We cannot detect only mount points to figure out whether cgroup v2 is available because systemd uses cgroup v2 for process tracking and all controllers are mounted as cgroup v1 controllers. To make sure that this is no the situation we need to check 'cgroup.controllers' file if it's not empty to make sure that cgroup v2 is not mounted only for process tracking. Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 59 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 59 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index 23bf81dae2..83944602d6 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -19,6 +19,10 @@ */ #include <config.h> +#ifdef __linux__ +# include <mntent.h> +#endif /* __linux__ */ + #include "internal.h" #define __VIR_CGROUP_ALLOW_INCLUDE_PRIV_H__ @@ -28,7 +32,9 @@ #include "vircgroup.h" #include "vircgroupbackend.h" #include "vircgroupv2.h" +#include "virfile.h" #include "virlog.h" +#include "virstring.h" VIR_LOG_INIT("util.cgroup"); @@ -41,8 +47,61 @@ VIR_ENUM_IMPL(virCgroupV2Controller, VIR_CGROUP_CONTROLLER_LAST, #ifdef __linux__ +/* We're looking for one 'cgroup2' fs mount which has some + * controllers enabled. */ +static bool +virCgroupV2Available(void) +{ + bool ret = false; + FILE *mounts = NULL; + struct mntent entry; + char buf[CGROUP_MAX_VAL]; + + if (!(mounts = fopen("/proc/mounts", "r"))) + return false; + + while (getmntent_r(mounts, &entry, buf, sizeof(buf)) != NULL) { + if (STREQ(entry.mnt_type, "cgroup2")) { + ret = true; + break; + } + } + + /* Systemd uses cgroup v2 for process tracking but no controller is + * available. We should consider this configuration as cgroup v2 is + * not available. */ + if (ret) { + int rc; + VIR_AUTOFREE(char *) contFile = NULL; + VIR_AUTOFREE(char *) contStr = NULL; + + if (virAsprintf(&contFile, "%s/cgroup.controllers", entry.mnt_dir) < 0) { + ret = false; + goto cleanup; + } + + rc = virFileReadAll(contFile, 1024 * 1024, &contStr); + if (rc < 0) { + ret = false; + goto cleanup; + } + + if (STREQ(contStr, "")) { + ret = false; + goto cleanup; + } + } + + cleanup: + VIR_FORCE_FCLOSE(mounts); + return ret; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, + + .available = virCgroupV2Available, }; -- 2.17.1

When reconnecting to a domain we are validating the cgroup name. In case of cgroup v2 we need to validate only the new format for host without systemd '{machinename}.libvirt-{drivername}' or scope name generated by systemd. Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 43 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 43 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index 83944602d6..77e7695ac5 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -35,6 +35,7 @@ #include "virfile.h" #include "virlog.h" #include "virstring.h" +#include "virsystemd.h" VIR_LOG_INIT("util.cgroup"); @@ -98,10 +99,52 @@ virCgroupV2Available(void) } +static bool +virCgroupV2ValidateMachineGroup(virCgroupPtr group, + const char *name ATTRIBUTE_UNUSED, + const char *drivername, + const char *machinename) +{ + VIR_AUTOFREE(char *) partmachinename = NULL; + VIR_AUTOFREE(char *) scopename = NULL; + char *tmp; + + if (virAsprintf(&partmachinename, "%s.libvirt-%s", machinename, + drivername) < 0) { + return false; + } + + if (virCgroupPartitionEscape(&partmachinename) < 0) + return false; + + if (!(scopename = virSystemdMakeScopeName(machinename, drivername, + false))) { + return false; + } + + if (virCgroupPartitionEscape(&scopename) < 0) + return false; + + if (!