chiark / gitweb /
elogind.git
11 years agotmpfiles: support globbing for w option
Dave Reisner [Mon, 3 Sep 2012 21:13:18 +0000 (17:13 -0400)]
tmpfiles: support globbing for w option

Break out the write logic into a separate function and simply use it as
a callback to glob_item.

This allows users to consolidate writes to sysfs with multiple similar
pathnames, e.g.

  w /sys/class/block/sd[a-z]/queue/read_ahead_kb - - - - 1024

11 years agotest: extend unit-name test a bit
Lennart Poettering [Wed, 12 Sep 2012 06:31:08 +0000 (08:31 +0200)]
test: extend unit-name test a bit

11 years agologin: reshuffle meaning of require_active parameter
Lennart Poettering [Wed, 12 Sep 2012 06:30:35 +0000 (08:30 +0200)]
login: reshuffle meaning of require_active parameter

11 years agologind: Avoid unnecesary rewrite of user file when switching sessions of the same...
Colin Guthrie [Tue, 4 Sep 2012 00:37:28 +0000 (01:37 +0100)]
logind: Avoid unnecesary rewrite of user file when switching sessions of the same user.

11 years agologind: Ensure the user, seat and session files are updated when the session is closing.
Colin Guthrie [Tue, 4 Sep 2012 00:37:27 +0000 (01:37 +0100)]
logind: Ensure the user, seat and session files are updated when the session is closing.

11 years agologind: Add a two new variables to the user session tracking file.
Colin Guthrie [Mon, 3 Sep 2012 22:57:58 +0000 (23:57 +0100)]
logind: Add a two new variables to the user session tracking file.

This counts 'online sessions' in addition to 'active sessions' and 'sessions'.

In this context, an 'online session' covers all session in the 'active' state
in addition to the explicit 'online' state.

This provides an easy machanism to determin all relevant sessions easily
(i.e. those that are not 'closing') and adds new semantics to the sd-login.c
APIs sd_uid_get_sessions() and sd_uid_get_seats() where the require_active
argument can be supplied as a value 2 which only lists sessions which are
'online'.

This functionality should allow client applications to avoid deadlocks where
they only exit when all sessions are complete, such as a the problem where
PulseAudio will not exit until all sessions are gone, but in itself prevents
the session from exiting.

11 years agologind: Properly list the ACTIVE_SEATS in the user session tracking file.
Colin Guthrie [Mon, 3 Sep 2012 22:47:02 +0000 (23:47 +0100)]
logind: Properly list the ACTIVE_SEATS in the user session tracking file.

Prevsiouly the first active seat for a user would never be listed and
any subsequent seats would be concatenated on without any spaces.

11 years agologind: If all user sessions are in closing state, set the overall status to closing.
Colin Guthrie [Mon, 3 Sep 2012 22:47:01 +0000 (23:47 +0100)]
logind: If all user sessions are in closing state, set the overall status to closing.

PulseAudio for example will keep a client connection open provided
at least one session exists. However, if all sessions are currently
in the process of closing, we should flag that as the overall state
appropriately to better reflect what is happening.

Although this does better reflect the status for any given user, it does
not actually solve the overall problem of PulseAudio still finding some
sessions active and thus not exiting and therefore actually preventing
the session from closing. Future commits will extend sd-login to cope
with this situation.

11 years agopam: Add session class to the debug log.
Colin Guthrie [Mon, 3 Sep 2012 22:47:00 +0000 (23:47 +0100)]
pam: Add session class to the debug log.

11 years agounit-name: rework unit_name_replace_instance function()
Lennart Poettering [Wed, 12 Sep 2012 02:46:38 +0000 (04:46 +0200)]
unit-name: rework unit_name_replace_instance function()

https://bugzilla.redhat.com/show_bug.cgi?id=855863

11 years agojournal: don't set seal flag if we don't have gcrypt built-in
Lennart Poettering [Tue, 11 Sep 2012 01:03:36 +0000 (03:03 +0200)]
journal: don't set seal flag if we don't have gcrypt built-in

11 years agojournal: don't try to compress without XZ
Michael Olbrich [Mon, 3 Sep 2012 13:46:44 +0000 (15:46 +0200)]
journal: don't try to compress without XZ

otherwise the header contains the HEADER_INCOMPATIBLE_COMPRESSED
flag even though the data is not compressed and reading the journal
fails.

