summaryrefslogtreecommitdiff
path: root/src
AgeCommit message (Collapse)Author
2013-11-07bus: indicate in .pc file that we need libsystemd-id128 definitionsLennart Poettering
2013-11-07api: add C++ guardsLennart Poettering
2013-11-07rtnl: headers in src/systemd/ may not use #pragma onceLennart Poettering
2013-11-07rtnl: headers in src/systemd/ may not include internal headersLennart Poettering
Hence including "util.h" from sd-rtnl.h is not OK. Let's minimize our headers we pull in a bit.
2013-11-07rtnl: internal headers must include pragma once protectionLennart Poettering
2013-11-07build-sys: add .pc file for libsystemd-busLennart Poettering
2013-11-07build-sys: merge sd-event into sd-busLennart Poettering
The sd-event APIs should be available only as part of libsystemd-bus so that the utility calls are not linked into each independently and we can minimize the number of libraries we have.
2013-11-07build-sys: add Makefile symlink for src/activate/Lennart Poettering
2013-11-07utf8: export utf8 validation functions as part of sd-busLennart Poettering
To write useful bus code clients need to validate utf8 frequently since the bus reacts allergic to it. Since glibc does not provide any calls for this, let's provide it as part of libsystemd-bus.
2013-11-07utf8: ascii_filter() is unused, let's remove itLennart Poettering
2013-11-07consistently use "int" when retrieving "bool" from bus messagesKay Sievers
2013-11-06bus: add forgotten _public_Zbigniew Jędrzejewski-Szmek
2013-11-06pam_systemd: dup the fd received from logindZbigniew Jędrzejewski-Szmek
Otherwise sd_bus_message cleanup would close it.
2013-11-07machinectl: we need glibc's socket.h, not our ownLennart Poettering
2013-11-07pam: externally our booleans are ints, not unsignedsLennart Poettering
2013-11-07activate: port to use new getpeername_pretty()/getsockname_pretty() callsLennart Poettering
2013-11-07socket: rework things to have only one sockaddr formatterLennart Poettering
2013-11-07event: make sure we keep a reference to all events we dispatch while we do so.Lennart Poettering
2013-11-07activate: fix uninitialized memoryLennart Poettering
2013-11-07pam_systemd: pass 'unsigned' instead of 'bool' to sd_bus_message_read()Kay Sievers
2013-11-06mkdir: modernize header ifdefsLennart Poettering
2013-11-06socket-proxyd: rework to support multiple sockets and splice()-based ↵Lennart Poettering
zero-copy network IO This also drops --ignore-env, which can't really work anymore if we allow multiple fds. Also adds support for pretty printing of peer identities for debug purposes, and abstract namespace UNIX sockets. Also ensures that we never take more connections than a certain limit.
2013-11-06active: rework make_socket_fd() to be based on socket_address_listen()Lennart Poettering
Among other things this makes sure we set SO_REUSEADDR which is immensely useful.
2013-11-06activate: reduce log settings to normalLennart Poettering
2013-11-06activate: get rid of redundant definiton of fd_cloexec()Lennart Poettering
2013-11-06pam_systemd: port to libsystemd-busZbigniew Jędrzejewski-Szmek
2013-11-06systemd_pam: treat debug as debug=1 and parse all paramsZbigniew Jędrzejewski-Szmek
systemd_pam would ignore all params after the first invalid one. Instead ignore just this one, and parse the rest. There's just one now, but as a matter of principle ;) Also, allow debug as an alias for debug=1, and don't treat invalid debug= options as fatal.
2013-11-06bus: add public libsystemd-eventZbigniew Jędrzejewski-Szmek
2013-11-06bus: export libsystemd-bus as a public libraryZbigniew Jędrzejewski-Szmek
Old static libsystemd-bus.la becomes libsystemd-bus-internal.la. memfd functions are also exported in the same library. (Best viewed with --color-words -U0).
2013-11-06machinectl: use initialization instead of zeroingZbigniew Jędrzejewski-Szmek
2013-11-06efi-boot-generator: simplifyZbigniew Jędrzejewski-Szmek
2013-11-06socket-proxy: clean-up indentingLennart Poettering
2013-11-06clients: unify how we invoke getopt_long()Lennart Poettering
Among other things this makes sure we always expose a --version command and show it in the help texts.
2013-11-06path-util: paths_check_timestamp() opimizationsLennart Poettering
2013-11-06inhibit: don't return error code on --helpLennart Poettering
2013-11-06busctl: port over to use bus_open_transport()Lennart Poettering
2013-11-06clients: various simplificationsLennart Poettering
2013-11-06run: support --system to match other commands, even if redundantLennart Poettering
2013-11-06analyze: port over to use bus_open_transport()Lennart Poettering
2013-11-06analyze: simplify thingsLennart Poettering
2013-11-06analyze: fix indentationLennart Poettering
2013-11-06Revert "hostnamed: prevent it from crashing if the chassis is unknown"Kay Sievers
This reverts commit 95b015de3849d8d8d03d9441ae1682a04c38833b. "bus: handle serialization of NULL strings" (cd6f997f71c3aba16aa08226d423d14cbc787f82) should fix this issue.
2013-11-06Comment spelling fixes.David Strauss
2013-11-06util: unify reading of /proc/cmdlineLennart Poettering
Instead of individually checking for containers in each user do this once in a new call proc_cmdline() that read the file only if we are not in a container.
2013-11-06nspawn: explicitly terminate machines when we exit nspawnLennart Poettering
https://bugs.freedesktop.org/show_bug.cgi?id=68370 https://bugzilla.redhat.com/show_bug.cgi?id=988883
2013-11-06machined: keep track of the initial leader PID of a machineLennart Poettering
This way we can without races always determine the machine for a leader PID. This allows machine managers to query the machine for a forked off container/VM without a race where the child might already have died before we could read the cgroup information from /proc/$PID/cgroup.
2013-11-06bus: handle serialization of NULL stringsLennart Poettering
Instead of simply crashing be somewhat nicer and serialize a NULL string into the empty string and generate an error on signature and object path strings.
2013-11-06shutdown: trim the cgroup tree on loop iterationLennart Poettering
This way we leave the cgroup empty before exiting in a container which makes sure the container manager will get cgroup notification event https://bugs.freedesktop.org/show_bug.cgi?id=68370 https://bugzilla.redhat.com/show_bug.cgi?id=988883
2013-11-06cgroup: run PID 1 in the root cgroupLennart Poettering
This way cleaning up the cgroup tree on shutdown is a lot easier since we are in the root dir. Also PID 1 was previously artificially placed in system.slice, even though our rule actually was not to have processes in slices. The root slice otoh is magic anyway, so having PID 1 in there sounds less surprising. Of course, this means that PID is scheduled against the three top-level slices.
2013-11-06bus: test-bus-marshal - make dbus-1 optionalKay Sievers