shaba/libvirt
1
0
Fork 0
mirror of https://gitlab.com/libvirt/libvirt.git synced 2025-01-17 06:03:52 +03:00
Code Releases Activity
libvirt/tests/vircgrouptest.c

605 lines
22 KiB
C
Raw Normal View History

Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
/*
* Copyright (C) 2013 Red Hat, Inc.
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library. If not, see
* <http://www.gnu.org/licenses/>.
*
* Author: Daniel P. Berrange <berrange@redhat.com>
*/
#include <config.h>
#include "testutils.h"
#ifdef __linux__
# include <stdlib.h>
# define __VIR_CGROUP_ALLOW_INCLUDE_PRIV_H__
# include "vircgrouppriv.h"
virutil: Move string related functions to virstring.c The source code base needs to be adapted as well. Some files include virutil.h just for the string related functions (here, the include is substituted to match the new file), some include virutil.h without any need (here, the include is removed), and some require both.
2013-04-03 12:36:23 +02:00
# include "virstring.h"
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
# include "virerror.h"
# include "virlog.h"
# include "virfile.h"
# define VIR_FROM_THIS VIR_FROM_NONE
static int validateCgroup(virCgroupPtr cgroup,
const char *expectPath,
const char **expectMountPoint,
Track symlinks for co-mounted cgroup controllers If a cgroup controller is co-mounted with another, eg /sys/fs/cgroup/cpu,cpuacct Then it is a requirement that there exist symlinks at /sys/fs/cgroup/cpu /sys/fs/cgroup/cpuacct pointing to the real mount point. Add support to virCgroupPtr to detect and track these symlinks Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-05 11:28:04 +01:00
const char **expectLinkPoint,
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
const char **expectPlacement)
{
Convert 'int i' to 'size_t i' in tests/ files Convert the type of loop iterators named 'i', 'j', k', 'ii', 'jj', 'kk', to be 'size_t' instead of 'int' or 'unsigned int', also santizing 'ii', 'jj', 'kk' to use the normal 'i', 'j', 'k' naming Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-08 15:09:33 +01:00
size_t i;
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
if (STRNEQ(cgroup->path, expectPath)) {
fprintf(stderr, "Wrong path '%s', expected '%s'\n",
cgroup->path, expectPath);
return -1;
}
tests/: Remove the whitespace before ";"
2013-05-21 15:53:48 +08:00
for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
if (STRNEQ_NULLABLE(expectMountPoint[i],
cgroup->controllers[i].mountPoint)) {
fprintf(stderr, "Wrong mount '%s', expected '%s' for '%s'\n",
cgroup->controllers[i].mountPoint,
expectMountPoint[i],
virCgroupControllerTypeToString(i));
return -1;
}
Track symlinks for co-mounted cgroup controllers If a cgroup controller is co-mounted with another, eg /sys/fs/cgroup/cpu,cpuacct Then it is a requirement that there exist symlinks at /sys/fs/cgroup/cpu /sys/fs/cgroup/cpuacct pointing to the real mount point. Add support to virCgroupPtr to detect and track these symlinks Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-05 11:28:04 +01:00
if (STRNEQ_NULLABLE(expectLinkPoint[i],
cgroup->controllers[i].linkPoint)) {
fprintf(stderr, "Wrong link '%s', expected '%s' for '%s'\n",
cgroup->controllers[i].linkPoint,
expectLinkPoint[i],
virCgroupControllerTypeToString(i));
return -1;
}
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
if (STRNEQ_NULLABLE(expectPlacement[i],
cgroup->controllers[i].placement)) {
fprintf(stderr, "Wrong placement '%s', expected '%s' for '%s'\n",
cgroup->controllers[i].placement,
expectPlacement[i],
virCgroupControllerTypeToString(i));
return -1;
}
}
return 0;
}
const char *mountsSmall[VIR_CGROUP_CONTROLLER_LAST] = {
[VIR_CGROUP_CONTROLLER_CPU] = "/not/really/sys/fs/cgroup/cpu,cpuacct",
[VIR_CGROUP_CONTROLLER_CPUACCT] = "/not/really/sys/fs/cgroup/cpu,cpuacct",
[VIR_CGROUP_CONTROLLER_CPUSET] = NULL,
[VIR_CGROUP_CONTROLLER_MEMORY] = "/not/really/sys/fs/cgroup/memory",
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
[VIR_CGROUP_CONTROLLER_FREEZER] = NULL,
[VIR_CGROUP_CONTROLLER_BLKIO] = NULL,
Add support for systemd cgroup mount Systemd uses a named cgroup mount for tracking processes. Add it as another type of controller, albeit one which we have to special case in a number of places. In particular we must never create/delete directories there, nor add tasks. Essentially the systemd mount is to be considered read-only for libvirt. With this change both the virCgroupDetectPlacement and virCgroupCopyPlacement methods must be invoked. The copy placement method will copy setup for resource controllers only. The detect placement method will probe for any named controllers, or resource controllers not already setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-25 19:13:44 +01:00
[VIR_CGROUP_CONTROLLER_SYSTEMD] = NULL,
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
};
const char *mountsFull[VIR_CGROUP_CONTROLLER_LAST] = {
[VIR_CGROUP_CONTROLLER_CPU] = "/not/really/sys/fs/cgroup/cpu,cpuacct",
[VIR_CGROUP_CONTROLLER_CPUACCT] = "/not/really/sys/fs/cgroup/cpu,cpuacct",
[VIR_CGROUP_CONTROLLER_CPUSET] = "/not/really/sys/fs/cgroup/cpuset",
[VIR_CGROUP_CONTROLLER_MEMORY] = "/not/really/sys/fs/cgroup/memory",
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
[VIR_CGROUP_CONTROLLER_FREEZER] = "/not/really/sys/fs/cgroup/freezer",
[VIR_CGROUP_CONTROLLER_BLKIO] = "/not/really/sys/fs/cgroup/blkio",
Add support for systemd cgroup mount Systemd uses a named cgroup mount for tracking processes. Add it as another type of controller, albeit one which we have to special case in a number of places. In particular we must never create/delete directories there, nor add tasks. Essentially the systemd mount is to be considered read-only for libvirt. With this change both the virCgroupDetectPlacement and virCgroupCopyPlacement methods must be invoked. The copy placement method will copy setup for resource controllers only. The detect placement method will probe for any named controllers, or resource controllers not already setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-25 19:13:44 +01:00
[VIR_CGROUP_CONTROLLER_SYSTEMD] = "/not/really/sys/fs/cgroup/systemd",
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
};
