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
eb56cc6f88
libvirt/tests/vircgroupmock.c

746 lines
25 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
/*
* Copyright (C) 2013 Red Hat, Inc.
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library. If not, see
* <http://www.gnu.org/licenses/>.
*
* Author: Daniel P. Berrange <berrange@redhat.com>
*/
#include <config.h>
#ifdef __linux__
vircgroupmock: Adapt to virmock.h Instead of introducing our own wrapper for dlsym() we can use the one provided by virmock.h. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-13 13:46:35 +03:00
# include "virmock.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 <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# include <fcntl.h>
# include <sys/stat.h>
Include sysmacros.h where needed So in glibc-2.23 sys/sysmacros.h is no longer included from sys/types.h and we don't build because of the usage of major/minor/makedev macros. Autoconf already has AC_HEADER_MAJOR macro that check where exactly these functions/macros are defined, so let's use that. Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
2016-04-14 16:11:45 +03:00
# ifdef MAJOR_IN_MKDEV
# include <sys/mkdev.h>
# elif MAJOR_IN_SYSMACROS
# include <sys/sysmacros.h>
# endif
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 <stdarg.h>
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
# include "testutilslxc.h"
tests: cgroupmock: Use the temporary directory as fake root We might need to mock files living outside SYSFS_PREFIX later on, so it's better to treat the temporary directory we are passed via the environment as the root of the fake filesystem and create SYSFS_PREFIX inside it. The environment variable name will be changed to reflect the new use we're making of it in a later commit.
2015-12-04 12:54:29 +03:00
# include "virstring.h"
# include "virfile.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
vircgroupmock: Adapt to virmock.h Instead of introducing our own wrapper for dlsym() we can use the one provided by virmock.h. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-13 13:46:35 +03:00
static int (*real_open)(const char *path, int flags, ...);
static FILE *(*real_fopen)(const char *path, const char *mode);
static int (*real_access)(const char *path, int mode);
static int (*real_stat)(const char *path, struct stat *sb);
static int (*real___xstat)(int ver, const char *path, struct stat *sb);
static int (*real_lstat)(const char *path, struct stat *sb);
static int (*real___lxstat)(int ver, const char *path, struct stat *sb);
static int (*real_mkdir)(const char *path, mode_t mode);
Workaround issue with clang and inline functions with static vars Clang does not like it when you pass a static variable to an inline function vircgroupmock.c:462:22: error: static variable 'fakesysfsdir' is used in an inline function with external linkage [-Werror,-Wstatic-in-inline] Just make the var non-static to avoid this Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-05-13 15:43:08 +04:00
/* Don't make static, since it causes problems with clang
* when passed as an arg to asprintf()
* vircgroupmock.c:462:22: error: static variable 'fakesysfsdir' is used in an inline function with external linkage [-Werror,-Wstatic-in-inline]
*/
tests: cgroupmock: Use the temporary directory as fake root We might need to mock files living outside SYSFS_PREFIX later on, so it's better to treat the temporary directory we are passed via the environment as the root of the fake filesystem and create SYSFS_PREFIX inside it. The environment variable name will be changed to reflect the new use we're making of it in a later commit.
2015-12-04 12:54:29 +03:00
char *fakerootdir;
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
char *fakesysfscgroupdir;
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
const char *fakedevicedir0 = FAKEDEVDIR0;
const char *fakedevicedir1 = FAKEDEVDIR1;
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: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
# define SYSFS_CGROUP_PREFIX "/not/really/sys/fs/cgroup/"
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
# define SYSFS_CPU_PRESENT "/sys/devices/system/cpu/present"
# define SYSFS_CPU_PRESENT_MOCKED "devices_system_cpu_present"
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
/*
* The plan:
*
* We fake out /proc/mounts, so make it look as is cgroups
* are mounted on /not/really/sys/fs/cgroup. We don't
* use /sys/fs/cgroup, because we want to make it easy to
* detect places where we've not mocked enough syscalls.
*
* In any open/acces/mkdir calls we look at path and if
* it starts with /not/really/sys/fs/cgroup, we rewrite
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
* the path to point at a subdirectory of the temporary
* directory referred to by LIBVIRT_FAKE_ROOT_DIR env
* variable that is set by the main test suite
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
*
* In mkdir() calls, we simulate the cgroups behaviour
* whereby creating the directory auto-creates a bunch
* of files beneath it
*/
/*
* Intentionally missing the 'devices' mount.
