coverity: #1425858

free allocated memory

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425859

check return value of snprintf()

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425860

remove logically dead code

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425861

free allocated memory

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425862

initialize handler

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425863

remove logically dead code

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425866

free allocated memory

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425867

do not pass NULL pointer to chdir()

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425869

do not unmap prematurely

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425870

check snprintf() return value

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425879

do not double close file descriptor

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425883

ensure \0-termination

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425884

free allocated memory

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1428855

remove logically dead code

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425886

free allocated memory

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425893

used calculated string length when copying into buffer

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1426029

fix use after free

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

Merge pull request #2009 from lifeng68/add_sysctl

confile: add lxc.sysctl config

confile: Add lxc.sysctl config

Signed-off-by: LiFeng <lifeng68@huawei.com>

Merge pull request #2019 from brauner/2017-12-11/fix_legacy_network_parser

legacy confile: fix legacy network parser

conf: fix lxc.prlimit clearing

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

legacy confile: fix legacy network parser

This fixes a bug introduced by:

commit 94f0035bf636ba853451d59c129e26d94850c04d
Author: Christian Brauner <christian.brauner@ubuntu.com>
Date:   Thu Dec 7 15:07:26 2017 +0100

    coverity: #1425924

    remove logically dead condition

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>
Coverity's bug analysis is correct but my fix wasn't.

This commit fixes a bunch of other bugs I just spotted as well.

This unblocks #2009.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

travis: Fix build failure

Signed-off-by: Stéphane Graber <stgraber@ubuntu.com>

Merge pull request #2016 from flx42/lxc-oci-misc-improvements

lxc-oci misc improvements

lxc-oci: add support for WorkingDir

Signed-off-by: Felix Abecassis <fabecassis@nvidia.com>

Merge pull request #2012 from marcosps/pthread_mtx_msg

lxc: Distinguish pthread_mutex_unlock error messages

Merge pull request #1950 from brauner/2017-11-27/criu_fixes

bugfixes

lxc-oci: add IPv6 support to /etc/hosts

Signed-off-by: Felix Abecassis <fabecassis@nvidia.com>

lxc-oci: add basic handling of numerical uid/gid

Signed-off-by: Felix Abecassis <fabecassis@nvidia.com>

lxc: Distinguish pthread_mutex_unlock error messages

The same message exists in lxclock.c and cgmanager.c, so print the
filename along with the message.

Before this patch:
lxc-destroy -n u1
pthread_mutex_unlock returned:1 Operation not permitted

After this patch:
xc-destroy -n u1
lxclock.c: pthread_mutex_unlock returned:1 Operation not permitted

Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>

Merge pull request #2011 from brauner/generic/coverity

coverity: bugfixes

coverity: #1425888

check return value of getifaddrs()

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425889

do not call close on bad file descriptor

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425890

silence complaint about unchecked return value from library function

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425895

free allocated memory

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425921

free allocated memory

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425922

free allocated memory

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425923

avoid NULL-pointer dereference

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425924

remove logically dead condition

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425928

Avoid NULL-pointer dereference. Apparently monitor.{c,h} calls
lxc_check_inherited() with NULL passed for the config. This isn't really a big
issue since monitor.{c,h} is effectively dead for all liblxc versions that have
the state client patch. Also, the patch that introduces the relevant lines into
lxc_check_inherited() is only in master and yet unreleased.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425929

free allocated memory

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

Merge pull request #2007 from brauner/2017-12-06/bugfixes

coverity: bugfixes

Merge pull request #1934 from brauner/2017-11-21/implement_do_lxc_reboot_correctly

commands: improve and simplify locking + lxccontainer: add reboot2() API extension

coverity: #1425743

remove logically dead code

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425742

fix error condition check

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425740

free allocated memory

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425739

check snprintf() return value

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425737

free allocated memory

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425735

use correct inequality comparison

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

coverity: #1425734

free memory on error

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

tests: expand tests for shortlived init processes

This adds additional test for app containers.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

Merge pull request #2006 from Ricardo-Yoshi-LXC/master

Move lxc_init up to lxc/src and add argument parser

lxc_init: changed minor details

Signed-off-by: RicardoSanchezA <ricardo.sanchez@utexas.edu>

lxc_init: add custom argument parser

lxc_init.c should not depend on tools/arguments.{c,h}, thus it needs its own custom argument parser

Signed-off-by: RicardoSanchezA <ricardo.sanchez@utexas.edu>

lxc_init: move up to src/lxc

lxc_init.c is actually not a tool, thus it doesn't belong to the tools/ subfolder

Signed-off-by: RicardoSanchezA <ricardo.sanchez@utexas.edu>

lxccontainer: handle execute containers correctly

It doesn't make sense to error out when an app container doesn't pass explicit
arguments through c->start{l}(). This is especially true since we implemented
lxc.execute.cmd. However, even before we could have always relied on
lxc.init.cmd and errored out after that.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

execute: pass logfile to lxc-init

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

commands: return -ECONNRESET to caller

Callers can then make a decision whether they want to consider the peer closing
the connection an error or not. For example, a c->wait(c, "STOPPED", -1) call
can then consider a ECONNRESET not an error but rather see it - correctly - as
a container exiting before being able to register a state client.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

commands: tell mainloop to reap client fd on error

This is the proper way to handle errors.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

lxccontainer: restore non-blocking shutdown

If timeout is set to 0 don't block.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

commands: tweak locking

Take the lock on the list after we've done all necessary work and check state.
If we are in requested state, do cleanup and return without adding the state
client to the state client list.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

test: add state server tests

This checks whether multiple concurrent waiters all get notified by the state
server.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

test: add test for reboot2() API extension

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

lxccontainer: add reboot2() API extension

This adds reboot2() as a new API extension. This function properly wait until a
reboot succeeded. It takes a timeout argument. When set to > 0 reboot2() will
block until the timeout is reached, if timeout is set to zero reboot2() will
not block, if set to -1 reboot2() will block indefinitly.