Fix cgroups when all are mounted on /sys/fs/cgroup Some users in Ubuntu/Debian seem to have a setup where all the cgroup controllers are mounted on /sys/fs/cgroup rather than any /sys/fs/cgroup/<controller> name. In the loop which detects which controllers are present for a mount point we were modifying 'mnt_dir' field in the 'struct mntent' var, but not always restoring the original value. This caused detection to break in the all-in-one mount setup. Fix that logic bug and add test case coverage for this mount setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-10 14:31:53 +01:00
const char *mountsAllInOne[VIR_CGROUP_CONTROLLER_LAST] = {
[VIR_CGROUP_CONTROLLER_CPU] = "/not/really/sys/fs/cgroup",
[VIR_CGROUP_CONTROLLER_CPUACCT] = "/not/really/sys/fs/cgroup",
[VIR_CGROUP_CONTROLLER_CPUSET] = "/not/really/sys/fs/cgroup",
[VIR_CGROUP_CONTROLLER_MEMORY] = "/not/really/sys/fs/cgroup",
[VIR_CGROUP_CONTROLLER_DEVICES] = "/not/really/sys/fs/cgroup",
[VIR_CGROUP_CONTROLLER_FREEZER] = NULL,
[VIR_CGROUP_CONTROLLER_BLKIO] = "/not/really/sys/fs/cgroup",
[VIR_CGROUP_CONTROLLER_SYSTEMD] = NULL,
};
Fix launching of VMs on when only logind part of systemd is present Debian systems may run the 'systemd-logind' daemon, which causes the /sys/fs/cgroup/systemd mount to be setup, but no other cgroup controllers are created. While the LXC driver considers cgroups to be mandatory, the QEMU driver is supposed to accept them as optional. We detect whether they are present by looking in /proc/mounts for any mounts of type 'cgroups', but this is not sufficient. We need to skip any named mounts (as seen by a name=XXX string in the mount options), so that we only detect actual resource controllers. http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=721979 Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-11 19:15:52 +01:00
const char *mountsLogind[VIR_CGROUP_CONTROLLER_LAST] = {
[VIR_CGROUP_CONTROLLER_CPU] = NULL,
[VIR_CGROUP_CONTROLLER_CPUACCT] = NULL,
[VIR_CGROUP_CONTROLLER_CPUSET] = NULL,
[VIR_CGROUP_CONTROLLER_MEMORY] = NULL,
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
[VIR_CGROUP_CONTROLLER_FREEZER] = NULL,
[VIR_CGROUP_CONTROLLER_BLKIO] = NULL,
[VIR_CGROUP_CONTROLLER_SYSTEMD] = "/not/really/sys/fs/cgroup/systemd",
};
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
Track symlinks for co-mounted cgroup controllers If a cgroup controller is co-mounted with another, eg /sys/fs/cgroup/cpu,cpuacct Then it is a requirement that there exist symlinks at /sys/fs/cgroup/cpu /sys/fs/cgroup/cpuacct pointing to the real mount point. Add support to virCgroupPtr to detect and track these symlinks Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-05 11:28:04 +01:00
const char *links[VIR_CGROUP_CONTROLLER_LAST] = {
[VIR_CGROUP_CONTROLLER_CPU] = "/not/really/sys/fs/cgroup/cpu",
[VIR_CGROUP_CONTROLLER_CPUACCT] = "/not/really/sys/fs/cgroup/cpuacct",
[VIR_CGROUP_CONTROLLER_CPUSET] = NULL,
[VIR_CGROUP_CONTROLLER_MEMORY] = NULL,
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
[VIR_CGROUP_CONTROLLER_FREEZER] = NULL,
[VIR_CGROUP_CONTROLLER_BLKIO] = NULL,
Fix cgroups when all are mounted on /sys/fs/cgroup Some users in Ubuntu/Debian seem to have a setup where all the cgroup controllers are mounted on /sys/fs/cgroup rather than any /sys/fs/cgroup/<controller> name. In the loop which detects which controllers are present for a mount point we were modifying 'mnt_dir' field in the 'struct mntent' var, but not always restoring the original value. This caused detection to break in the all-in-one mount setup. Fix that logic bug and add test case coverage for this mount setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-10 14:31:53 +01:00
[VIR_CGROUP_CONTROLLER_SYSTEMD] = NULL,
};
const char *linksAllInOne[VIR_CGROUP_CONTROLLER_LAST] = {
[VIR_CGROUP_CONTROLLER_CPU] = NULL,
[VIR_CGROUP_CONTROLLER_CPUACCT] = NULL,
[VIR_CGROUP_CONTROLLER_CPUSET] = NULL,
[VIR_CGROUP_CONTROLLER_MEMORY] = NULL,
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
[VIR_CGROUP_CONTROLLER_FREEZER] = NULL,
[VIR_CGROUP_CONTROLLER_BLKIO] = NULL,
[VIR_CGROUP_CONTROLLER_SYSTEMD] = NULL,
Track symlinks for co-mounted cgroup controllers If a cgroup controller is co-mounted with another, eg /sys/fs/cgroup/cpu,cpuacct Then it is a requirement that there exist symlinks at /sys/fs/cgroup/cpu /sys/fs/cgroup/cpuacct pointing to the real mount point. Add support to virCgroupPtr to detect and track these symlinks Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-05 11:28:04 +01:00
};
Fix launching of VMs on when only logind part of systemd is present Debian systems may run the 'systemd-logind' daemon, which causes the /sys/fs/cgroup/systemd mount to be setup, but no other cgroup controllers are created. While the LXC driver considers cgroups to be mandatory, the QEMU driver is supposed to accept them as optional. We detect whether they are present by looking in /proc/mounts for any mounts of type 'cgroups', but this is not sufficient. We need to skip any named mounts (as seen by a name=XXX string in the mount options), so that we only detect actual resource controllers. http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=721979 Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-11 19:15:52 +01:00
const char *linksLogind[VIR_CGROUP_CONTROLLER_LAST] = {
[VIR_CGROUP_CONTROLLER_CPU] = NULL,
[VIR_CGROUP_CONTROLLER_CPUACCT] = NULL,
[VIR_CGROUP_CONTROLLER_CPUSET] = NULL,
[VIR_CGROUP_CONTROLLER_MEMORY] = NULL,
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
[VIR_CGROUP_CONTROLLER_FREEZER] = NULL,
[VIR_CGROUP_CONTROLLER_BLKIO] = NULL,
[VIR_CGROUP_CONTROLLER_SYSTEMD] = NULL,
};
Track symlinks for co-mounted cgroup controllers If a cgroup controller is co-mounted with another, eg /sys/fs/cgroup/cpu,cpuacct Then it is a requirement that there exist symlinks at /sys/fs/cgroup/cpu /sys/fs/cgroup/cpuacct pointing to the real mount point. Add support to virCgroupPtr to detect and track these symlinks Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-05 11:28:04 +01:00
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
static int testCgroupNewForSelf(const void *args ATTRIBUTE_UNUSED)
{
virCgroupPtr cgroup = NULL;
int ret = -1;