* Co-mounting cpu & cpuacct controllers
* An anonymous controller for systemd
*/
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
const char *procmounts =
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
"rootfs / rootfs rw 0 0\n"
"tmpfs /run tmpfs rw,seclabel,nosuid,nodev,mode=755 0 0\n"
"tmpfs /not/really/sys/fs/cgroup tmpfs rw,seclabel,nosuid,nodev,noexec,mode=755 0 0\n"
"cgroup /not/really/sys/fs/cgroup/systemd cgroup rw,nosuid,nodev,noexec,relatime,release_agent=/usr/lib/systemd/systemd-cgroups-agent,name=systemd 0 0\n"
"cgroup /not/really/sys/fs/cgroup/cpuset cgroup rw,nosuid,nodev,noexec,relatime,cpuset 0 0\n"
"cgroup /not/really/sys/fs/cgroup/cpu,cpuacct cgroup rw,nosuid,nodev,noexec,relatime,cpuacct,cpu 0 0\n"
"cgroup /not/really/sys/fs/cgroup/freezer cgroup rw,nosuid,nodev,noexec,relatime,freezer 0 0\n"
"cgroup /not/really/sys/fs/cgroup/blkio cgroup rw,nosuid,nodev,noexec,relatime,blkio 0 0\n"
"cgroup /not/really/sys/fs/cgroup/memory cgroup rw,nosuid,nodev,noexec,relatime,memory 0 0\n"
"/dev/sda1 /boot ext4 rw,seclabel,relatime,data=ordered 0 0\n"
"tmpfs /tmp tmpfs rw,seclabel,relatime,size=1024000k 0 0\n";
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
const char *procselfcgroups =
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
"115:memory:/\n"
"8:blkio:/\n"
"6:freezer:/\n"
"3:cpuacct,cpu:/system\n"
"2:cpuset:/\n"
"1:name=systemd:/user/berrange/123\n";
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
const char *proccgroups =
"#subsys_name hierarchy num_cgroups enabled\n"
"cpuset 2 4 1\n"
"cpu 3 48 1\n"
"cpuacct 3 48 1\n"
"memory 4 4 1\n"
"devices 5 4 1\n"
"freezer 6 4 1\n"
"blkio 8 4 1\n";
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 *procmountsallinone =
"rootfs / rootfs rw 0 0\n"
"sysfs /sys sysfs rw,nosuid,nodev,noexec,relatime 0 0\n"
"proc /proc proc rw,nosuid,nodev,noexec,relatime 0 0\n"
"udev /dev devtmpfs rw,relatime,size=16458560k,nr_inodes=4114640,mode=755 0 0\n"
"devpts /dev/pts devpts rw,nosuid,noexec,relatime,gid=5,mode=620,ptmxmode=000 0 0\n"
"nfsd /proc/fs/nfsd nfsd rw,relatime 0 0\n"
"cgroup /not/really/sys/fs/cgroup cgroup rw,relatime,blkio,devices,memory,cpuacct,cpu,cpuset 0 0\n";
const char *procselfcgroupsallinone =
"6:blkio,devices,memory,cpuacct,cpu,cpuset:/";
const char *proccgroupsallinone =
"#subsys_name hierarchy num_cgroups enabled\n"
"cpuset 6 1 1\n"
"cpu 6 1 1\n"
"cpuacct 6 1 1\n"
"memory 6 1 1\n"
"devices 6 1 1\n"
"blkio 6 1 1\n";
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 *procmountslogind =
"none /not/really/sys/fs/cgroup tmpfs rw,rootcontext=system_u:object_r:sysfs_t:s0,seclabel,relatime,size=4k,mode=755 0 0\n"
"systemd /not/really/sys/fs/cgroup/systemd cgroup rw,nosuid,nodev,noexec,relatime,name=systemd 0 0\n";
const char *procselfcgroupslogind =
"1:name=systemd:/\n";
const char *proccgroupslogind =
"#subsys_name hierarchy num_cgroups enabled\n"
"cpuset 0 1 1\n"
"cpu 0 1 1\n"
"cpuacct 0 1 1\n"
"memory 0 1 0\n"
"devices 0 1 1\n"
"freezer 0 1 1\n"
"net_cls 0 1 1\n"
"blkio 0 1 1\n"
"perf_event 0 1 1\n";
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 make_file(const char *path,
const char *name,
const char *value)
{
int fd = -1;
int ret = -1;
char *filepath = NULL;
if (asprintf(&filepath, "%s/%s", path, name) < 0)
return -1;