11 years agoutil: rm_rf_children_dangerous: delete all descendants dangerously
Tom Gundersen [Mon, 3 Sep 2012 13:40:37 +0000 (15:40 +0200)]
util: rm_rf_children_dangerous: delete all descendants dangerously

Call rm_rf_children_dangerous() recursively rather than falling back to
rm_rf_children(). This fixes a bug in systemd-tmpfiles.

The problem can easily be reproduced by:

  # mount /dev/sda1 /mnt
  # mkdir /mnt/test
  # echo "D /mnt" > /root/test.conf
  # systemd-tmpfiles --remove /root/test.conf
  Attempted to remove disk file system, and we can't allow that.
  rm_rf(/root/test): Operation not permitted

Reported-by: Lukas Jirkovsky <l.jirkovsky@gmail.com>
11 years agoupdate TODO
Lennart Poettering [Mon, 10 Sep 2012 23:45:27 +0000 (01:45 +0200)]
update TODO

11 years agobuild-sys: perform autodetection of quota-tools binaries
Dave Reisner [Sun, 2 Sep 2012 21:28:30 +0000 (17:28 -0400)]
build-sys: perform autodetection of quota-tools binaries

Use AC_PATH_PROG to try and locate the quotaon and quotacheck binaries,
falling back on hardcoded defaults when they can't be found.

11 years agocondition: add ConditionFileNotEmpty=
Lennart Poettering [Mon, 10 Sep 2012 23:29:46 +0000 (01:29 +0200)]
condition: add ConditionFileNotEmpty=

https://bugs.freedesktop.org/show_bug.cgi?id=54448

11 years agowhen determining unit file list, include invalid unit names in an "invalid" state
Lennart Poettering [Mon, 10 Sep 2012 23:11:32 +0000 (01:11 +0200)]
when determining unit file list, include invalid unit names in an "invalid" state

11 years agoupdate TODO
Lennart Poettering [Mon, 10 Sep 2012 22:23:05 +0000 (00:23 +0200)]
update TODO

11 years agoupdate TODO
Lennart Poettering [Mon, 10 Sep 2012 20:45:20 +0000 (22:45 +0200)]
update TODO

11 years agoman: document that assignemnts of devices to seats are persistent
Lennart Poettering [Mon, 10 Sep 2012 11:20:45 +0000 (13:20 +0200)]
man: document that assignemnts of devices to seats are persistent

11 years agojournald: don't do debug logging by default
Lennart Poettering [Mon, 10 Sep 2012 11:18:39 +0000 (13:18 +0200)]
journald: don't do debug logging by default

11 years agojournald: don't complain if fss key does not exist
Lennart Poettering [Mon, 10 Sep 2012 11:18:22 +0000 (13:18 +0200)]
journald: don't complain if fss key does not exist

11 years agoselinux: close stdin/stdout/stderr before loading selinux policy
Lennart Poettering [Mon, 10 Sep 2012 11:14:56 +0000 (13:14 +0200)]
selinux: close stdin/stdout/stderr before loading selinux policy

Let's make sure libselinux doesn't spew to stderr

11 years agojournald: detect invalid header pointers correctly
Lennart Poettering [Mon, 10 Sep 2012 09:58:00 +0000 (11:58 +0200)]
journald: detect invalid header pointers correctly

11 years agounit: don't allow units to be gc'ed that still are referenced via UnitRef
Lennart Poettering [Mon, 10 Sep 2012 08:12:10 +0000 (10:12 +0200)]
unit: don't allow units to be gc'ed that still are referenced via UnitRef

11 years agoinhibit: close all fds when invoking inhibited tool
Lennart Poettering [Mon, 10 Sep 2012 07:45:55 +0000 (09:45 +0200)]
inhibit: close all fds when invoking inhibited tool

11 years agodbus: minor coding style fixes
Lennart Poettering [Mon, 10 Sep 2012 07:38:49 +0000 (09:38 +0200)]
dbus: minor coding style fixes

11 years agoMake systemd-inhibit --list work
Matthias Clasen [Sat, 8 Sep 2012 02:56:35 +0000 (22:56 -0400)]
Make systemd-inhibit --list work

The code in the print_inhibitors function had an unintended
unconditional early exit, causing it to never print any
inhibitors.

11 years agobuild-sys: add libsystemd-id128-internal.la as dep to libsystemd-core.la
Lennart Poettering [Mon, 10 Sep 2012 07:28:42 +0000 (09:28 +0200)]
build-sys: add libsystemd-id128-internal.la as dep to libsystemd-core.la

condition.c makes use of sd_id128_get_machine() after all.