const char *placement[VIR_CGROUP_CONTROLLER_LAST] = {
[VIR_CGROUP_CONTROLLER_CPU] = "/system",
[VIR_CGROUP_CONTROLLER_CPUACCT] = "/system",
Refactor cgroups internal data structures Currently the virCgroupPtr struct contains 3 pieces of information - path - path of the cgroup, relative to current process' cgroup placement - placement - current process' placement in each controller - mounts - mount point of each controller When reading/writing cgroup settings, the path & placement strings are combined to form the file path. This approach only works if we assume all cgroups will be relative to the current process' cgroup placement. To allow support for managing cgroups at any place in the heirarchy a change is needed. The 'placement' data should reflect the absolute path to the cgroup, and the 'path' value should no longer be used to form the paths to the cgroup attribute files. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-22 11:11:34 +00:00
[VIR_CGROUP_CONTROLLER_CPUSET] = "/",
[VIR_CGROUP_CONTROLLER_MEMORY] = "/",
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
Refactor cgroups internal data structures Currently the virCgroupPtr struct contains 3 pieces of information - path - path of the cgroup, relative to current process' cgroup placement - placement - current process' placement in each controller - mounts - mount point of each controller When reading/writing cgroup settings, the path & placement strings are combined to form the file path. This approach only works if we assume all cgroups will be relative to the current process' cgroup placement. To allow support for managing cgroups at any place in the heirarchy a change is needed. The 'placement' data should reflect the absolute path to the cgroup, and the 'path' value should no longer be used to form the paths to the cgroup attribute files. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-22 11:11:34 +00:00
[VIR_CGROUP_CONTROLLER_FREEZER] = "/",
[VIR_CGROUP_CONTROLLER_BLKIO] = "/",
Add support for systemd cgroup mount Systemd uses a named cgroup mount for tracking processes. Add it as another type of controller, albeit one which we have to special case in a number of places. In particular we must never create/delete directories there, nor add tasks. Essentially the systemd mount is to be considered read-only for libvirt. With this change both the virCgroupDetectPlacement and virCgroupCopyPlacement methods must be invoked. The copy placement method will copy setup for resource controllers only. The detect placement method will probe for any named controllers, or resource controllers not already setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-25 19:13:44 +01:00
[VIR_CGROUP_CONTROLLER_SYSTEMD] = "/user/berrange/123",
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
};
if (virCgroupNewSelf(&cgroup) < 0) {
fprintf(stderr, "Cannot create cgroup for self\n");
goto cleanup;
}
Track symlinks for co-mounted cgroup controllers If a cgroup controller is co-mounted with another, eg /sys/fs/cgroup/cpu,cpuacct Then it is a requirement that there exist symlinks at /sys/fs/cgroup/cpu /sys/fs/cgroup/cpuacct pointing to the real mount point. Add support to virCgroupPtr to detect and track these symlinks Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-05 11:28:04 +01:00
ret = validateCgroup(cgroup, "", mountsFull, links, placement);
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
cleanup:
virCgroupFree(&cgroup);
return ret;
}
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
# define ENSURE_ERRNO(en) \
do { \
if (!virLastErrorIsSystemErrno(en)) { \
virErrorPtr err = virGetLastError(); \
fprintf(stderr, "Did not get " #en " error code: %d:%d\n", \
err ? err->code : 0, err ? err->int1 : 0); \
goto cleanup; \
} } while (0)
/* Asking for impossible combination since CPU is co-mounted */
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
static int testCgroupNewForPartition(const void *args ATTRIBUTE_UNUSED)
{
virCgroupPtr cgroup = NULL;
int ret = -1;
int rv;
const char *placementSmall[VIR_CGROUP_CONTROLLER_LAST] = {
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
[VIR_CGROUP_CONTROLLER_CPU] = "/virtualmachines.partition",
[VIR_CGROUP_CONTROLLER_CPUACCT] = "/virtualmachines.partition",
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
[VIR_CGROUP_CONTROLLER_CPUSET] = NULL,
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
[VIR_CGROUP_CONTROLLER_MEMORY] = "/virtualmachines.partition",
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
[VIR_CGROUP_CONTROLLER_FREEZER] = NULL,
[VIR_CGROUP_CONTROLLER_BLKIO] = NULL,
Add support for systemd cgroup mount Systemd uses a named cgroup mount for tracking processes. Add it as another type of controller, albeit one which we have to special case in a number of places. In particular we must never create/delete directories there, nor add tasks. Essentially the systemd mount is to be considered read-only for libvirt. With this change both the virCgroupDetectPlacement and virCgroupCopyPlacement methods must be invoked. The copy placement method will copy setup for resource controllers only. The detect placement method will probe for any named controllers, or resource controllers not already setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-25 19:13:44 +01:00
[VIR_CGROUP_CONTROLLER_SYSTEMD] = NULL,
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
};
const char *placementFull[VIR_CGROUP_CONTROLLER_LAST] = {
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
[VIR_CGROUP_CONTROLLER_CPU] = "/virtualmachines.partition",
[VIR_CGROUP_CONTROLLER_CPUACCT] = "/virtualmachines.partition",
[VIR_CGROUP_CONTROLLER_CPUSET] = "/virtualmachines.partition",
[VIR_CGROUP_CONTROLLER_MEMORY] = "/virtualmachines.partition",
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
[VIR_CGROUP_CONTROLLER_FREEZER] = "/virtualmachines.partition",
[VIR_CGROUP_CONTROLLER_BLKIO] = "/virtualmachines.partition",