vircgroupmock: Adapt to virmock.h Instead of introducing our own wrapper for dlsym() we can use the one provided by virmock.h. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-13 13:46:35 +03:00
if ((fd = real_open(filepath, O_CREAT|O_WRONLY, 0600)) < 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
goto cleanup;
if (write(fd, value, strlen(value)) != strlen(value))
goto cleanup;
ret = 0;
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
if (fd != -1 && close(fd) < 0)
ret = -1;
free(filepath);
return ret;
}
static int make_controller(const char *path, mode_t mode)
{
int ret = -1;
const char *controller;
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
if (!STRPREFIX(path, fakesysfscgroupdir)) {
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
errno = EINVAL;
return -1;
}
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
controller = path + strlen(fakesysfscgroupdir) + 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 (STREQ(controller, "cpu"))
return symlink("cpu,cpuacct", path);
if (STREQ(controller, "cpuacct"))
return symlink("cpu,cpuacct", path);
vircgroupmock: Adapt to virmock.h Instead of introducing our own wrapper for dlsym() we can use the one provided by virmock.h. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-13 13:46:35 +03:00
if (real_mkdir(path, mode) < 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
goto cleanup;
# define MAKE_FILE(name, value) \
do { \
if (make_file(path, name, value) < 0) \
goto cleanup; \
} while (0)
if (STRPREFIX(controller, "cpu,cpuacct")) {
MAKE_FILE("cpu.cfs_period_us", "100000\n");
MAKE_FILE("cpu.cfs_quota_us", "-1\n");
MAKE_FILE("cpu.rt_period_us", "1000000\n");
MAKE_FILE("cpu.rt_runtime_us", "950000\n");
MAKE_FILE("cpu.shares", "1024\n");
MAKE_FILE("cpu.stat",
"nr_periods 0\n"
"nr_throttled 0\n"
"throttled_time 0\n");
MAKE_FILE("cpuacct.stat",
"user 216687025\n"
"system 43421396\n");
MAKE_FILE("cpuacct.usage", "2787788855799582\n");
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
MAKE_FILE("cpuacct.usage_percpu",
"7059492996 0 0 0 0 0 0 0 4180532496 0 0 0 0 0 0 0 "
"1957541268 0 0 0 0 0 0 0 2065932204 0 0 0 0 0 0 0 "
"18228689414 0 0 0 0 0 0 0 4245525148 0 0 0 0 0 0 0 "
"2911161568 0 0 0 0 0 0 0 1407758136 0 0 0 0 0 0 0 "
"1836807700 0 0 0 0 0 0 0 1065296618 0 0 0 0 0 0 0 "
"2046213266 0 0 0 0 0 0 0 747889778 0 0 0 0 0 0 0 "
"709566900 0 0 0 0 0 0 0 444777342 0 0 0 0 0 0 0 "
"5683512916 0 0 0 0 0 0 0 635751356 0 0 0 0 0 0 0\n");
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
} else if (STRPREFIX(controller, "cpuset")) {
MAKE_FILE("cpuset.cpu_exclusive", "1\n");
if (STREQ(controller, "cpuset"))
MAKE_FILE("cpuset.cpus", "0-1");
else
MAKE_FILE("cpuset.cpus", ""); /* Values don't inherit */
MAKE_FILE("cpuset.mem_exclusive", "1\n");
MAKE_FILE("cpuset.mem_hardwall", "0\n");
MAKE_FILE("cpuset.memory_migrate", "0\n");
MAKE_FILE("cpuset.memory_pressure", "0\n");
MAKE_FILE("cpuset.memory_pressure_enabled", "0\n");
MAKE_FILE("cpuset.memory_spread_page", "0\n");
MAKE_FILE("cpuset.memory_spread_slab", "0\n");
if (STREQ(controller, "cpuset"))
MAKE_FILE("cpuset.mems", "0");
else
MAKE_FILE("cpuset.mems", ""); /* Values don't inherit */
MAKE_FILE("cpuset.sched_load_balance", "1\n");
MAKE_FILE("cpuset.sched_relax_domain_level", "-1\n");
} else if (STRPREFIX(controller, "memory")) {
MAKE_FILE("memory.failcnt", "0\n");