Spotted by Khem Raj.

11 years agopath-util: fall back to stat() if kernel does not support name_to_handle_at()
Mantas Mikulėnas [Sun, 9 Sep 2012 12:16:19 +0000 (15:16 +0300)]
path-util: fall back to stat() if kernel does not support name_to_handle_at()

Fixes instant hang on kernels that do not have CONFIG_FHANDLE enabled.

11 years agopath-util: do not fail in path_is_mountpoint() if path doesn't exist
Mantas Mikulėnas [Sun, 9 Sep 2012 12:16:18 +0000 (15:16 +0300)]
path-util: do not fail in path_is_mountpoint() if path doesn't exist

This was accidentally lost in commit 1640a0b6b05b.

11 years agojournald: make splitting up of journal files per-user configurable
Lennart Poettering [Fri, 7 Sep 2012 21:40:00 +0000 (23:40 +0200)]
journald: make splitting up of journal files per-user configurable

11 years agojournal: add call to determine current journal file disk usage
Lennart Poettering [Fri, 7 Sep 2012 21:20:28 +0000 (23:20 +0200)]
journal: add call to determine current journal file disk usage

11 years agomain: fix syntax of net_prio cgroup controller
Lennart Poettering [Fri, 7 Sep 2012 20:22:15 +0000 (22:22 +0200)]
main: fix syntax of net_prio cgroup controller

11 years agojournald: avoid logging to kmsg in the normal paths
Lennart Poettering [Thu, 6 Sep 2012 07:32:51 +0000 (00:32 -0700)]
journald: avoid logging to kmsg in the normal paths

11 years agolocaled: system-setup-keyboard is no more on fedora
Lennart Poettering [Thu, 6 Sep 2012 06:54:09 +0000 (23:54 -0700)]
localed: system-setup-keyboard is no more on fedora

11 years agoman: document that ExecStart= doesn't understand shell command lines
Lennart Poettering [Thu, 6 Sep 2012 06:51:19 +0000 (23:51 -0700)]
man: document that ExecStart= doesn't understand shell command lines

11 years agotmpfiles: don't attempt creation of device nodes when we run in a container
Lennart Poettering [Thu, 6 Sep 2012 06:39:55 +0000 (23:39 -0700)]
tmpfiles: don't attempt creation of device nodes when we run in a container

11 years agosystemctl: properly build flags for show_journal_by_unit()
Lennart Poettering [Thu, 6 Sep 2012 00:05:04 +0000 (17:05 -0700)]
systemctl: properly build flags for show_journal_by_unit()

11 years agojournalctl: show "Reboot" markers in output only when showing local-only entries
Lennart Poettering [Wed, 5 Sep 2012 23:52:46 +0000 (16:52 -0700)]
journalctl: show "Reboot" markers in output only when showing local-only entries

11 years agojournalctl: replace --local by --merge, i.e. don't interleave remote journals by...
Lennart Poettering [Wed, 5 Sep 2012 23:49:00 +0000 (16:49 -0700)]
journalctl: replace --local by --merge, i.e. don't interleave remote journals by default

11 years agonspawn: handle poweroff/reboot nicely in containers
Lennart Poettering [Wed, 5 Sep 2012 23:23:41 +0000 (16:23 -0700)]
nspawn: handle poweroff/reboot nicely in containers

11 years agoshutdown: in containers, invoke reboot(2), too. Then fallback to exit() if CAP_SYS_BO...
Lennart Poettering [Wed, 5 Sep 2012 22:37:18 +0000 (15:37 -0700)]
shutdown: in containers, invoke reboot(2), too. Then fallback to exit() if CAP_SYS_BOOT is missing

The kernel's PID namespaces support reboot(2) just fine, so let's make
use of it if possible.

11 years agoshutdown: remove explicit sync() invocations
Lennart Poettering [Wed, 5 Sep 2012 22:32:57 +0000 (15:32 -0700)]
shutdown: remove explicit sync() invocations

The kernel implicitly does sync() anyway, hence there is no need to do
that in userspace explicitly. This makes the "-n" switch to halt(8) a
noop.

11 years agonspawn: don't provide /dev/rtc0 in the container
Lennart Poettering [Wed, 5 Sep 2012 22:27:07 +0000 (15:27 -0700)]
nspawn: don't provide /dev/rtc0 in the container

Since RTCs are hardware devices and are very much shared resources we
should avoid to provide them in each container.

