shaba/libvirt
1
0
Fork 0
mirror of https://gitlab.com/libvirt/libvirt.git synced 2024-12-23 21:34:54 +03:00
Code Releases Activity
35c3aab44d
libvirt/tests/vircgrouptest.c

956 lines
34 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 18:36:52 +04:00
/*
tests: fix build on old 32-bit platforms gcc 4.1.2 (hello RHEL 5) on 32-bit platforms complains: vircgrouptest.c: In function 'testCgroupGetPercpuStats': vircgrouptest.c:627: warning: integer constant is too large for 'long' type vircgrouptest.c:628: warning: this decimal constant is unsigned only in ISO C90 vircgrouptest.c:634: warning: integer constant is too large for 'long' type vircgrouptest.c:635: warning: this decimal constant is unsigned only in ISO C90 vircgrouptest.c:636: warning: this decimal constant is unsigned only in ISO C90 vircgrouptest.c:644: warning: integer constant is too large for 'long' type * tests/vircgrouptest.c (testCgroupGetPercpuStats): Use ULL suffix. Signed-off-by: Eric Blake <eblake@redhat.com>
2015-04-15 21:50:23 +03:00
* Copyright (C) 2013-2015 Red Hat, Inc.
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 18:36:52 +04:00
*
* 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 14:36:23 +04:00
# 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 18:36:52 +04:00
# include "virerror.h"
# include "virlog.h"
# include "virfile.h"
tests: Add tests for virCgroupDetectMounts Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2015-03-17 00:36:52 +03:00
# include "virbuffer.h"
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
# include "testutilslxc.h"
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
# include "nodeinfo.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 18:36:52 +04:00
# define VIR_FROM_THIS VIR_FROM_NONE
Add virLogSource variables to all source files Any source file which calls the logging APIs now needs to have a VIR_LOG_INIT("source.name") declaration at the start of the file. This provides a static variable of the virLogSource type. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2014-02-28 16:16:17 +04:00
VIR_LOG_INIT("tests.cgrouptest");
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 18:36:52 +04:00
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 14:28:04 +04:00
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 18:36:52 +04:00
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 18:09:33 +04:00
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 18:36:52 +04:00
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 11:53:48 +04:00
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 18:36:52 +04:00
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 14:28:04 +04:00
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 18:36:52 +04:00
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 22:13:44 +04:00
[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 18:36:52 +04:00
};
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 22:13:44 +04:00
[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 18:36:52 +04:00
};
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 17:31:53 +04:00
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 22:15:52 +04:00
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 18:36:52 +04:00
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 14:28:04 +04:00
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 17:31:53 +04:00
[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 14:28:04 +04:00
};
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 22:15:52 +04:00
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 14:28:04 +04:00
tests: Add tests for virCgroupDetectMounts Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2015-03-17 00:36:52 +03:00
static int
testCgroupDetectMounts(const void *args)
{
int result = -1;
const char *file = args;
char *mounts = NULL;
char *parsed = NULL;
const char *actual;
virCgroupPtr group = NULL;
virBuffer buf = VIR_BUFFER_INITIALIZER;
size_t i;
if (virAsprintf(&mounts, "%s/vircgroupdata/%s.mounts",
abs_srcdir, file) < 0 ||
virAsprintf(&parsed, "%s/vircgroupdata/%s.parsed",
abs_srcdir, file) < 0 ||
VIR_ALLOC(group) < 0)
goto cleanup;
if (virCgroupDetectMountsFromFile(group, mounts, false) < 0)
goto cleanup;
for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
virBufferAsprintf(&buf, "%-12s %s\n",
virCgroupControllerTypeToString(i),
NULLSTR(group->controllers[i].mountPoint));
}
if (virBufferCheckError(&buf) < 0)
goto cleanup;
actual = virBufferCurrentContent(&buf);
tests: Add virtTestCompareToFile Replaces a common pattern used in many test files
2015-04-23 18:14:26 +03:00
if (virtTestCompareToFile(actual, parsed) < 0)
tests: Add tests for virCgroupDetectMounts Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2015-03-17 00:36:52 +03:00
goto cleanup;
result = 0;
cleanup:
VIR_FREE(mounts);
VIR_FREE(parsed);
virCgroupFree(&group);
virBufferFreeAndReset(&buf);
return result;
}