MAKE_FILE("memory.force_empty", ""); /* Write only */
MAKE_FILE("memory.kmem.tcp.failcnt", "0\n");
MAKE_FILE("memory.kmem.tcp.limit_in_bytes", "9223372036854775807\n");
MAKE_FILE("memory.kmem.tcp.max_usage_in_bytes", "0\n");
MAKE_FILE("memory.kmem.tcp.usage_in_bytes", "16384\n");
MAKE_FILE("memory.limit_in_bytes", "9223372036854775807\n");
MAKE_FILE("memory.max_usage_in_bytes", "0\n");
MAKE_FILE("memory.memsw.failcnt", ""); /* Not supported */
MAKE_FILE("memory.memsw.limit_in_bytes", ""); /* Not supported */
MAKE_FILE("memory.memsw.max_usage_in_bytes", ""); /* Not supported */
MAKE_FILE("memory.memsw.usage_in_bytes", ""); /* Not supported */
MAKE_FILE("memory.move_charge_at_immigrate", "0\n");
MAKE_FILE("memory.numa_stat",
"total=367664 N0=367664\n"
"file=314764 N0=314764\n"
"anon=51999 N0=51999\n"
"unevictable=901 N0=901\n");
MAKE_FILE("memory.oom_control",
"oom_kill_disable 0\n"
"under_oom 0\n");
MAKE_FILE("memory.soft_limit_in_bytes", "9223372036854775807\n");
MAKE_FILE("memory.stat",
"cache 1336619008\n"
"rss 97792000\n"
"mapped_file 42090496\n"
"pgpgin 13022605027\n"
"pgpgout 13023820533\n"
"pgfault 54429417056\n"
"pgmajfault 315715\n"
"inactive_anon 145887232\n"
"active_anon 67100672\n"
"inactive_file 627400704\n"
"active_file 661872640\n"
"unevictable 3690496\n"
"hierarchical_memory_limit 9223372036854775807\n"
"total_cache 1336635392\n"
"total_rss 118689792\n"
"total_mapped_file 42106880\n"
"total_pgpgin 13022606816\n"
"total_pgpgout 13023820793\n"
"total_pgfault 54429422313\n"
"total_pgmajfault 315715\n"
"total_inactive_anon 145891328\n"
"total_active_anon 88010752\n"
"total_inactive_file 627400704\n"
"total_active_file 661872640\n"
"total_unevictable 3690496\n"
"recent_rotated_anon 112807028\n"
"recent_rotated_file 2547948\n"
"recent_scanned_anon 113796164\n"
"recent_scanned_file 8199863\n");
MAKE_FILE("memory.swappiness", "60\n");
MAKE_FILE("memory.usage_in_bytes", "1455321088\n");
MAKE_FILE("memory.use_hierarchy", "0\n");
} else if (STRPREFIX(controller, "freezer")) {
MAKE_FILE("freezer.state", "THAWED");
} else if (STRPREFIX(controller, "blkio")) {
MAKE_FILE("blkio.io_merged",
"8:0 Read 1100949\n"
"8:0 Write 2248076\n"
"8:0 Sync 63063\n"
"8:0 Async 3285962\n"
"8:0 Total 3349025\n");
MAKE_FILE("blkio.io_queued",
"8:0 Read 0\n"
"8:0 Write 0\n"
"8:0 Sync 0\n"
"8:0 Async 0\n"
"8:0 Total 0\n");
MAKE_FILE("blkio.io_service_bytes",
"8:0 Read 59542078464\n"
"8:0 Write 397369182208\n"
"8:0 Sync 234080922624\n"
"8:0 Async 222830338048\n"
"8:0 Total 456911260672\n");
MAKE_FILE("blkio.io_serviced",
"8:0 Read 3402504\n"
"8:0 Write 14966516\n"
"8:0 Sync 12064031\n"
"8:0 Async 6304989\n"
"8:0 Total 18369020\n");
MAKE_FILE("blkio.io_service_time",
"8:0 Read 10747537542349\n"
"8:0 Write 9200028590575\n"
"8:0 Sync 6449319855381\n"
"8:0 Async 13498246277543\n"
"8:0 Total 19947566132924\n");
MAKE_FILE("blkio.io_wait_time",
"8:0 Read 14687514824889\n"
"8:0 Write 357748452187691\n"
"8:0 Sync 55296974349413\n"
"8:0 Async 317138992663167\n"
"8:0 Total 372435967012580\n");
MAKE_FILE("blkio.reset_stats", ""); /* Write only */
MAKE_FILE("blkio.sectors", "8:0 892404806\n");
MAKE_FILE("blkio.throttle.io_service_bytes",
"8:0 Read 59542107136\n"
"8:0 Write 411440480256\n"
"8:0 Sync 248486822912\n"
"8:0 Async 222495764480\n"
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