Add support for systemd cgroup mount Systemd uses a named cgroup mount for tracking processes. Add it as another type of controller, albeit one which we have to special case in a number of places. In particular we must never create/delete directories there, nor add tasks. Essentially the systemd mount is to be considered read-only for libvirt. With this change both the virCgroupDetectPlacement and virCgroupCopyPlacement methods must be invoked. The copy placement method will copy setup for resource controllers only. The detect placement method will probe for any named controllers, or resource controllers not already setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-25 19:13:44 +01:00
[VIR_CGROUP_CONTROLLER_SYSTEMD] = "/user/berrange/123",
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
};
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
if ((rv = virCgroupNewPartition("/virtualmachines", false, -1, &cgroup)) != -1) {
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
fprintf(stderr, "Unexpected found /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
ENSURE_ERRNO(ENOENT);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
/* Asking for impossible combination since CPU is co-mounted */
if ((rv = virCgroupNewPartition("/virtualmachines", true,
(1 << VIR_CGROUP_CONTROLLER_CPU),
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
&cgroup)) != -1) {
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
fprintf(stderr, "Should not have created /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
ENSURE_ERRNO(EINVAL);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
/* Asking for impossible combination since devices is not mounted */
if ((rv = virCgroupNewPartition("/virtualmachines", true,
(1 << VIR_CGROUP_CONTROLLER_DEVICES),
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
&cgroup)) != -1) {
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
fprintf(stderr, "Should not have created /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
ENSURE_ERRNO(ENXIO);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
/* Asking for small combination since devices is not mounted */
if ((rv = virCgroupNewPartition("/virtualmachines", true,
(1 << VIR_CGROUP_CONTROLLER_CPU) |
(1 << VIR_CGROUP_CONTROLLER_CPUACCT) |
(1 << VIR_CGROUP_CONTROLLER_MEMORY),
&cgroup)) != 0) {
fprintf(stderr, "Cannot create /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
ret = validateCgroup(cgroup, "/virtualmachines.partition", mountsSmall, links, placementSmall);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
virCgroupFree(&cgroup);
if ((rv = virCgroupNewPartition("/virtualmachines", true, -1, &cgroup)) != 0) {
fprintf(stderr, "Cannot create /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
ret = validateCgroup(cgroup, "/virtualmachines.partition", mountsFull, links, placementFull);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
cleanup:
virCgroupFree(&cgroup);
return ret;
}
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
static int testCgroupNewForPartitionNested(const void *args ATTRIBUTE_UNUSED)
{
virCgroupPtr cgroup = NULL;
int ret = -1;
int rv;
const char *placementFull[VIR_CGROUP_CONTROLLER_LAST] = {
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
[VIR_CGROUP_CONTROLLER_CPU] = "/deployment.partition/production.partition",
[VIR_CGROUP_CONTROLLER_CPUACCT] = "/deployment.partition/production.partition",
[VIR_CGROUP_CONTROLLER_CPUSET] = "/deployment.partition/production.partition",
[VIR_CGROUP_CONTROLLER_MEMORY] = "/deployment.partition/production.partition",
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
[VIR_CGROUP_CONTROLLER_FREEZER] = "/deployment.partition/production.partition",
[VIR_CGROUP_CONTROLLER_BLKIO] = "/deployment.partition/production.partition",
Add support for systemd cgroup mount Systemd uses a named cgroup mount for tracking processes. Add it as another type of controller, albeit one which we have to special case in a number of places. In particular we must never create/delete directories there, nor add tasks. Essentially the systemd mount is to be considered read-only for libvirt. With this change both the virCgroupDetectPlacement and virCgroupCopyPlacement methods must be invoked. The copy placement method will copy setup for resource controllers only. The detect placement method will probe for any named controllers, or resource controllers not already setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-25 19:13:44 +01:00
[VIR_CGROUP_CONTROLLER_SYSTEMD] = "/user/berrange/123",
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
};
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
if ((rv = virCgroupNewPartition("/deployment/production", false, -1, &cgroup)) != -1) {
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
fprintf(stderr, "Unexpected found /deployment/production cgroup: %d\n", -rv);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
goto cleanup;
}
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
ENSURE_ERRNO(ENOENT);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
/* Should not work, since we require /deployment to be pre-created */
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
if ((rv = virCgroupNewPartition("/deployment/production", true, -1, &cgroup)) != -1) {
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
fprintf(stderr, "Unexpected created /deployment/production cgroup: %d\n", -rv);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
goto cleanup;
}
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
ENSURE_ERRNO(ENOENT);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
if ((rv = virCgroupNewPartition("/deployment", true, -1, &cgroup)) != 0) {
fprintf(stderr, "Failed to create /deployment cgroup: %d\n", -rv);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
goto cleanup;
}
/* Should now work */
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
if ((rv = virCgroupNewPartition("/deployment/production", true, -1, &cgroup)) != 0) {
fprintf(stderr, "Failed to create /deployment/production cgroup: %d\n", -rv);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
goto cleanup;
}
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
ret = validateCgroup(cgroup, "/deployment.partition/production.partition",
mountsFull, links, placementFull);
cleanup:
virCgroupFree(&cgroup);
return ret;
}
static int testCgroupNewForPartitionNestedDeep(const void *args ATTRIBUTE_UNUSED)
{
virCgroupPtr cgroup = NULL;
int ret = -1;