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 18:36:52 +04:00
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 15:11:34 +04:00
[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 18:36:52 +04:00
[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 15:11:34 +04:00
[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 22:13:44 +04:00
[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 18:36:52 +04:00
};
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 14:28:04 +04:00
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 18:36:52 +04:00
Indent top-level labels by one space in tests/
2014-03-25 10:53:44 +04:00
cleanup:
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 18:36:52 +04:00
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 19:49:24 +04:00
# 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 22:08:39 +04:00
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 13:23:51 +04:00
[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 22:08:39 +04:00
[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 13:23:51 +04:00
[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 22:08:39 +04:00
[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 22:13:44 +04:00
[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 22:08:39 +04:00
};
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 13:23:51 +04:00
[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 22:08:39 +04:00
[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 13:23:51 +04:00
[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 22:13:44 +04:00
[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 22:08:39 +04:00
};
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 19:49:24 +04:00
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 22:08:39 +04:00
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 19:49:24 +04:00
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 22:08:39 +04:00
/* 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 19:49:24 +04:00
&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 22:08:39 +04:00
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 19:49:24 +04:00
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 22:08:39 +04:00
/* 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 19:49:24 +04:00
&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 22:08:39 +04:00
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 19:49:24 +04:00
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 22:08:39 +04:00
/* 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 13:23:51 +04:00
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 22:08:39 +04:00
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 13:23:51 +04:00
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 22:08:39 +04:00
Indent top-level labels by one space in tests/
2014-03-25 10:53:44 +04:00
cleanup:
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 22:08:39 +04:00
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 18:36:52 +04:00
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 22:08:39 +04:00
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 13:23:51 +04:00
[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 22:08:39 +04:00
[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 13:23:51 +04:00
[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 22:13:44 +04:00
[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 22:08:39 +04:00
};
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 18:36:52 +04:00
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 19:49:24 +04:00
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 13:23:51 +04:00
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 22:08:39 +04:00
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 19:49:24 +04:00
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 22:08:39 +04:00
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 13:23:51 +04:00
/* 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 19:49:24 +04:00
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 13:23:51 +04:00
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 22:08:39 +04:00
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 19:49:24 +04:00
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 22:08:39 +04:00
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 13:23:51 +04:00
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 22:08:39 +04:00
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 13:23:51 +04:00
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 22:08:39 +04:00
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 13:23:51 +04:00
ret = validateCgroup(cgroup, "/deployment.partition/production.partition",
mountsFull, links, placementFull);
Indent top-level labels by one space in tests/
2014-03-25 10:53:44 +04:00
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 13:23:51 +04:00
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 22:13:44 +04:00