"8:0 Total 470982587392\n"
"9:0 Read 59542107137\n"
"9:0 Write 411440480257\n"
"9:0 Sync 248486822912\n"
"9:0 Async 222495764480\n"
"9:0 Total 470982587392\n");
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
MAKE_FILE("blkio.throttle.io_serviced",
"8:0 Read 4832583\n"
"8:0 Write 36641903\n"
"8:0 Sync 30723171\n"
"8:0 Async 10751315\n"
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
"8:0 Total 41474486\n"
"9:0 Read 4832584\n"
"9:0 Write 36641904\n"
"9:0 Sync 30723171\n"
"9:0 Async 10751315\n"
"9:0 Total 41474486\n");
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
MAKE_FILE("blkio.throttle.read_bps_device", "");
MAKE_FILE("blkio.throttle.read_iops_device", "");
MAKE_FILE("blkio.throttle.write_bps_device", "");
MAKE_FILE("blkio.throttle.write_iops_device", "");
MAKE_FILE("blkio.time", "8:0 61019089\n");
MAKE_FILE("blkio.weight", "1000\n");
MAKE_FILE("blkio.weight_device", "");
} else {
errno = EINVAL;
goto cleanup;
}
ret = 0;
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
return ret;
}
static void init_syms(void)
{
vircgroupmock: Adapt to virmock.h Instead of introducing our own wrapper for dlsym() we can use the one provided by virmock.h. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-13 13:46:35 +03:00
if (real_fopen)
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
return;
vircgroupmock: Adapt to virmock.h Instead of introducing our own wrapper for dlsym() we can use the one provided by virmock.h. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-13 13:46:35 +03:00
VIR_MOCK_REAL_INIT(fopen);
VIR_MOCK_REAL_INIT(access);
VIR_MOCK_REAL_INIT_ALT(lstat, __lxstat);
VIR_MOCK_REAL_INIT_ALT(stat, __xstat);
VIR_MOCK_REAL_INIT(mkdir);
VIR_MOCK_REAL_INIT(open);
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 void init_sysfs(void)
{
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
if (fakerootdir && fakesysfscgroupdir)
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
return;
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 (!(fakerootdir = getenv("LIBVIRT_FAKE_ROOT_DIR"))) {
fprintf(stderr, "Missing LIBVIRT_FAKE_ROOT_DIR env variable\n");
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: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
if (virAsprintfQuiet(&fakesysfscgroupdir, "%s%s",
fakerootdir, SYSFS_CGROUP_PREFIX) < 0)
tests: cgroupmock: Use the temporary directory as fake root We might need to mock files living outside SYSFS_PREFIX later on, so it's better to treat the temporary directory we are passed via the environment as the root of the fake filesystem and create SYSFS_PREFIX inside it. The environment variable name will be changed to reflect the new use we're making of it in a later commit.
2015-12-04 12:54:29 +03:00
abort();
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
if (virFileMakePath(fakesysfscgroupdir) < 0) {
fprintf(stderr, "Cannot create %s\n", fakesysfscgroupdir);
tests: cgroupmock: Use the temporary directory as fake root We might need to mock files living outside SYSFS_PREFIX later on, so it's better to treat the temporary directory we are passed via the environment as the root of the fake filesystem and create SYSFS_PREFIX inside it. The environment variable name will be changed to reflect the new use we're making of it in a later commit.