The struct state_client gets rename to lxc_state_client since it's more in line
with other declarations. It also gets moved from the lxc_handler to the
lxc_conf struct so that the state clients waiting for reboots don't get
deallocated on reboot since the handler is deallocated on reboot.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

commands: allow waiting for all states

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

start: don't lock setting the state

- setting the handler->state value is atomic on any POSIX implementation since
  we're dealing with an integer (enum/lxc_state_t)
- while the state clients are served it is not possible for lxc_set_state() to
  transition to the next state anyway so there's no danger in moving to the
  next state with clients missing it
- we only care about the list being modified

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

commands: don't lock the whole command

There are multiple reasons why this is not required:
- every command is transactional
- we only care about the list being modified not the memory allocation and
  other costly operations

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

commands: don't lock atomic operations

We're dealing with an integer (lxc_state_t which is an enum). Any POSIX
implementation makes those operations atomic so there's not need in locking
this.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

commands: don't traverse whole list

When we remove a state client fd there's not reason to walk the whole list. We
can simply break once we found and removed the fd.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

Merge pull request #1994 from brauner/2017-12-04/bugfixes

attach: do not fail on non-existing namespaces

Merge pull request #1996 from brauner/2017-12-04/reenable_coverity_scan

test: Coverity scan integration

Merge pull request #2004 from brauner/2017-12-06/fix_log_env_variable

start: set loglevel correctly

start: set loglevel correctly

We want the loglevel that the user specified when starting the container
because it overrides the default one set in the config.

Closes #2003.

Reported-by: Felix Abecassis <fabecassis@nvidia.com>
Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

arguments: move to tools/ subdirectory

The arguments.{c,h} file should not be a part of liblxc itself. It should only
be used for the lxc tools.

Signed-off-by: RicardoSanchezA <ricardo.sanchez@utexas.edu>
Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

Merge pull request #2002 from tenforward/japanese

Add lxc.console.*, lxc.namespace.*, and lxc.init.cwd to Japanese lxc.container.conf(5)

doc: Add lxc.init.cwd to Japanese lxc.container.conf(5)

Signed-off-by: KATOH Yasufumi <karma@jazz.email.ne.jp>

doc: Add lxc.namespace.[namespace identifier] to Japanese lxc.container.conf(5)

and fix the description of namespace inheritance

Signed-off-by: KATOH Yasufumi <karma@jazz.email.ne.jp>

doc: Add lxc.console.buffer.* and lxc.console.rotate to Japanese lxc.container.conf(5)

and update the description of lxc.console.logfile

Signed-off-by: KATOH Yasufumi <karma@jazz.email.ne.jp>

Merge pull request #1992 from lifeng68/add_init_cwd

confile:add lxc.init.cwd

confile:add lxc.init.cwd

Signed-off-by: LiFeng <lifeng68@huawei.com>

Merge pull request #2000 from hallyn/2017-12-04/execargs

lxc_execute: properly figure out number of needed arguments

lxc_execute: properly figure out number of needed arguments

The lxc_log args have noghing to do with lxc.rootfs.path, and
we need room for a NULL at end of arguments.

How this bug was hidden for so long I don't know - I can only
trigger it on certain systems under certain conditions, but it's
definately wrong as is.

Signed-off-by: Serge Hallyn <shallyn@cisco.com>

test: reenable Coverity integration

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

attach: do not fail on non-existing namespaces

Closes #1993.

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

Merge pull request #1979 from marcosps/issue_494

lxc_unshare: Add uid_mapping when creating userns

Merge pull request #1988 from abbasally5/feature-improve-bash-completion-1236

Add bash completion to list backing store types for lxc-create -B

Merge pull request #1989 from p-l-/fix-lxc-checkconfig

Fix lxc-checkconfig output

Merge pull request #1981 from brauner/2017-12-01/fix_android_sethostname

utils: declare sethostname() static inline

criu: silence static analysis

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

confile: remove dead assignment

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

criu: initialize status

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

confile_legacy: prevent null pointer deref

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

conf: prevent null pointer dereference

Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>

Insert missing "echo" after "is_enabled"

Signed-off-by: Pierre LALET <pierre.lalet@cea.fr>

Fix SETCOLOR_FAILURE evaluation

Signed-off-by: Pierre LALET <pierre.lalet@cea.fr>

Add bash completion to list backing store types for lxc-create -B
- Backing Store types are hard-coded (Not sure how to get programmatically)
- Closes #1236

Signed-off-by: Abbas Ally <abbasally5@yahoo.com>

Merge pull request #1984 from CC-Hsu/patch-1

Add new dependency to Slackware template

Add new dependency to Slackware template

I followed the [changelog of Slackware-current]<http://www.slackware.com/changelog/>,
and found that Slackware-current split hostname utility from util-linux package in Nov 17 2017.
So I add the new package to the template.

Signed-off-by: Chia-Chun Hsu <a12321aabb@gmail.com>

Merge pull request #1983 from iDarkTemplar/update_template

Update gentoo.moresecure.conf.

Update gentoo.moresecure.conf.

Closes https://github.com/lxc/lxc/issues/1928

Signed-off-by: i.Dark_Templar <darktemplar@dark-templar-archives.net>