[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 13:23:51 +04:00
};
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 19:49:24 +04:00
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 13:23:51 +04:00
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 19:49:24 +04:00
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 13:23:51 +04:00
/* 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 19:49:24 +04:00
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 13:23:51 +04:00
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 19:49:24 +04:00
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 13:23:51 +04:00
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 22:08:39 +04:00
Indent top-level labels by one space in tests/
2014-03-25 10:53:44 +04:00
cleanup:
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 22:08:39 +04:00
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 13:23:51 +04:00
[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 22:08:39 +04:00
[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 13:23:51 +04:00
[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 22:13:44 +04:00
[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 22:08:39 +04:00
};
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 13:23:51 +04:00
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 22:08:39 +04:00
Indent top-level labels by one space in tests/
2014-03-25 10:53:44 +04:00
cleanup:
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 22:08:39 +04:00
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 13:50:24 +04:00
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 22:13:44 +04:00
[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 13:50:24 +04:00
};
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);
Indent top-level labels by one space in tests/
2014-03-25 10:53:44 +04:00
cleanup:
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 13:50:24 +04:00
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 17:31:53 +04:00
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);
Indent top-level labels by one space in tests/
2014-03-25 10:53:44 +04:00
cleanup:
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 17:31:53 +04:00
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 22:15:52 +04:00
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);
Indent top-level labels by one space in tests/
2014-03-25 10:53:44 +04:00
cleanup:
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 22:15:52 +04:00
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;
}
vircgroup: Introduce virCgroupControllerAvailable This new internal API checks if given CGroup controller is available. It is going to be needed later when we need to make a decision whether pin domain memory onto NUMA nodes using cpuset CGroup controller or using numa_set_membind(). Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2015-03-31 12:39:13 +03:00
static int testCgroupControllerAvailable(const void *args ATTRIBUTE_UNUSED)
{
int ret = 0;
# define CHECK_CONTROLLER(c, present) \
if ((present && !virCgroupControllerAvailable(c)) || \
(!present && virCgroupControllerAvailable(c))) { \
fprintf(stderr, present ? \
"Expected controller %s not available\n" : \
"Unexpected controller %s available\n", #c); \
ret = -1; \
}
CHECK_CONTROLLER(VIR_CGROUP_CONTROLLER_CPU, true)
CHECK_CONTROLLER(VIR_CGROUP_CONTROLLER_CPUACCT, true)
CHECK_CONTROLLER(VIR_CGROUP_CONTROLLER_CPUSET, true)
CHECK_CONTROLLER(VIR_CGROUP_CONTROLLER_MEMORY, true)
CHECK_CONTROLLER(VIR_CGROUP_CONTROLLER_DEVICES, false)
CHECK_CONTROLLER(VIR_CGROUP_CONTROLLER_FREEZER, true)
CHECK_CONTROLLER(VIR_CGROUP_CONTROLLER_BLKIO, true)
CHECK_CONTROLLER(VIR_CGROUP_CONTROLLER_NET_CLS, false)
CHECK_CONTROLLER(VIR_CGROUP_CONTROLLER_PERF_EVENT, false)
CHECK_CONTROLLER(VIR_CGROUP_CONTROLLER_SYSTEMD, true)
# undef CHECK_CONTROLLER
return ret;
}
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
static int testCgroupGetPercpuStats(const void *args ATTRIBUTE_UNUSED)
{
virCgroupPtr cgroup = NULL;
size_t i;
int rv, ret = -1;
Fix virCgroupGetPercpuStats with non-continuous present CPUs Per-cpu stats are only shown for present CPUs in the cgroups, but we were only parsing the largest CPU number from /sys/devices/system/cpu/present and looking for stats even for non-present CPUs. This resulted in: internal error: cpuacct parse error
2015-01-22 13:54:50 +03:00
virTypedParameterPtr params = NULL;
# define EXPECTED_NCPUS 160
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
Fix build with older gcc My commit af1c98e4 broke the build on RHEL-6: vircgrouptest.c: In function 'testCgroupGetPercpuStats': vircgrouptest.c:566: error: nested extern declaration of '_gl_verify_function2' [-Wnested-externs] The only thing that needs checking is that the array size is at least EXPECTED_NCPUS, to prevent access beyond the array. We can ensure the minimum size also by specifying the array size upfront.