2015-12-04 12:54:29 +03:00
abort();
}
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
# define MAKE_CONTROLLER(subpath) \
do { \
char *path; \
if (asprintf(&path, "%s/%s", fakesysfscgroupdir, subpath) < 0) \
abort(); \
if (make_controller(path, 0755) < 0) { \
fprintf(stderr, "Cannot initialize %s\n", path); \
tests: fix some resource leaks Found by running valgrind for these tests. Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
2017-04-09 14:28:07 +03:00
free(path); \
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
abort(); \
} \
tests: fix some resource leaks Found by running valgrind for these tests. Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
2017-04-09 14:28:07 +03:00
free(path); \
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
} while (0)
MAKE_CONTROLLER("cpu");
MAKE_CONTROLLER("cpuacct");
MAKE_CONTROLLER("cpu,cpuacct");
MAKE_CONTROLLER("cpu,cpuacct/system");
MAKE_CONTROLLER("cpuset");
MAKE_CONTROLLER("blkio");
MAKE_CONTROLLER("memory");
MAKE_CONTROLLER("freezer");
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
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
if (make_file(fakesysfscgroupdir,
SYSFS_CPU_PRESENT_MOCKED, "8-23,48-159\n") < 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
abort();
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
}
FILE *fopen(const char *path, const char *mode)
{
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 *mock;
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
bool allinone = false, logind = false;
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
init_syms();
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
mock = getenv("VIR_CGROUP_MOCK_MODE");
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
if (mock) {
if (STREQ(mock, "allinone"))
allinone = true;
else if (STREQ(mock, "logind"))
logind = true;
}
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
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 (STREQ(path, "/proc/mounts")) {
if (STREQ(mode, "r")) {
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
if (allinone)
return fmemopen((void *)procmountsallinone,
strlen(procmountsallinone), mode);
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
else if (logind)
return fmemopen((void *)procmountslogind,
strlen(procmountslogind), mode);
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
else
return fmemopen((void *)procmounts, strlen(procmounts), mode);
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
} else {
errno = EACCES;
return NULL;
}
}
if (STREQ(path, "/proc/cgroups")) {
if (STREQ(mode, "r")) {
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
if (allinone)
return fmemopen((void *)proccgroupsallinone,
strlen(proccgroupsallinone), mode);
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
else if (logind)
return fmemopen((void *)proccgroupslogind,
strlen(proccgroupslogind), mode);
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
else
return fmemopen((void *)proccgroups, strlen(proccgroups), 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
} else {
errno = EACCES;
return NULL;
}
}
if (STREQ(path, "/proc/self/cgroup")) {
if (STREQ(mode, "r")) {
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
if (allinone)
return fmemopen((void *)procselfcgroupsallinone,
strlen(procselfcgroupsallinone), mode);
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
else if (logind)
return fmemopen((void *)procselfcgroupslogind,
strlen(procselfcgroupslogind), mode);
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
else
return fmemopen((void *)procselfcgroups, strlen(procselfcgroups), 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
} else {
errno = EACCES;
return NULL;
}
}
vircgroupmock: Adapt to virmock.h Instead of introducing our own wrapper for dlsym() we can use the one provided by virmock.h. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-13 13:46:35 +03:00
return real_fopen(path, 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
}
int access(const char *path, int mode)
{
int ret;
init_syms();
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
if (STRPREFIX(path, SYSFS_CGROUP_PREFIX)) {
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
init_sysfs();
char *newpath;
if (asprintf(&newpath, "%s/%s",
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
fakesysfscgroupdir,
path + strlen(SYSFS_CGROUP_PREFIX)) < 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
errno = ENOMEM;
return -1;
}
vircgroupmock: Adapt to virmock.h Instead of introducing our own wrapper for dlsym() we can use the one provided by virmock.h. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-13 13:46:35 +03:00
ret = real_access(newpath, 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
free(newpath);
vircgroupmock: Mock access() to some more files Currently, if access(path, mode) is invoked, we check if @path has this special prefix SYSFS_PREFIX. If it does, we modify the path a bit and call realaccess. If it doesn't we act just like a wrapper and call realaccess directly. However, we are mocking fopen() as well. And as one can clearly see there, fopen("/proc/cgroups") will succeed. Hence, we have an error in our mocked access(): We need to check whether @path is not equal to /proc/cgroups as it may not exists on real system we're running however we definitely know how to fopen() it. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2013-10-22 16:15:21 +04:00
} else if (STREQ(path, "/proc/cgroups") ||