(tmp = strrchr(group->unified.placement, '/'))) + return false; + tmp++; + + if (STRNEQ(tmp, partmachinename) && + STRNEQ(tmp, scopename)) { + VIR_DEBUG("Name '%s' for unified does not match '%s' or '%s'", + tmp, partmachinename, scopename); + return false; + } + + return true; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, .available = virCgroupV2Available, + .validateMachineGroup = virCgroupV2ValidateMachineGroup, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 27 +++++++++++++++++++++++++++ 1 file changed, 27 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index b4c551d1d2..28f4dd6ff0 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -148,12 +148,39 @@ virCgroupV2CopyMounts(virCgroupPtr group, } +static int +virCgroupV2CopyPlacement(virCgroupPtr group, + const char *path, + virCgroupPtr parent) +{ + if (path[0] == '/') { + if (VIR_STRDUP(group->unified.placement, path) < 0) + return -1; + } else { + /* + * parent == "/" + path="" => "/" + * parent == "/libvirt.service" + path == "" => "/libvirt.service" + * parent == "/libvirt.service" + path == "foo" => "/libvirt.service/foo" + */ + if (virAsprintf(&group->unified.placement, "%s%s%s", + parent->unified.placement, + (STREQ(parent->unified.placement, "/") || + STREQ(path, "") ? "" : "/"), + path) < 0) + return -1; + } + + return 0; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, .available = virCgroupV2Available, .validateMachineGroup = virCgroupV2ValidateMachineGroup, .copyMounts = virCgroupV2CopyMounts, + .copyPlacement = virCgroupV2CopyPlacement, }; -- 2.17.1

If the placement was copied from parent or set to absolute path there is nothing to do, otherwise set the placement based on process placement from /proc/self/cgroup or /proc/{pid}/cgroup. Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 26 ++++++++++++++++++++++++++ 1 file changed, 26 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index f13c5fc669..800fc083f2 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -189,6 +189,31 @@ virCgroupV2DetectMounts(virCgroupPtr group, } +static int +virCgroupV2DetectPlacement(virCgroupPtr group, + const char *path, + const char *controllers ATTRIBUTE_UNUSED, + const char *selfpath) +{ + if (group->unified.placement) + return 0; + + /* + * selfpath == "/" + path="" -> "/" + * selfpath == "/libvirt.service" + path == "" -> "/libvirt.service" + * selfpath == "/libvirt.service" + path == "foo" -> "/libvirt.service/foo" + */ + if (virAsprintf(&group->unified.placement, + "%s%s%s", selfpath, + (STREQ(selfpath, "/") || + STREQ(path, "") ? "" : "/"), + path) < 0) + return -1; + + return 0; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -197,6 +222,7 @@ virCgroupBackend virCgroupV2Backend = { .copyMounts = virCgroupV2CopyMounts, .copyPlacement = virCgroupV2CopyPlacement, .detectMounts = virCgroupV2DetectMounts, + .detectPlacement = virCgroupV2DetectPlacement, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index 307a0b5142..a06226e414 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -228,6 +228,17 @@ virCgroupV2ValidatePlacement(virCgroupPtr group, } +static char * +virCgroupV2StealPlacement(virCgroupPtr group) +{ + char *ret; + + VIR_STEAL_PTR(ret, group->unified.placement); + + return ret; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -238,6 +249,7 @@ virCgroupBackend virCgroupV2Backend = { .detectMounts = virCgroupV2DetectMounts, .detectPlacement = virCgroupV2DetectPlacement, .validatePlacement = virCgroupV2ValidatePlacement, + .stealPlacement = virCgroupV2StealPlacement, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index 77f3c3e978..c4e315c91a 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -307,6 +307,14 @@ virCgroupV2DetectControllers(virCgroupPtr group, } +static bool +virCgroupV2HasController(virCgroupPtr group, + int controller) +{ + return group->unified.controllers & (1 << controller); +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -319,6 +327,7 @@ virCgroupBackend virCgroupV2Backend = { .validatePlacement = virCgroupV2ValidatePlacement, .stealPlacement = virCgroupV2StealPlacement, .detectControllers = virCgroupV2DetectControllers, + .hasController = virCgroupV2HasController, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index e17ceccac8..29d320e630 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -323,6 +323,29 @@ virCgroupV2GetAnyController(virCgroupPtr group) } +static int +virCgroupV2PathOfController(virCgroupPtr group, + int controller, + const char *key, + char **path) +{ + if (!virCgroupV2HasController(group, controller)) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("v2 controller '%s' is not available"), + virCgroupV2ControllerTypeToString(controller)); + return -1; + } + + if (virAsprintf(path, "%s%s/%s", + group->unified.mountPoint, + group->unified.placement, + key ? key : "") < 0) + return -1; + + return 0; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -337,6 +360,7 @@ virCgroupBackend virCgroupV2Backend = { .detectControllers = virCgroupV2DetectControllers, .hasController = virCgroupV2HasController, .getAnyController = virCgroupV2GetAnyController, + .pathOfController = virCgroupV2PathOfController, }; -- 2.17.1