2015-01-23 12:30:01 +03:00
unsigned long long expected[EXPECTED_NCPUS] = {
Fix virCgroupGetPercpuStats with non-continuous present CPUs Per-cpu stats are only shown for present CPUs in the cgroups, but we were only parsing the largest CPU number from /sys/devices/system/cpu/present and looking for stats even for non-present CPUs. This resulted in: internal error: cpuacct parse error
2015-01-22 13:54:50 +03:00
0, 0, 0, 0, 0, 0, 0, 0,
tests: fix build on old 32-bit platforms gcc 4.1.2 (hello RHEL 5) on 32-bit platforms complains: vircgrouptest.c: In function 'testCgroupGetPercpuStats': vircgrouptest.c:627: warning: integer constant is too large for 'long' type vircgrouptest.c:628: warning: this decimal constant is unsigned only in ISO C90 vircgrouptest.c:634: warning: integer constant is too large for 'long' type vircgrouptest.c:635: warning: this decimal constant is unsigned only in ISO C90 vircgrouptest.c:636: warning: this decimal constant is unsigned only in ISO C90 vircgrouptest.c:644: warning: integer constant is too large for 'long' type * tests/vircgrouptest.c (testCgroupGetPercpuStats): Use ULL suffix. Signed-off-by: Eric Blake <eblake@redhat.com>
2015-04-15 21:50:23 +03:00
7059492996ULL, 0, 0, 0, 0, 0, 0, 0,
4180532496ULL, 0, 0, 0, 0, 0, 0, 0,
Fix virCgroupGetPercpuStats with non-continuous present CPUs Per-cpu stats are only shown for present CPUs in the cgroups, but we were only parsing the largest CPU number from /sys/devices/system/cpu/present and looking for stats even for non-present CPUs. This resulted in: internal error: cpuacct parse error
2015-01-22 13:54:50 +03:00
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
tests: fix build on old 32-bit platforms gcc 4.1.2 (hello RHEL 5) on 32-bit platforms complains: vircgrouptest.c: In function 'testCgroupGetPercpuStats': vircgrouptest.c:627: warning: integer constant is too large for 'long' type vircgrouptest.c:628: warning: this decimal constant is unsigned only in ISO C90 vircgrouptest.c:634: warning: integer constant is too large for 'long' type vircgrouptest.c:635: warning: this decimal constant is unsigned only in ISO C90 vircgrouptest.c:636: warning: this decimal constant is unsigned only in ISO C90 vircgrouptest.c:644: warning: integer constant is too large for 'long' type * tests/vircgrouptest.c (testCgroupGetPercpuStats): Use ULL suffix. Signed-off-by: Eric Blake <eblake@redhat.com>
2015-04-15 21:50:23 +03:00
1957541268ULL, 0, 0, 0, 0, 0, 0, 0,
2065932204ULL, 0, 0, 0, 0, 0, 0, 0,
18228689414ULL, 0, 0, 0, 0, 0, 0, 0,
4245525148ULL, 0, 0, 0, 0, 0, 0, 0,
2911161568ULL, 0, 0, 0, 0, 0, 0, 0,
1407758136ULL, 0, 0, 0, 0, 0, 0, 0,
1836807700ULL, 0, 0, 0, 0, 0, 0, 0,
1065296618ULL, 0, 0, 0, 0, 0, 0, 0,
2046213266ULL, 0, 0, 0, 0, 0, 0, 0,
747889778ULL, 0, 0, 0, 0, 0, 0, 0,
709566900ULL, 0, 0, 0, 0, 0, 0, 0,
444777342ULL, 0, 0, 0, 0, 0, 0, 0,
5683512916ULL, 0, 0, 0, 0, 0, 0, 0,
635751356ULL, 0, 0, 0, 0, 0, 0, 0,
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
};
Fix virCgroupGetPercpuStats with non-continuous present CPUs Per-cpu stats are only shown for present CPUs in the cgroups, but we were only parsing the largest CPU number from /sys/devices/system/cpu/present and looking for stats even for non-present CPUs. This resulted in: internal error: cpuacct parse error
2015-01-22 13:54:50 +03:00
if (VIR_ALLOC_N(params, EXPECTED_NCPUS) < 0)
goto cleanup;
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
if ((rv = virCgroupNewPartition("/virtualmachines", true,
(1 << VIR_CGROUP_CONTROLLER_CPU) |
(1 << VIR_CGROUP_CONTROLLER_CPUACCT),
&cgroup)) < 0) {
fprintf(stderr, "Could not create /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
nodeinfo: Add sysfs_prefix to nodeGetCPUCount Add the sysfs_prefix argument to the call to allow for setting the path for tests to something other than SYSFS_SYSTEM_PATH.