vircgroupmock: Mock access("/sys/devices/system/cpu/present") There's been a report on the upstream list [1] describing we access /sys/devices/system/cpu/present directly on the host from within our test suite. This may end up in unpredictable results as no all linux systems are required to have that file. Mock access to the file. libvirt.git/tests $ ../run strace vircgrouptest ... access("/sys/devices/system/cpu/present", F_OK) = 0 ... Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-02-25 10:44:46 +03:00
STREQ(path, "/proc/self/cgroup") ||
STREQ(path, SYSFS_CPU_PRESENT)) {
vircgroupmock: Mock access() to some more files Currently, if access(path, mode) is invoked, we check if @path has this special prefix SYSFS_PREFIX. If it does, we modify the path a bit and call realaccess. If it doesn't we act just like a wrapper and call realaccess directly. However, we are mocking fopen() as well. And as one can clearly see there, fopen("/proc/cgroups") will succeed. Hence, we have an error in our mocked access(): We need to check whether @path is not equal to /proc/cgroups as it may not exists on real system we're running however we definitely know how to fopen() it. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2013-10-22 16:15:21 +04:00
/* These files are readable for all. */
ret = (mode == F_OK || mode == R_OK) ? 0 : -1;
} else if (STREQ(path, "/proc/mounts")) {
/* This one is accessible anytime for anybody. In fact, it's just
* a symlink to /proc/self/mounts. */
ret = 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
} else {
vircgroupmock: Adapt to virmock.h Instead of introducing our own wrapper for dlsym() we can use the one provided by virmock.h. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-13 13:46:35 +03:00
ret = real_access(path, 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
}
return ret;
}
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
int __lxstat(int ver, const char *path, struct stat *sb)
{
int ret;
init_syms();
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
if (STRPREFIX(path, SYSFS_CGROUP_PREFIX)) {
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
init_sysfs();
char *newpath;
if (asprintf(&newpath, "%s/%s",
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
fakesysfscgroupdir,
path + strlen(SYSFS_CGROUP_PREFIX)) < 0) {
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
errno = ENOMEM;
return -1;
}
tests: Drop VIR_MOCK_CALL_STAT It wasn't as great idea as I thought. Thing around stat() are more complicated than that. Therefore we need to revert 86d1705a8ace349b plus drop use of the macro as introduced in later patches. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-16 18:15:46 +03:00
ret = real___lxstat(ver, newpath, sb);
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
free(newpath);
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
} else if (STRPREFIX(path, fakedevicedir0)) {
sb->st_mode = S_IFBLK;
sb->st_rdev = makedev(8, 0);
return 0;
} else if (STRPREFIX(path, fakedevicedir1)) {
sb->st_mode = S_IFBLK;
sb->st_rdev = makedev(9, 0);
return 0;
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
} else {
tests: Drop VIR_MOCK_CALL_STAT It wasn't as great idea as I thought. Thing around stat() are more complicated than that. Therefore we need to revert 86d1705a8ace349b plus drop use of the macro as introduced in later patches. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-16 18:15:46 +03:00
ret = real___lxstat(ver, path, sb);
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
}
return ret;
}
int lstat(const char *path, struct stat *sb)
{
int ret;
init_syms();
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
if (STRPREFIX(path, SYSFS_CGROUP_PREFIX)) {
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
init_sysfs();
char *newpath;
if (asprintf(&newpath, "%s/%s",
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
fakesysfscgroupdir,
path + strlen(SYSFS_CGROUP_PREFIX)) < 0) {
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
errno = ENOMEM;
return -1;
}
tests: Drop VIR_MOCK_CALL_STAT It wasn't as great idea as I thought. Thing around stat() are more complicated than that. Therefore we need to revert 86d1705a8ace349b plus drop use of the macro as introduced in later patches. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-16 18:15:46 +03:00
ret = real_lstat(newpath, sb);
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
free(newpath);
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
} else if (STRPREFIX(path, fakedevicedir0)) {
sb->st_mode = S_IFBLK;
sb->st_rdev = makedev(8, 0);
return 0;
} else if (STRPREFIX(path, fakedevicedir1)) {
sb->st_mode = S_IFBLK;
sb->st_rdev = makedev(9, 0);
return 0;
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
} else {
tests: Drop VIR_MOCK_CALL_STAT It wasn't as great idea as I thought. Thing around stat() are more complicated than that. Therefore we need to revert 86d1705a8ace349b plus drop use of the macro as introduced in later patches. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-16 18:15:46 +03:00
ret = real_lstat(path, sb);
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
}
return ret;
}
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
int __xstat(int ver, const char *path, struct stat *sb)
{
int ret;
init_syms();
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
if (STRPREFIX(path, SYSFS_CGROUP_PREFIX)) {
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
init_sysfs();
char *newpath;
if (asprintf(&newpath, "%s/%s",
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
fakesysfscgroupdir,
path + strlen(SYSFS_CGROUP_PREFIX)) < 0) {
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
errno = ENOMEM;
return -1;
}