int rv;
const char *placementFull[VIR_CGROUP_CONTROLLER_LAST] = {
[VIR_CGROUP_CONTROLLER_CPU] = "/user/berrange.user/production.partition",
[VIR_CGROUP_CONTROLLER_CPUACCT] = "/user/berrange.user/production.partition",
[VIR_CGROUP_CONTROLLER_CPUSET] = "/user/berrange.user/production.partition",
[VIR_CGROUP_CONTROLLER_MEMORY] = "/user/berrange.user/production.partition",
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
[VIR_CGROUP_CONTROLLER_FREEZER] = "/user/berrange.user/production.partition",
[VIR_CGROUP_CONTROLLER_BLKIO] = "/user/berrange.user/production.partition",
Add support for systemd cgroup mount Systemd uses a named cgroup mount for tracking processes. Add it as another type of controller, albeit one which we have to special case in a number of places. In particular we must never create/delete directories there, nor add tasks. Essentially the systemd mount is to be considered read-only for libvirt. With this change both the virCgroupDetectPlacement and virCgroupCopyPlacement methods must be invoked. The copy placement method will copy setup for resource controllers only. The detect placement method will probe for any named controllers, or resource controllers not already setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-25 19:13:44 +01:00
[VIR_CGROUP_CONTROLLER_SYSTEMD] = "/user/berrange/123",
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
};
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
if ((rv = virCgroupNewPartition("/user/berrange.user/production", false, -1, &cgroup)) != -1) {
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
fprintf(stderr, "Unexpected found /user/berrange.user/production cgroup: %d\n", -rv);
goto cleanup;
}
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
ENSURE_ERRNO(ENOENT);
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
/* Should not work, since we require /user/berrange.user to be pre-created */
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
if ((rv = virCgroupNewPartition("/user/berrange.user/production", true, -1, &cgroup)) != -1) {
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
fprintf(stderr, "Unexpected created /user/berrange.user/production cgroup: %d\n", -rv);
goto cleanup;
}
Report full errors from virCgroupNew* Instead of returning raw errno values, report full libvirt errors in virCgroupNew* functions. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-04 16:49:24 +01:00
ENSURE_ERRNO(ENOENT);
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
if ((rv = virCgroupNewPartition("/user", true, -1, &cgroup)) != 0) {
fprintf(stderr, "Failed to create /user/berrange.user cgroup: %d\n", -rv);
goto cleanup;
}
if ((rv = virCgroupNewPartition("/user/berrange.user", true, -1, &cgroup)) != 0) {
fprintf(stderr, "Failed to create /user/berrange.user cgroup: %d\n", -rv);
goto cleanup;
}
/* Should now work */
if ((rv = virCgroupNewPartition("/user/berrange.user/production", true, -1, &cgroup)) != 0) {
fprintf(stderr, "Failed to create /user/berrange.user/production cgroup: %d\n", -rv);
goto cleanup;
}
ret = validateCgroup(cgroup, "/user/berrange.user/production.partition",
mountsFull, links, placementFull);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
cleanup:
virCgroupFree(&cgroup);
return ret;
}
static int testCgroupNewForPartitionDomain(const void *args ATTRIBUTE_UNUSED)
{
virCgroupPtr partitioncgroup = NULL;
virCgroupPtr domaincgroup = NULL;
int ret = -1;
int rv;
const char *placement[VIR_CGROUP_CONTROLLER_LAST] = {
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
[VIR_CGROUP_CONTROLLER_CPU] = "/production.partition/foo.libvirt-lxc",
[VIR_CGROUP_CONTROLLER_CPUACCT] = "/production.partition/foo.libvirt-lxc",
[VIR_CGROUP_CONTROLLER_CPUSET] = "/production.partition/foo.libvirt-lxc",
[VIR_CGROUP_CONTROLLER_MEMORY] = "/production.partition/foo.libvirt-lxc",
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
[VIR_CGROUP_CONTROLLER_FREEZER] = "/production.partition/foo.libvirt-lxc",
[VIR_CGROUP_CONTROLLER_BLKIO] = "/production.partition/foo.libvirt-lxc",
Add support for systemd cgroup mount Systemd uses a named cgroup mount for tracking processes. Add it as another type of controller, albeit one which we have to special case in a number of places. In particular we must never create/delete directories there, nor add tasks. Essentially the systemd mount is to be considered read-only for libvirt. With this change both the virCgroupDetectPlacement and virCgroupCopyPlacement methods must be invoked. The copy placement method will copy setup for resource controllers only. The detect placement method will probe for any named controllers, or resource controllers not already setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-25 19:13:44 +01:00
[VIR_CGROUP_CONTROLLER_SYSTEMD] = "/user/berrange/123",
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
};
if ((rv = virCgroupNewPartition("/production", true, -1, &partitioncgroup)) != 0) {
fprintf(stderr, "Failed to create /production cgroup: %d\n", -rv);
goto cleanup;
}
if ((rv = virCgroupNewDomainPartition(partitioncgroup, "lxc", "foo", true, &domaincgroup)) != 0) {
fprintf(stderr, "Cannot create LXC cgroup: %d\n", -rv);
goto cleanup;
}
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
ret = validateCgroup(domaincgroup, "/production.partition/foo.libvirt-lxc", mountsFull, links, placement);
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
cleanup:
virCgroupFree(&partitioncgroup);
virCgroupFree(&domaincgroup);
return ret;
}
Do proper escaping of cgroup resource partitions If a user cgroup name begins with "cgroup.", "_" or with any of the controllers from /proc/cgroups followed by a dot, then they need to be prefixed with a single underscore. eg if there is an object "cpu.service", then this would end up as "_cpu.service" in the cgroup filesystem tree, however, "waldo.service" would stay "waldo.service", at least as long as nobody comes up with a cgroup controller called "waldo". Since we require a '.XXXX' suffix on all partitions, there is no scope for clashing with the kernel 'tasks' and 'release_agent' files. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:50:24 +01:00
static int testCgroupNewForPartitionDomainEscaped(const void *args ATTRIBUTE_UNUSED)