2015-07-08 02:49:10 +03:00
if (nodeGetCPUCount(NULL) != EXPECTED_NCPUS) {
fprintf(stderr, "Unexpected: nodeGetCPUCount() yields: %d\n", nodeGetCPUCount(NULL));
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
goto cleanup;
}
if ((rv = virCgroupGetPercpuStats(cgroup,
params,
Fix virCgroupGetPercpuStats with non-continuous present CPUs Per-cpu stats are only shown for present CPUs in the cgroups, but we were only parsing the largest CPU number from /sys/devices/system/cpu/present and looking for stats even for non-present CPUs. This resulted in: internal error: cpuacct parse error
2015-01-22 13:54:50 +03:00
1, 0, EXPECTED_NCPUS, 0)) < 0) {
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
fprintf(stderr, "Failed call to virCgroupGetPercpuStats for /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
Fix virCgroupGetPercpuStats with non-continuous present CPUs Per-cpu stats are only shown for present CPUs in the cgroups, but we were only parsing the largest CPU number from /sys/devices/system/cpu/present and looking for stats even for non-present CPUs. This resulted in: internal error: cpuacct parse error
2015-01-22 13:54:50 +03:00
for (i = 0; i < EXPECTED_NCPUS; i++) {
Avoid using !STREQ and !STRNEQ We have macros for both positive and negative string matching. Therefore there is no need to use !STREQ or !STRNEQ. At the same time as we are dropping this, new syntax-check rule is introduced to make sure we won't introduce it again. Signed-off-by: Ishmanpreet Kaur Khera <khera.ishman@gmail.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2015-10-20 19:15:12 +03:00
if (STRNEQ(params[i].field, VIR_DOMAIN_CPU_STATS_CPUTIME)) {
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
fprintf(stderr,
Fix virCgroupGetPercpuStats with non-continuous present CPUs Per-cpu stats are only shown for present CPUs in the cgroups, but we were only parsing the largest CPU number from /sys/devices/system/cpu/present and looking for stats even for non-present CPUs. This resulted in: internal error: cpuacct parse error
2015-01-22 13:54:50 +03:00
"Wrong parameter name value from virCgroupGetPercpuStats at %zu (is: %s)\n",
i, params[i].field);
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
goto cleanup;
}
if (params[i].type != VIR_TYPED_PARAM_ULLONG) {
fprintf(stderr,
Fix virCgroupGetPercpuStats with non-continuous present CPUs Per-cpu stats are only shown for present CPUs in the cgroups, but we were only parsing the largest CPU number from /sys/devices/system/cpu/present and looking for stats even for non-present CPUs. This resulted in: internal error: cpuacct parse error
2015-01-22 13:54:50 +03:00
"Wrong parameter value type from virCgroupGetPercpuStats at %zu (is: %d)\n",
i, params[i].type);
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
goto cleanup;
}
if (params[i].value.ul != expected[i]) {
fprintf(stderr,
Fix virCgroupGetPercpuStats with non-continuous present CPUs Per-cpu stats are only shown for present CPUs in the cgroups, but we were only parsing the largest CPU number from /sys/devices/system/cpu/present and looking for stats even for non-present CPUs. This resulted in: internal error: cpuacct parse error
2015-01-22 13:54:50 +03:00
"Wrong value from virCgroupGetMemoryUsage at %zu (expected %llu)\n",
i, params[i].value.ul);
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
goto cleanup;
}
}
ret = 0;
Indent top-level labels by one space in tests/
2014-03-25 10:53:44 +04:00
cleanup:
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
virCgroupFree(&cgroup);
Fix a memory leak in virCgroupGetPercpuStats Coverity reports that my commit af1c98e introduced two memory leaks: the cpumap if ncpus == 0 in virCgroupGetPercpuStats and the params array in the test of the function.
2015-01-26 12:20:22 +03:00
VIR_FREE(params);
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
return ret;
}
Add unit test for virCgroupGetMemoryUsage.