11 years agojournalctl: rework JSON output mode
Lennart Poettering [Wed, 5 Sep 2012 22:25:32 +0000 (15:25 -0700)]
journalctl: rework JSON output mode

This splits the JSON output mode into different modes: json and
json-pretty. The former printing one entry per line, the latter showing
JSON objects nicely indented and in multiple lines to make it easier to
read for humans.

11 years agonspawn: generate a new randomized boot ID for each container
Lennart Poettering [Wed, 5 Sep 2012 21:39:16 +0000 (14:39 -0700)]
nspawn: generate a new randomized boot ID for each container

11 years agonspawn: if a file system comes pre-mounted, still do the read-only remounts
Lennart Poettering [Wed, 5 Sep 2012 21:16:41 +0000 (14:16 -0700)]
nspawn: if a file system comes pre-mounted, still do the read-only remounts

11 years agopath: fix invalid variable access in path_is_mount_point()
Lennart Poettering [Wed, 5 Sep 2012 00:05:23 +0000 (17:05 -0700)]
path: fix invalid variable access in path_is_mount_point()

11 years agopath-util: in path_is_mount_point() fall back to the classic stat() test if fs does...
Lennart Poettering [Tue, 4 Sep 2012 23:37:57 +0000 (16:37 -0700)]
path-util: in path_is_mount_point() fall back to the classic stat() test if fs does not support name_to_handle_at()

11 years agonspawn: skip mounts if already mounted
Lennart Poettering [Tue, 4 Sep 2012 23:33:13 +0000 (16:33 -0700)]
nspawn: skip mounts if already mounted

11 years agonspawn: mount a clean instance of sysfs
Lennart Poettering [Tue, 4 Sep 2012 23:32:43 +0000 (16:32 -0700)]
nspawn: mount a clean instance of sysfs

11 years agomissing: define name_to_handle_at on our own if it is missing
Lennart Poettering [Tue, 4 Sep 2012 22:20:04 +0000 (15:20 -0700)]
missing: define name_to_handle_at on our own if it is missing

11 years agoupdate TODO
Lennart Poettering [Tue, 4 Sep 2012 20:39:18 +0000 (13:39 -0700)]
update TODO

11 years agoshared: logs-show: fflush after each output type
Brandon Philips [Tue, 4 Sep 2012 01:37:46 +0000 (18:37 -0700)]
shared: logs-show: fflush after each output type

journalctl -f redirected to a pipe or file wasn't working for some
output formats but was working for json. It turns out only json was
doing an fflush.

Make all output formats flush.

11 years agoupdate TODO -- hackfest edition
Lennart Poettering [Tue, 4 Sep 2012 01:58:18 +0000 (18:58 -0700)]
update TODO -- hackfest edition

11 years agomain: jointly mount more controllers
Lennart Poettering [Tue, 4 Sep 2012 01:54:32 +0000 (18:54 -0700)]
main: jointly mount more controllers

After talking to the cgroup kernel folks at LPC we came to the
conclusion that it is probably a good idea to mount all CPU related
resp. all network related cgroup controllers together, both because they
are good defaults for admins and because this might prepare
for eventual kernel cleanups where the ability to mount them separately
is removed.

11 years agojournalctl: include logs from PID 1 about services in systemctl status
Lennart Poettering [Fri, 24 Aug 2012 22:55:22 +0000 (00:55 +0200)]
journalctl: include logs from PID 1 about services in systemctl status

11 years agoreadahead: a bit of reformatting
Lennart Poettering [Fri, 24 Aug 2012 22:16:19 +0000 (00:16 +0200)]
readahead: a bit of reformatting

11 years agoutil: make path_is_mount_point() recognize bind mounts, too
Lennart Poettering [Fri, 24 Aug 2012 22:10:17 +0000 (00:10 +0200)]
util: make path_is_mount_point() recognize bind mounts, too

11 years agojournal: don't pull in flush service from journald service
Lennart Poettering [Fri, 24 Aug 2012 21:23:08 +0000 (23:23 +0200)]
journal: don't pull in flush service from journald service

In the initrd we don't need the flush service hence don't attempt to
pull it in.