{
virCgroupPtr partitioncgroup1 = NULL;
virCgroupPtr partitioncgroup2 = NULL;
virCgroupPtr partitioncgroup3 = NULL;
virCgroupPtr domaincgroup = NULL;
int ret = -1;
int rv;
const char *placement[VIR_CGROUP_CONTROLLER_LAST] = {
[VIR_CGROUP_CONTROLLER_CPU] = "/_cgroup.evil/net_cls.evil/__evil.evil/_cpu.foo.libvirt-lxc",
[VIR_CGROUP_CONTROLLER_CPUACCT] = "/_cgroup.evil/net_cls.evil/__evil.evil/_cpu.foo.libvirt-lxc",
[VIR_CGROUP_CONTROLLER_CPUSET] = "/_cgroup.evil/net_cls.evil/__evil.evil/_cpu.foo.libvirt-lxc",
[VIR_CGROUP_CONTROLLER_MEMORY] = "/_cgroup.evil/net_cls.evil/__evil.evil/_cpu.foo.libvirt-lxc",
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
[VIR_CGROUP_CONTROLLER_FREEZER] = "/_cgroup.evil/net_cls.evil/__evil.evil/_cpu.foo.libvirt-lxc",
[VIR_CGROUP_CONTROLLER_BLKIO] = "/_cgroup.evil/net_cls.evil/__evil.evil/_cpu.foo.libvirt-lxc",
Add support for systemd cgroup mount Systemd uses a named cgroup mount for tracking processes. Add it as another type of controller, albeit one which we have to special case in a number of places. In particular we must never create/delete directories there, nor add tasks. Essentially the systemd mount is to be considered read-only for libvirt. With this change both the virCgroupDetectPlacement and virCgroupCopyPlacement methods must be invoked. The copy placement method will copy setup for resource controllers only. The detect placement method will probe for any named controllers, or resource controllers not already setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-25 19:13:44 +01:00
[VIR_CGROUP_CONTROLLER_SYSTEMD] = "/user/berrange/123",
Do proper escaping of cgroup resource partitions If a user cgroup name begins with "cgroup.", "_" or with any of the controllers from /proc/cgroups followed by a dot, then they need to be prefixed with a single underscore. eg if there is an object "cpu.service", then this would end up as "_cpu.service" in the cgroup filesystem tree, however, "waldo.service" would stay "waldo.service", at least as long as nobody comes up with a cgroup controller called "waldo". Since we require a '.XXXX' suffix on all partitions, there is no scope for clashing with the kernel 'tasks' and 'release_agent' files. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:50:24 +01:00
};
if ((rv = virCgroupNewPartition("/cgroup.evil", true, -1, &partitioncgroup1)) != 0) {
fprintf(stderr, "Failed to create /cgroup.evil cgroup: %d\n", -rv);
goto cleanup;
}
if ((rv = virCgroupNewPartition("/cgroup.evil/net_cls.evil", true, -1, &partitioncgroup2)) != 0) {
fprintf(stderr, "Failed to create /cgroup.evil/cpu.evil cgroup: %d\n", -rv);
goto cleanup;
}
if ((rv = virCgroupNewPartition("/cgroup.evil/net_cls.evil/_evil.evil", true, -1, &partitioncgroup3)) != 0) {
fprintf(stderr, "Failed to create /cgroup.evil cgroup: %d\n", -rv);
goto cleanup;
}
if ((rv = virCgroupNewDomainPartition(partitioncgroup3, "lxc", "cpu.foo", true, &domaincgroup)) != 0) {
fprintf(stderr, "Cannot create LXC cgroup: %d\n", -rv);
goto cleanup;
}
/* NB we're not expecting 'net_cls.evil' to be escaped,
* since our fake /proc/cgroups pretends this controller
* isn't compiled into the kernel
*/
ret = validateCgroup(domaincgroup, "/_cgroup.evil/net_cls.evil/__evil.evil/_cpu.foo.libvirt-lxc", mountsFull, links, placement);
cleanup:
virCgroupFree(&partitioncgroup3);
virCgroupFree(&partitioncgroup2);
virCgroupFree(&partitioncgroup1);
virCgroupFree(&domaincgroup);
return ret;
}
Fix cgroups when all are mounted on /sys/fs/cgroup Some users in Ubuntu/Debian seem to have a setup where all the cgroup controllers are mounted on /sys/fs/cgroup rather than any /sys/fs/cgroup/<controller> name. In the loop which detects which controllers are present for a mount point we were modifying 'mnt_dir' field in the 'struct mntent' var, but not always restoring the original value. This caused detection to break in the all-in-one mount setup. Fix that logic bug and add test case coverage for this mount setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-10 14:31:53 +01:00
static int testCgroupNewForSelfAllInOne(const void *args ATTRIBUTE_UNUSED)
{
virCgroupPtr cgroup = NULL;
int ret = -1;
const char *placement[VIR_CGROUP_CONTROLLER_LAST] = {
[VIR_CGROUP_CONTROLLER_CPU] = "/",
[VIR_CGROUP_CONTROLLER_CPUACCT] = "/",
[VIR_CGROUP_CONTROLLER_CPUSET] = "/",
[VIR_CGROUP_CONTROLLER_MEMORY] = "/",
[VIR_CGROUP_CONTROLLER_DEVICES] = "/",
[VIR_CGROUP_CONTROLLER_FREEZER] = NULL,
[VIR_CGROUP_CONTROLLER_BLKIO] = "/",
};
if (virCgroupNewSelf(&cgroup) < 0) {
fprintf(stderr, "Cannot create cgroup for self\n");
goto cleanup;
}
ret = validateCgroup(cgroup, "", mountsAllInOne, linksAllInOne, placement);
cleanup:
virCgroupFree(&cgroup);
return ret;
}
Fix launching of VMs on when only logind part of systemd is present Debian systems may run the 'systemd-logind' daemon, which causes the /sys/fs/cgroup/systemd mount to be setup, but no other cgroup controllers are created. While the LXC driver considers cgroups to be mandatory, the QEMU driver is supposed to accept them as optional. We detect whether they are present by looking in /proc/mounts for any mounts of type 'cgroups', but this is not sufficient. We need to skip any named mounts (as seen by a name=XXX string in the mount options), so that we only detect actual resource controllers. http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=721979 Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-11 19:15:52 +01:00
static int testCgroupNewForSelfLogind(const void *args ATTRIBUTE_UNUSED)
{
virCgroupPtr cgroup = NULL;
int ret = -1;
const char *placement[VIR_CGROUP_CONTROLLER_LAST] = {
[VIR_CGROUP_CONTROLLER_CPU] = NULL,
[VIR_CGROUP_CONTROLLER_CPUACCT] = NULL,
[VIR_CGROUP_CONTROLLER_CPUSET] = NULL,
[VIR_CGROUP_CONTROLLER_MEMORY] = NULL,
[VIR_CGROUP_CONTROLLER_DEVICES] = NULL,
[VIR_CGROUP_CONTROLLER_FREEZER] = NULL,
[VIR_CGROUP_CONTROLLER_BLKIO] = NULL,
[VIR_CGROUP_CONTROLLER_SYSTEMD] = "/",
};
if (virCgroupNewSelf(&cgroup) < 0) {
fprintf(stderr, "Cannot create cgroup for self\n");
goto cleanup;
}
ret = validateCgroup(cgroup, "", mountsLogind, linksLogind, placement);
cleanup:
virCgroupFree(&cgroup);
return ret;
}