2014-02-14 21:49:06 +04:00
static int testCgroupGetMemoryUsage(const void *args ATTRIBUTE_UNUSED)
{
virCgroupPtr cgroup = NULL;
int rv, ret = -1;
unsigned long kb;
if ((rv = virCgroupNewPartition("/virtualmachines", true,
(1 << VIR_CGROUP_CONTROLLER_MEMORY),
&cgroup)) < 0) {
fprintf(stderr, "Could not create /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
if ((rv = virCgroupGetMemoryUsage(cgroup, &kb)) < 0) {
fprintf(stderr, "Could not retrieve GetMemoryUsage for /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
if (kb != 1421212UL) {
fprintf(stderr,
"Wrong value from virCgroupGetMemoryUsage (expected %ld)\n",
1421212UL);
goto cleanup;
}
ret = 0;
Indent top-level labels by one space in tests/
2014-03-25 10:53:44 +04:00
cleanup:
Add unit test for virCgroupGetMemoryUsage.
2014-02-14 21:49:06 +04:00
virCgroupFree(&cgroup);
return ret;
}
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
static int testCgroupGetBlkioIoServiced(const void *args ATTRIBUTE_UNUSED)
{
virCgroupPtr cgroup = NULL;
size_t i;
int rv, ret = -1;
const long long expected_values[] = {
build: fix build on 32-bit hosts vircgrouptest.c: In function 'testCgroupGetPercpuStats': vircgrouptest.c:543: warning: integer constatnt is too large for 'long' type Signed-off-by: Eric Blake <eblake@redhat.com>
2014-02-22 04:21:58 +04:00
119084214273ULL,
822880960513ULL,
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
9665167,
73283807
};
const char* names[] = {
"bytes read",
"bytes written",
"requests read",
"requests written"
};
long long values[ARRAY_CARDINALITY(expected_values)];
if ((rv = virCgroupNewPartition("/virtualmachines", true,
(1 << VIR_CGROUP_CONTROLLER_BLKIO),
&cgroup)) < 0) {
fprintf(stderr, "Could not create /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
if ((rv = virCgroupGetBlkioIoServiced(cgroup,
values, &values[1],
&values[2], &values[3])) < 0) {
fprintf(stderr, "Could not retrieve BlkioIoServiced for /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
for (i = 0; i < ARRAY_CARDINALITY(expected_values); i++) {
if (expected_values[i] != values[i]) {
fprintf(stderr,
"Wrong value for %s from virCgroupBlkioIoServiced (expected %lld)\n",
names[i], expected_values[i]);
goto cleanup;
}
}
ret = 0;
Indent top-level labels by one space in tests/
2014-03-25 10:53:44 +04:00
cleanup:
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
virCgroupFree(&cgroup);
return ret;
}
static int testCgroupGetBlkioIoDeviceServiced(const void *args ATTRIBUTE_UNUSED)
{
virCgroupPtr cgroup = NULL;
size_t i;
int rv, ret = -1;
const long long expected_values0[] = {
build: fix build on 32-bit hosts vircgrouptest.c: In function 'testCgroupGetPercpuStats': vircgrouptest.c:543: warning: integer constatnt is too large for 'long' type Signed-off-by: Eric Blake <eblake@redhat.com>
2014-02-22 04:21:58 +04:00
59542107136ULL,
411440480256ULL,
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
4832583,
36641903
};
const long long expected_values1[] = {
build: fix build on 32-bit hosts vircgrouptest.c: In function 'testCgroupGetPercpuStats': vircgrouptest.c:543: warning: integer constatnt is too large for 'long' type Signed-off-by: Eric Blake <eblake@redhat.com>
2014-02-22 04:21:58 +04:00
59542107137ULL,
411440480257ULL,