11 years agounit: make sure structured logging is available before we enter the main boot phase
Lennart Poettering [Fri, 24 Aug 2012 21:18:35 +0000 (23:18 +0200)]
unit: make sure structured logging is available before we enter the main boot phase

11 years agoswap: don't try to shutdown swaps on shutdown in a container
Lennart Poettering [Fri, 24 Aug 2012 21:07:03 +0000 (23:07 +0200)]
swap: don't try to shutdown swaps on shutdown in a container

11 years agoswap: refuse start/stop of swaps in container
Lennart Poettering [Fri, 24 Aug 2012 21:00:13 +0000 (23:00 +0200)]
swap: refuse start/stop of swaps in container

11 years agoshutdown: don't attempt read-only mounts in a container
Lennart Poettering [Fri, 24 Aug 2012 20:50:16 +0000 (22:50 +0200)]
shutdown: don't attempt read-only mounts in a container

11 years agojournal: suppress structured messages if they'd go to the console
Lennart Poettering [Fri, 24 Aug 2012 20:43:33 +0000 (22:43 +0200)]
journal: suppress structured messages if they'd go to the console

11 years agojournal: generate structured journal messages for a number of events
Lennart Poettering [Fri, 24 Aug 2012 20:21:20 +0000 (22:21 +0200)]
journal: generate structured journal messages for a number of events

11 years agoshared: in code that might get called from suid programs use __secure_getenv() rather...
Lennart Poettering [Thu, 23 Aug 2012 16:47:01 +0000 (18:47 +0200)]
shared: in code that might get called from suid programs use __secure_getenv() rather than getenv()

It's better to be safe than sorry.

11 years agobuild-sys: xsltproc is required for man pages and gtk-doc
Zbigniew Jędrzejewski-Szmek [Fri, 24 Aug 2012 23:57:03 +0000 (01:57 +0200)]
build-sys: xsltproc is required for man pages and gtk-doc

11 years agojournal: add HAVE_XZ check to avoid build failure
Yin Kangkai [Fri, 24 Aug 2012 08:52:58 +0000 (16:52 +0800)]
journal: add HAVE_XZ check to avoid build failure

Without this, build fail with --disable-xz or xz not installed.

11 years agokeymaps: Add "e" key for Acer Aspire 5720
Martin Pitt [Fri, 24 Aug 2012 05:11:20 +0000 (07:11 +0200)]
keymaps: Add "e" key for Acer Aspire 5720

https://launchpad.net/bugs/1009836

11 years agojournald: add missing includes
Allin Cottrell [Thu, 23 Aug 2012 23:46:38 +0000 (01:46 +0200)]
journald: add missing includes

11 years agosd-journal: return null when mmap_cache_new fails
Lukas Nykryn [Thu, 23 Aug 2012 10:28:47 +0000 (12:28 +0200)]
sd-journal: return null when mmap_cache_new fails

11 years agohashmap: hashmap_contains does not need hashmap_entry
Lukas Nykryn [Thu, 23 Aug 2012 10:21:55 +0000 (12:21 +0200)]
hashmap: hashmap_contains does not need hashmap_entry

11 years agoreadahead-analyze: don't call fclose on null
Lukas Nykryn [Mon, 20 Aug 2012 13:15:40 +0000 (15:15 +0200)]
readahead-analyze: don't call fclose on null

11 years agocore: free word later in parse_proc_cmdline
Lukas Nykryn [Mon, 20 Aug 2012 12:52:07 +0000 (14:52 +0200)]
core: free word later in parse_proc_cmdline

11 years agologin: check return of parse_pid and parse_uid
Lukas Nykryn [Mon, 20 Aug 2012 12:39:08 +0000 (14:39 +0200)]
login: check return of parse_pid and parse_uid

11 years agoload-fragment: initialize bool invert before use
Lukas Nykryn [Mon, 20 Aug 2012 12:33:21 +0000 (14:33 +0200)]
load-fragment: initialize bool invert before use

11 years agobuild-sys: prepare release 189 v189
Lennart Poettering [Thu, 23 Aug 2012 00:46:22 +0000 (02:46 +0200)]
build-sys: prepare release 189

11 years agoservice: introduce a proper service result if the start limit is hit
Lennart Poettering [Wed, 22 Aug 2012 03:02:33 +0000 (05:02 +0200)]
service: introduce a proper service result if the start limit is hit

11 years agojournalctl: complain if unprivileged users attempt to access the journal and persista...
Lennart Poettering [Wed, 22 Aug 2012 02:45:17 +0000 (04:45 +0200)]
journalctl: complain if unprivileged users attempt to access the journal and persistant logging is off