static int testCgroupAvailable(const void *args)
{
bool got = virCgroupAvailable();
bool want = args == (void*)0x1;
if (got != want) {
fprintf(stderr, "Expected cgroup %savailable, but state was wrong\n",
want ? "" : "not ");
return -1;
}
return 0;
}
Fix cgroups when all are mounted on /sys/fs/cgroup Some users in Ubuntu/Debian seem to have a setup where all the cgroup controllers are mounted on /sys/fs/cgroup rather than any /sys/fs/cgroup/<controller> name. In the loop which detects which controllers are present for a mount point we were modifying 'mnt_dir' field in the 'struct mntent' var, but not always restoring the original value. This caused detection to break in the all-in-one mount setup. Fix that logic bug and add test case coverage for this mount setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-10 14:31:53 +01:00
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
# define FAKESYSFSDIRTEMPLATE abs_builddir "/fakesysfsdir-XXXXXX"
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
static int
mymain(void)
{
int ret = 0;
char *fakesysfsdir;
Adapt to VIR_STRDUP and VIR_STRNDUP in tests/*
2013-05-03 14:52:21 +02:00
if (VIR_STRDUP_QUIET(fakesysfsdir, FAKESYSFSDIRTEMPLATE) < 0) {
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
fprintf(stderr, "Out of memory\n");
abort();
}
if (!mkdtemp(fakesysfsdir)) {
fprintf(stderr, "Cannot create fakesysfsdir");
abort();
}
setenv("LIBVIRT_FAKE_SYSFS_DIR", fakesysfsdir, 1);
Remove test case average timing The test case average timing code has not been used by any test case ever. Delete it to remove complexity. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-20 19:13:35 +01:00
if (virtTestRun("New cgroup for self", testCgroupNewForSelf, NULL) < 0)
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
ret = -1;
Remove test case average timing The test case average timing code has not been used by any test case ever. Delete it to remove complexity. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-20 19:13:35 +01:00
if (virtTestRun("New cgroup for partition", testCgroupNewForPartition, NULL) < 0)
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
ret = -1;
Remove test case average timing The test case average timing code has not been used by any test case ever. Delete it to remove complexity. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-20 19:13:35 +01:00
if (virtTestRun("New cgroup for partition nested", testCgroupNewForPartitionNested, NULL) < 0)
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
ret = -1;
Remove test case average timing The test case average timing code has not been used by any test case ever. Delete it to remove complexity. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-20 19:13:35 +01:00
if (virtTestRun("New cgroup for partition nested deeply", testCgroupNewForPartitionNestedDeep, NULL) < 0)
Ensure all cgroup partitions have a suffix of ".partition" If the partition named passed in the XML does not already have a suffix, ensure it gets a '.partition' added to each component. The exceptions are /machine, /user and /system which do not need to have a suffix, since they are fixed partitions at the top level. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:23:51 +01:00
ret = -1;
Remove test case average timing The test case average timing code has not been used by any test case ever. Delete it to remove complexity. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-20 19:13:35 +01:00
if (virtTestRun("New cgroup for domain partition", testCgroupNewForPartitionDomain, NULL) < 0)
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
ret = -1;
Remove test case average timing The test case average timing code has not been used by any test case ever. Delete it to remove complexity. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-20 19:13:35 +01:00
if (virtTestRun("New cgroup for domain partition escaped", testCgroupNewForPartitionDomainEscaped, NULL) < 0)
Do proper escaping of cgroup resource partitions If a user cgroup name begins with "cgroup.", "_" or with any of the controllers from /proc/cgroups followed by a dot, then they need to be prefixed with a single underscore. eg if there is an object "cpu.service", then this would end up as "_cpu.service" in the cgroup filesystem tree, however, "waldo.service" would stay "waldo.service", at least as long as nobody comes up with a cgroup controller called "waldo". Since we require a '.XXXX' suffix on all partitions, there is no scope for clashing with the kernel 'tasks' and 'release_agent' files. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-04-26 10:50:24 +01:00
ret = -1;
Add a new virCgroupNewPartition for setting up resource partitions A resource partition is an absolute cgroup path, ignoring the current process placement. Expose a virCgroupNewPartition API for constructing such cgroups Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 18:08:39 +00:00
Remove test case average timing The test case average timing code has not been used by any test case ever. Delete it to remove complexity. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-20 19:13:35 +01:00
if (virtTestRun("Cgroup available", testCgroupAvailable, (void*)0x1) < 0)
Fix launching of VMs on when only logind part of systemd is present Debian systems may run the 'systemd-logind' daemon, which causes the /sys/fs/cgroup/systemd mount to be setup, but no other cgroup controllers are created. While the LXC driver considers cgroups to be mandatory, the QEMU driver is supposed to accept them as optional. We detect whether they are present by looking in /proc/mounts for any mounts of type 'cgroups', but this is not sufficient. We need to skip any named mounts (as seen by a name=XXX string in the mount options), so that we only detect actual resource controllers. http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=721979 Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-11 19:15:52 +01:00
ret = -1;