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
4832584,
36641904
};
const char* names[] = {
"bytes read",
"bytes written",
"requests read",
"requests written"
};
long long values[ARRAY_CARDINALITY(expected_values0)];
if ((rv = virCgroupNewPartition("/virtualmachines", true,
(1 << VIR_CGROUP_CONTROLLER_BLKIO),
&cgroup)) < 0) {
fprintf(stderr, "Could not create /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
if ((rv = virCgroupGetBlkioIoDeviceServiced(cgroup,
FAKEDEVDIR0,
values, &values[1],
&values[2], &values[3])) < 0) {
fprintf(stderr, "Could not retrieve BlkioIoDeviceServiced for /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
for (i = 0; i < ARRAY_CARDINALITY(expected_values0); i++) {
if (expected_values0[i] != values[i]) {
fprintf(stderr,
"Wrong value for %s from virCgroupGetBlkioIoDeviceServiced (expected %lld)\n",
names[i], expected_values0[i]);
goto cleanup;
}
}
if ((rv = virCgroupGetBlkioIoDeviceServiced(cgroup,
FAKEDEVDIR1,
values, &values[1],
&values[2], &values[3])) < 0) {
fprintf(stderr, "Could not retrieve BlkioIoDeviceServiced for /virtualmachines cgroup: %d\n", -rv);
goto cleanup;
}
for (i = 0; i < ARRAY_CARDINALITY(expected_values1); i++) {
if (expected_values1[i] != values[i]) {
fprintf(stderr,
"Wrong value for %s from virCgroupGetBlkioIoDeviceServiced (expected %lld)\n",
names[i], expected_values1[i]);
goto cleanup;
}
}
ret = 0;
Indent top-level labels by one space in tests/
2014-03-25 10:53:44 +04:00
cleanup:
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
virCgroupFree(&cgroup);
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 17:31:53 +04:00
tests: Rename LIBVIRT_FAKE_SYSFS_DIR to LIBVIRT_FAKE_ROOT_DIR The old name is no longer accurate, since now we're using its value as the root of the fake filesystem. No functional changes.
2015-12-04 15:38:16 +03:00
# define FAKEROOTDIRTEMPLATE abs_builddir "/fakerootdir-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 18:36:52 +04:00
static int
mymain(void)
{
int ret = 0;
tests: Rename LIBVIRT_FAKE_SYSFS_DIR to LIBVIRT_FAKE_ROOT_DIR The old name is no longer accurate, since now we're using its value as the root of the fake filesystem. No functional changes.
2015-12-04 15:38:16 +03:00
char *fakerootdir;
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 18:36:52 +04:00
tests: Rename LIBVIRT_FAKE_SYSFS_DIR to LIBVIRT_FAKE_ROOT_DIR The old name is no longer accurate, since now we're using its value as the root of the fake filesystem. No functional changes.
2015-12-04 15:38:16 +03:00
if (VIR_STRDUP_QUIET(fakerootdir, FAKEROOTDIRTEMPLATE) < 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 18:36:52 +04:00
fprintf(stderr, "Out of memory\n");
abort();
}
tests: Rename LIBVIRT_FAKE_SYSFS_DIR to LIBVIRT_FAKE_ROOT_DIR The old name is no longer accurate, since now we're using its value as the root of the fake filesystem. No functional changes.
2015-12-04 15:38:16 +03:00
if (!mkdtemp(fakerootdir)) {
fprintf(stderr, "Cannot create fakerootdir");
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 18:36:52 +04:00
abort();
}
tests: Rename LIBVIRT_FAKE_SYSFS_DIR to LIBVIRT_FAKE_ROOT_DIR The old name is no longer accurate, since now we're using its value as the root of the fake filesystem. No functional changes.