tests: Drop VIR_MOCK_CALL_STAT It wasn't as great idea as I thought. Thing around stat() are more complicated than that. Therefore we need to revert 86d1705a8ace349b plus drop use of the macro as introduced in later patches. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-16 18:15:46 +03:00
ret = real___xstat(ver, newpath, sb);
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
free(newpath);
} else if (STRPREFIX(path, fakedevicedir0)) {
sb->st_mode = S_IFBLK;
sb->st_rdev = makedev(8, 0);
return 0;
} else if (STRPREFIX(path, fakedevicedir1)) {
sb->st_mode = S_IFBLK;
sb->st_rdev = makedev(9, 0);
return 0;
} else {
tests: Drop VIR_MOCK_CALL_STAT It wasn't as great idea as I thought. Thing around stat() are more complicated than that. Therefore we need to revert 86d1705a8ace349b plus drop use of the macro as introduced in later patches. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-16 18:15:46 +03:00
ret = real___xstat(ver, path, sb);
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
}
return ret;
}
int stat(const char *path, struct stat *sb)
{
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
char *newpath = NULL;
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
int ret;
init_syms();
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 (STREQ(path, SYSFS_CPU_PRESENT)) {
init_sysfs();
if (asprintf(&newpath, "%s/%s",
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
fakesysfscgroupdir,
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
SYSFS_CPU_PRESENT_MOCKED) < 0) {
errno = ENOMEM;
return -1;
}
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
} else if (STRPREFIX(path, SYSFS_CGROUP_PREFIX)) {
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
init_sysfs();
if (asprintf(&newpath, "%s/%s",
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
fakesysfscgroupdir,
path + strlen(SYSFS_CGROUP_PREFIX)) < 0) {
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
errno = ENOMEM;
return -1;
}
} else if (STRPREFIX(path, fakedevicedir0)) {
sb->st_mode = S_IFBLK;
sb->st_rdev = makedev(8, 0);
return 0;
} else if (STRPREFIX(path, fakedevicedir1)) {
sb->st_mode = S_IFBLK;
sb->st_rdev = makedev(9, 0);
return 0;
} else {
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 (!(newpath = strdup(path)))
return -1;
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
}
tests: Drop VIR_MOCK_CALL_STAT It wasn't as great idea as I thought. Thing around stat() are more complicated than that. Therefore we need to revert 86d1705a8ace349b plus drop use of the macro as introduced in later patches. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-16 18:15:46 +03:00
ret = real_stat(newpath, sb);
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
free(newpath);
Add unit test for virCgroupGetBlkioIo*Serviced
2014-02-14 21:49:05 +04:00
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
int mkdir(const char *path, mode_t mode)
{
int ret;
init_syms();
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
if (STRPREFIX(path, SYSFS_CGROUP_PREFIX)) {
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
init_sysfs();
char *newpath;
if (asprintf(&newpath, "%s/%s",
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
fakesysfscgroupdir,
path + strlen(SYSFS_CGROUP_PREFIX)) < 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
errno = ENOMEM;
return -1;
}
ret = make_controller(newpath, mode);
free(newpath);
} else {
vircgroupmock: Adapt to virmock.h Instead of introducing our own wrapper for dlsym() we can use the one provided by virmock.h. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-13 13:46:35 +03:00
ret = real_mkdir(path, 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
}
return ret;
}
int open(const char *path, int flags, ...)
{
int ret;
char *newpath = NULL;
init_syms();
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 (STREQ(path, SYSFS_CPU_PRESENT)) {
init_sysfs();
if (asprintf(&newpath, "%s/%s",
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
fakesysfscgroupdir,
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
SYSFS_CPU_PRESENT_MOCKED) < 0) {
errno = ENOMEM;
return -1;
}
}
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
if (STRPREFIX(path, SYSFS_CGROUP_PREFIX)) {
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
init_sysfs();
if (asprintf(&newpath, "%s/%s",
tests: Use more specific names for variables Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and fakesysfscgroupdir. This makes it explicit what part of the fake sysfs filesystem they're referring to, and also leaves open the possibility of handling files in two unrelated parts of the fake sysfs filesystem. No functional changes.
2015-12-04 16:51:08 +03:00
fakesysfscgroupdir,
path + strlen(SYSFS_CGROUP_PREFIX)) < 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
errno = ENOMEM;
return -1;
}
}
if (flags & O_CREAT) {
va_list ap;
mode_t mode;
va_start(ap, flags);
mode = va_arg(ap, mode_t);
va_end(ap);
vircgroupmock: Adapt to virmock.h Instead of introducing our own wrapper for dlsym() we can use the one provided by virmock.h. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-13 13:46:35 +03:00
ret = real_open(newpath ? newpath : path, flags, 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
} else {
vircgroupmock: Adapt to virmock.h Instead of introducing our own wrapper for dlsym() we can use the one provided by virmock.h. Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-13 13:46:35 +03:00
ret = real_open(newpath ? newpath : path, flags);
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
}
free(newpath);
return ret;
}
#else
/* Nothing to override on non-__linux__ platforms */
#endif
Copy Permalink