Fix cgroups when all are mounted on /sys/fs/cgroup Some users in Ubuntu/Debian seem to have a setup where all the cgroup controllers are mounted on /sys/fs/cgroup rather than any /sys/fs/cgroup/<controller> name. In the loop which detects which controllers are present for a mount point we were modifying 'mnt_dir' field in the 'struct mntent' var, but not always restoring the original value. This caused detection to break in the all-in-one mount setup. Fix that logic bug and add test case coverage for this mount setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-10 14:31:53 +01:00
setenv("VIR_CGROUP_MOCK_MODE", "allinone", 1);
Remove test case average timing The test case average timing code has not been used by any test case ever. Delete it to remove complexity. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-20 19:13:35 +01:00
if (virtTestRun("New cgroup for self (allinone)", testCgroupNewForSelfAllInOne, NULL) < 0)
Fix cgroups when all are mounted on /sys/fs/cgroup Some users in Ubuntu/Debian seem to have a setup where all the cgroup controllers are mounted on /sys/fs/cgroup rather than any /sys/fs/cgroup/<controller> name. In the loop which detects which controllers are present for a mount point we were modifying 'mnt_dir' field in the 'struct mntent' var, but not always restoring the original value. This caused detection to break in the all-in-one mount setup. Fix that logic bug and add test case coverage for this mount setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-10 14:31:53 +01:00
ret = -1;
Remove test case average timing The test case average timing code has not been used by any test case ever. Delete it to remove complexity. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-20 19:13:35 +01:00
if (virtTestRun("Cgroup available", testCgroupAvailable, (void*)0x1) < 0)
Fix launching of VMs on when only logind part of systemd is present Debian systems may run the 'systemd-logind' daemon, which causes the /sys/fs/cgroup/systemd mount to be setup, but no other cgroup controllers are created. While the LXC driver considers cgroups to be mandatory, the QEMU driver is supposed to accept them as optional. We detect whether they are present by looking in /proc/mounts for any mounts of type 'cgroups', but this is not sufficient. We need to skip any named mounts (as seen by a name=XXX string in the mount options), so that we only detect actual resource controllers. http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=721979 Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-11 19:15:52 +01:00
ret = -1;
unsetenv("VIR_CGROUP_MOCK_MODE");
setenv("VIR_CGROUP_MOCK_MODE", "logind", 1);
Remove test case average timing The test case average timing code has not been used by any test case ever. Delete it to remove complexity. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-20 19:13:35 +01:00
if (virtTestRun("New cgroup for self (logind)", testCgroupNewForSelfLogind, NULL) < 0)
Fix launching of VMs on when only logind part of systemd is present Debian systems may run the 'systemd-logind' daemon, which causes the /sys/fs/cgroup/systemd mount to be setup, but no other cgroup controllers are created. While the LXC driver considers cgroups to be mandatory, the QEMU driver is supposed to accept them as optional. We detect whether they are present by looking in /proc/mounts for any mounts of type 'cgroups', but this is not sufficient. We need to skip any named mounts (as seen by a name=XXX string in the mount options), so that we only detect actual resource controllers. http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=721979 Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-11 19:15:52 +01:00
ret = -1;
Remove test case average timing The test case average timing code has not been used by any test case ever. Delete it to remove complexity. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-20 19:13:35 +01:00
if (virtTestRun("Cgroup available", testCgroupAvailable, (void*)0x0) < 0)
Fix launching of VMs on when only logind part of systemd is present Debian systems may run the 'systemd-logind' daemon, which causes the /sys/fs/cgroup/systemd mount to be setup, but no other cgroup controllers are created. While the LXC driver considers cgroups to be mandatory, the QEMU driver is supposed to accept them as optional. We detect whether they are present by looking in /proc/mounts for any mounts of type 'cgroups', but this is not sufficient. We need to skip any named mounts (as seen by a name=XXX string in the mount options), so that we only detect actual resource controllers. http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=721979 Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-11 19:15:52 +01:00
ret = -1;
Fix cgroups when all are mounted on /sys/fs/cgroup Some users in Ubuntu/Debian seem to have a setup where all the cgroup controllers are mounted on /sys/fs/cgroup rather than any /sys/fs/cgroup/<controller> name. In the loop which detects which controllers are present for a mount point we were modifying 'mnt_dir' field in the 'struct mntent' var, but not always restoring the original value. This caused detection to break in the all-in-one mount setup. Fix that logic bug and add test case coverage for this mount setup. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-09-10 14:31:53 +01:00
unsetenv("VIR_CGROUP_MOCK_MODE");
Add a test suite for cgroups functionality Some aspects of the cgroups setup / detection code are quite subtle and easy to break. It would greatly benefit from unit testing, but this is difficult because the test suite won't have privileges to play around with cgroups. The solution is to use monkey patching via LD_PRELOAD to override the fopen, open, mkdir, access functions to redirect access of cgroups files to some magic stubs in the test suite. Using this we provide custom content for the /proc/cgroup and /proc/self/mounts files which report a fixed cgroup setup. We then override open/mkdir/access so that access to the cgroups filesystem gets redirected into files in a temporary directory tree in the test suite build dir. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-28 14:36:52 +00:00
if (getenv("LIBVIRT_SKIP_CLEANUP") == NULL)
virFileDeleteTree(fakesysfsdir);
VIR_FREE(fakesysfsdir);
return ret==0 ? EXIT_SUCCESS : EXIT_FAILURE;
}
VIRT_TEST_MAIN_PRELOAD(mymain, abs_builddir "/.libs/vircgroupmock.so")
#else
int
main(void)
{
return EXIT_AM_SKIP;
}
#endif
Copy Permalink