2015-12-04 15:38:16 +03:00
setenv("LIBVIRT_FAKE_ROOT_DIR", fakerootdir, 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 18:36:52 +04:00
tests: Add tests for virCgroupDetectMounts Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2015-03-17 00:36:52 +03:00
# define DETECT_MOUNTS(file) \
do { \
if (virtTestRun("Detect cgroup mounts for " file, \
testCgroupDetectMounts, \
file) < 0) \
ret = -1; \
} while (0)
DETECT_MOUNTS("ovirt-node-6.6");
DETECT_MOUNTS("ovirt-node-7.1");
DETECT_MOUNTS("fedora-18");
DETECT_MOUNTS("fedora-21");
DETECT_MOUNTS("rhel-7.1");
DETECT_MOUNTS("cgroups1");
DETECT_MOUNTS("cgroups2");
DETECT_MOUNTS("cgroups3");
DETECT_MOUNTS("all-in-one");
DETECT_MOUNTS("no-cgroups");
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 22:13:35 +04:00
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 18:36:52 +04:00
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 22:13:35 +04:00
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 18:36:52 +04:00
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 22:13:35 +04:00
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 22:08:39 +04:00
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 22:13:35 +04:00
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 13:23:51 +04:00
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 22:13:35 +04:00
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 22:08:39 +04:00
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 22:13:35 +04:00
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 13:50:24 +04:00
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 22:08:39 +04:00
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 22:13:35 +04:00
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 22:15:52 +04:00
ret = -1;
vircgroup: Introduce virCgroupControllerAvailable This new internal API checks if given CGroup controller is available. It is going to be needed later when we need to make a decision whether pin domain memory onto NUMA nodes using cpuset CGroup controller or using numa_set_membind(). Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2015-03-31 12:39:13 +03:00
if (virtTestRun("Cgroup controller available", testCgroupControllerAvailable, NULL) < 0)
ret = -1;
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
if (virtTestRun("virCgroupGetBlkioIoServiced works", testCgroupGetBlkioIoServiced, NULL) < 0)
ret = -1;
if (virtTestRun("virCgroupGetBlkioIoDeviceServiced works", testCgroupGetBlkioIoDeviceServiced, NULL) < 0)
ret = -1;
Add unit test for virCgroupGetMemoryUsage.
2014-02-14 21:49:06 +04:00
if (virtTestRun("virCgroupGetMemoryUsage works", testCgroupGetMemoryUsage, NULL) < 0)
ret = -1;
Add unit test for virCgroupGetPercpuStats.
2014-02-14 21:49:08 +04:00
if (virtTestRun("virCgroupGetPercpuStats works", testCgroupGetPercpuStats, NULL) < 0)
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 17:31:53 +04:00
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 22:13:35 +04:00
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 17:31:53 +04:00
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 22:13:35 +04:00
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 22:15:52 +04:00
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 22:13:35 +04:00
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 22:15:52 +04:00
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 22:13:35 +04:00
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 22:15:52 +04:00
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 17:31:53 +04:00
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 18:36:52 +04:00
if (getenv("LIBVIRT_SKIP_CLEANUP") == NULL)
tests: Rename LIBVIRT_FAKE_SYSFS_DIR to LIBVIRT_FAKE_ROOT_DIR The old name is no longer accurate, since now we're using its value as the root of the fake filesystem. No functional changes.
2015-12-04 15:38:16 +03:00
virFileDeleteTree(fakerootdir);
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 18:36:52 +04:00
tests: Rename LIBVIRT_FAKE_SYSFS_DIR to LIBVIRT_FAKE_ROOT_DIR The old name is no longer accurate, since now we're using its value as the root of the fake filesystem. No functional changes.
2015-12-04 15:38:16 +03:00
VIR_FREE(fakerootdir);
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 18:36:52 +04:00
Require spaces around equality comparisons Commit a1cbe4b5 added a check for spaces around assignments and this patch extends it to checks for spaces around '=='. One exception is virAssertCmpInt where comma after '==' is acceptable (since it is a macro and '==' is its argument). Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
2014-03-17 13:38:38 +04:00
return ret == 0 ? EXIT_SUCCESS : EXIT_FAILURE;
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 18:36:52 +04:00
}
VIRT_TEST_MAIN_PRELOAD(mymain, abs_builddir "/.libs/vircgroupmock.so")
#else
int
main(void)
{
return EXIT_AM_SKIP;
}
#endif
Copy Permalink