11 years agojournal: the ratelimiter is part of journald
Lennart Poettering [Wed, 22 Aug 2012 01:43:45 +0000 (03:43 +0200)]
journal: the ratelimiter is part of journald

11 years agojournald: split off native protocol support into its own .c file
Lennart Poettering [Wed, 22 Aug 2012 01:42:23 +0000 (03:42 +0200)]
journald: split off native protocol support into its own .c file

11 years agojournald: split console transport stuff into its own file
Lennart Poettering [Wed, 22 Aug 2012 01:36:29 +0000 (03:36 +0200)]
journald: split console transport stuff into its own file

11 years agojournald: move stream protocol into its own .c file
Lennart Poettering [Wed, 22 Aug 2012 01:32:41 +0000 (03:32 +0200)]
journald: move stream protocol into its own .c file

11 years agojournal: move syslog specific calls out of util.c
Lennart Poettering [Wed, 22 Aug 2012 01:24:46 +0000 (03:24 +0200)]
journal: move syslog specific calls out of util.c

11 years agojournald: splitt of syslog protocol support into its own file
Lennart Poettering [Wed, 22 Aug 2012 01:21:03 +0000 (03:21 +0200)]
journald: splitt of syslog protocol support into its own file

11 years agojournald: split /dev/kmsg related stuff into its own .c file
Lennart Poettering [Wed, 22 Aug 2012 01:09:44 +0000 (03:09 +0200)]
journald: split /dev/kmsg related stuff into its own .c file

11 years agojournald: write tags also to user journal files
Lennart Poettering [Wed, 22 Aug 2012 00:53:51 +0000 (02:53 +0200)]
journald: write tags also to user journal files

11 years agojournald: augment journal entries from the kernel with data from udev
Lennart Poettering [Wed, 22 Aug 2012 00:49:17 +0000 (02:49 +0200)]
journald: augment journal entries from the kernel with data from udev

11 years agounit: add new ConditionHost= condition type
Lennart Poettering [Tue, 21 Aug 2012 23:51:53 +0000 (01:51 +0200)]
unit: add new ConditionHost= condition type

11 years agomain: fix PR_SET_CHILD_SUBREAPER warning version
Lennart Poettering [Tue, 21 Aug 2012 21:07:21 +0000 (23:07 +0200)]
main: fix PR_SET_CHILD_SUBREAPER warning version

11 years agojournalctl: be more friendly when informing about seal verification
Lennart Poettering [Tue, 21 Aug 2012 21:03:20 +0000 (23:03 +0200)]
journalctl: be more friendly when informing about seal verification

11 years agoshared/util: add fstab support for partuuid/partlabel
Dave Reisner [Sat, 11 Aug 2012 17:32:29 +0000 (13:32 -0400)]
shared/util: add fstab support for partuuid/partlabel

udev has supported this since 172, so it should be a safe (and welcome)
addition for users of GPT partitioned disks.

11 years agoshared/util: refactor fstab_node_to_udev_node
Dave Reisner [Sat, 11 Aug 2012 17:31:50 +0000 (13:31 -0400)]
shared/util: refactor fstab_node_to_udev_node

Since this is purely duplicated logic, separate it out into a small
static function.

11 years agoupdate TODO
Lennart Poettering [Tue, 21 Aug 2012 15:38:29 +0000 (17:38 +0200)]
update TODO

11 years agodev-setup: make NULL as parameter for dev_setup() equivalent to ""
Lennart Poettering [Tue, 21 Aug 2012 15:23:03 +0000 (17:23 +0200)]
dev-setup: make NULL as parameter for dev_setup() equivalent to ""

11 years agonspawn: add /dev FD symlinks in container setup
Dave Reisner [Wed, 15 Aug 2012 00:00:31 +0000 (20:00 -0400)]
nspawn: add /dev FD symlinks in container setup

This creates /dev/fd, /dev/stdin, /dev/stdout, /dev/stderr, and
/dev/core as symlinks to /proc on container creation. Except for
/dev/core, these are needed for shells like bash to be fully functional.

11 years agodev-setup: allow a path prefix for use in chroots
Dave Reisner [Wed, 15 Aug 2012 00:00:30 +0000 (20:00 -0400)]
dev-setup: allow a path prefix for use in chroots

With this adjustment, we can reuse this code elsewhere, such as in
nspawn.

11 years agoupdate TODO
Lennart Poettering [Tue, 21 Aug 2012 15:19:32 +0000 (17:19 +0200)]
update TODO