19.

https://bugs.freedesktop.org/show_bug.cgi?id=87037
https://bugs.freedesktop.org/show_bug.cgi?id=87072
https://bugs.freedesktop.org/show_bug.cgi?id=87162

9 years agoNEWS: add contributors list
Lennart Poettering [Tue, 9 Dec 2014 23:30:19 +0000 (00:30 +0100)]
NEWS: add contributors list

9 years agoNEWS: prepare NEWS for new release
Lennart Poettering [Tue, 9 Dec 2014 23:27:26 +0000 (00:27 +0100)]
NEWS: prepare NEWS for new release

9 years agonspawn: create the macvlan MAC addresses in an arch independent stable way
Lennart Poettering [Tue, 9 Dec 2014 23:26:16 +0000 (00:26 +0100)]
nspawn: create the macvlan MAC addresses in an arch independent stable way

9 years agojournal: optimize iteration: skip files that cannot improve current candidate entry
Michal Schmidt [Tue, 9 Dec 2014 19:27:26 +0000 (20:27 +0100)]
journal: optimize iteration: skip files that cannot improve current candidate entry

Suppose that while iterating we have already looked into a journal file
and got a candidate for the next entry. And we are considering to look
into another journal file because it may contain an entry that is nearer
to the current location than the candidate.

We should skip the whole journal file if we can tell by looking at its
header that none of its entries can precede the candidate.

Before:
$ time ./journalctl --since=2014-06-01 --until=2014-07-01 > /dev/null

real    0m20.518s
user    0m19.989s
sys     0m0.328s

After:
$ time ./journalctl --since=2014-06-01 --until=2014-07-01 > /dev/null

real    0m9.445s
user    0m9.228s
sys     0m0.213s

9 years agojournal: optimize iteration: skip whole files behind current location
Michal Schmidt [Tue, 9 Dec 2014 19:27:19 +0000 (20:27 +0100)]
journal: optimize iteration: skip whole files behind current location

Interleaving of entries from many journal files is expensive. But there
is room for optimization.

We can skip looking into journal files whose entries all lie before the
current iterating location. We can tell if that's the case from looking
at the journal file header. This saves a huge amount of work if one has
many of mostly not interleaved journal files.

On my workstation with 90 journal files in /var/log/journal/ID/
totalling 3.4 GB I get these results:

Before:
$ time ./journalctl --since=2014-06-01 --until=2014-07-01 > /dev/null

real    5m54.258s
user    2m4.263s
sys     3m48.965s

After:
$ time ./journalctl --since=2014-06-01 --until=2014-07-01 > /dev/null

real    0m20.518s
user    0m19.989s
sys     0m0.328s

The high "sys" time in the original was caused by putting more stress on
the mmap-cache than it could handle. With the patch the working set
now consists of fewer mmap windows and mmap-cache is not thrashing.

9 years agonetworkd: remove unused variable
Thomas Hindoe Paaboel Andersen [Tue, 9 Dec 2014 19:50:37 +0000 (20:50 +0100)]
networkd: remove unused variable

It is no longer used after 45af44d47da6933b260c734ad9ff721f63f80a4d

9 years agounit: update unit dropin paths and time when dropin file is written.
WaLyong Cho [Tue, 9 Dec 2014 12:46:30 +0000 (21:46 +0900)]
unit: update unit dropin paths and time when dropin file is written.

If a unit is set property by "systemctl set-property", a new dropin
file is generated. But the unit's dropin_paths and dropin_mtime are
not updated. So the unit is shown as need daemon reload.
Update unit dropin_paths and dropin_mtime also when dropin file is
written.

9 years agobus-proxy: cloning smack label
Przemyslaw Kedzierski [Tue, 9 Dec 2014 11:17:24 +0000 (12:17 +0100)]
bus-proxy: cloning smack label

When dbus client connects to systemd-bus-proxyd through
Unix domain socket proxy takes client's smack label and sets for itself.

It is done before and independent of dropping privileges.

The reason of such soluton is fact that tests of access rights
performed by lsm may take place inside kernel, not only
in userspace of recipient of message.

The bus-proxyd needs CAP_MAC_ADMIN to manipulate its label.

In case of systemd running in system mode, CAP_MAC_ADMIN
should be added to CapabilityBoundingSet in service file of bus-proxyd.

In case of systemd running in user mode ('systemd --user')
it can be achieved by addition
Capabilities=cap_mac_admin=i and SecureBits=keep-caps
to user@.service file
and setting cap_mac_admin+ei on bus-proxyd binary.

9 years agorun: introduce timer support option
WaLyong Cho [Tue, 9 Dec 2014 07:07:16 +0000 (16:07 +0900)]
run: introduce timer support option

Support timer options --on-active=, --on-boot=, --on-startup=,
--on-unit-active=, --on-unit-inactive=, --on-calendar=. Each options
corresponding with OnActiveSec=, OnBootSec=, OnStartupSec=,
OnUnitActiveSec=, OnUnitInactiveSec=, OnCalendar= of timer
respectively. And OnCalendar= and WakeSystem= supported by
--timer-property= option like --property= of systemd-run.

And if --unit= option and timer options are specified the command can
be omitted. In this case, systemd-run assumes the target service is
already loaded. And just try to generate transient timer unit only.

9 years agoupdate TODO
Lennart Poettering [Tue, 9 Dec 2014 17:17:35 +0000 (18:17 +0100)]
update TODO

9 years agosd-bus: get rid of PID starttime concept
Lennart Poettering [Tue, 9 Dec 2014 17:16:54 +0000 (18:16 +0100)]
sd-bus: get rid of PID starttime concept

As kdbus no longer exports this, remove all traces from sd-bus too

9 years agosd-bus: properly support passing memfds around with offset/size parameters
Lennart Poettering [Tue, 9 Dec 2014 17:07:46 +0000 (18:07 +0100)]
sd-bus: properly support passing memfds around with offset/size parameters

9 years agobus: sync with kdbus-git (ABI break)
David Herrmann [Tue, 9 Dec 2014 10:12:41 +0000 (11:12 +0100)]
bus: sync with kdbus-git (ABI break)

kdbus-git gained two new features:
 * memfd offsets: This allows to specify a 'start' offset in kdbus_memfd
                  so you can send partial memfd hunks instead of always
                  the full memfd
 * KDBUS_HELLO_UNPRIVILEGED: If passed during HELLO, the client will be
                             treated as unprivileged.

9 years agolibsystemd-network: fix writing of routes in dhcp lease file
Dan Winship [Tue, 18 Nov 2014 13:59:42 +0000 (08:59 -0500)]
libsystemd-network: fix writing of routes in dhcp lease file

inet_ntoa() uses a static buffer, so you can't call it twice in the
same fprintf() call.

9 years agosleep: drop redundant MESSAGE= prefix
Lennart Poettering [Tue, 9 Dec 2014 02:58:20 +0000 (03:58 +0100)]
sleep: drop redundant MESSAGE= prefix

9 years agoupdate TODO
Lennart Poettering [Tue, 9 Dec 2014 02:58:10 +0000 (03:58 +0100)]
update TODO

9 years agojournalctl: respect --after-cursor semantics with --follow in all cases
Wesley Dawson [Mon, 1 Dec 2014 07:27:00 +0000 (23:27 -0800)]
journalctl: respect --after-cursor semantics with --follow in all cases

In the case where no entries have been added to the journal after the specified
cursor, set need_seek before the main loop to prevent display of the entry at
said cursor.

9 years agocore: rename unit_destroy_cgroup() to unit_destroy_cgroup_if_empty() since it's not...
Lennart Poettering [Tue, 9 Dec 2014 01:31:42 +0000 (02:31 +0100)]
core: rename unit_destroy_cgroup() to unit_destroy_cgroup_if_empty() since it's not quite as destructive as it sounds nowadays

9 years agocgroup: Handle error when destroying cgroup
Ross Lagerwall [Sat, 29 Nov 2014 15:27:14 +0000 (15:27 +0000)]
cgroup: Handle error when destroying cgroup

If a cgroup fails to be destroyed (most likely because there are still
processes running as part of a service after the main pid exits), don't
free and remove the cgroup unit from the manager.  This fixes a
regression introduced by the cgroup rework in v205 where systemd would
forget about processes still running after the unit becomes inactive.
(This can happen when the main pid exits and KillMode=process or none).

9 years agofstab-generator: Allow mount.usr without mount.usrflags, honor rw/ro
Michael Marineau [Mon, 8 Dec 2014 19:05:31 +0000 (11:05 -0800)]
fstab-generator: Allow mount.usr without mount.usrflags, honor rw/ro

There is no need to require mount.usrflags. The original implementation
assumed that a btrfs subvolume would always be needed but that is not
applicable to systems that do not use btrfs for /usr.

Similar to using rootflags= for the default of mount.usrflags=, append
the classic 'ro' and 'rw' flags to the mount options.

9 years agofstab-generator: free all allocated strings
Lennart Poettering [Tue, 9 Dec 2014 01:22:44 +0000 (02:22 +0100)]
fstab-generator: free all allocated strings