In cgroups v2 we need to handle threads and processes differently. If you need to move a process you need to write its pid into cgrou.procs file and it will move the process with all its threads as well. The whole process will be moved if you use tid of any thread. In order to move only threads at first we need to create threaded group and after that we can write the relevant thread tids into cgroup.threads file. Threads can be moved only into cgroups that are children of cgroup of its process. Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index dcfc735a84..3e8238ba1d 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -440,6 +440,20 @@ virCgroupV2Remove(virCgroupPtr group) } +static int +virCgroupV2AddTask(virCgroupPtr group, + pid_t pid, + unsigned int flags) +{ + int controller = virCgroupV2GetAnyController(group); + + if (flags & VIR_CGROUP_TASK_THREAD) + return virCgroupSetValueI64(group, controller, "cgroup.threads", pid); + else + return virCgroupSetValueI64(group, controller, "cgroup.procs", pid); +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -457,6 +471,7 @@ virCgroupBackend virCgroupV2Backend = { .pathOfController = virCgroupV2PathOfController, .makeGroup = virCgroupV2MakeGroup, .remove = virCgroupV2Remove, + .addTask = virCgroupV2AddTask, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 34 ++++++++++++++++++++++++++++++++++ 1 file changed, 34 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index d20221cf9d..2e296b738a 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -21,6 +21,7 @@ #ifdef __linux__ # include <mntent.h> +# include <sys/mount.h> #endif /* __linux__ */ #include "internal.h" @@ -470,6 +471,38 @@ virCgroupV2HasEmptyTasks(virCgroupPtr cgroup, } +static int +virCgroupV2BindMount(virCgroupPtr group, + const char *oldroot, + const char *mountopts) +{ + VIR_AUTOFREE(char *) opts = NULL; + VIR_AUTOFREE(char *) src = NULL; + + VIR_DEBUG("Mounting cgroups at '%s'", group->unified.mountPoint); + + if (virFileMakePath(group->unified.mountPoint) < 0) { + virReportSystemError(errno, _("Unable to create directory %s"), + group->unified.mountPoint); + return -1; + } + + if (virAsprintf(&opts, "mode=755,size=65536%s", mountopts) < 0) + return -1; + + if (virAsprintf(&src, "%s%s", oldroot, group->unified.mountPoint) < 0) + return -1; + + if (mount(src, group->unified.mountPoint, "none", MS_BIND, NULL) < 0) { + virReportSystemError(errno, _("Failed to bind cgroup '%s' on '%s'"), + src, group->unified.mountPoint); + return -1; + } + + return 0; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -489,6 +522,7 @@ virCgroupBackend virCgroupV2Backend = { .remove = virCgroupV2Remove, .addTask = virCgroupV2AddTask, .hasEmptyTasks = virCgroupV2HasEmptyTasks, + .bindMount = virCgroupV2BindMount, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 49 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 49 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index 976e5e449b..d10d691858 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -529,6 +529,52 @@ virCgroupV2SetOwner(virCgroupPtr cgroup, } +static int +virCgroupV2SetBlkioWeight(virCgroupPtr group, + unsigned int weight) +{ + VIR_AUTOFREE(char *) value = NULL; + + if (virAsprintf(&value, "default %u", weight) < 0) + return -1; + + return virCgroupSetValueStr(group, + VIR_CGROUP_CONTROLLER_BLKIO, + "io.weight", + value); +} + + +static int +virCgroupV2GetBlkioWeight(virCgroupPtr group, + unsigned int *weight) +{ + VIR_AUTOFREE(char *) value = NULL; + char *tmp; + + if (virCgroupGetValueStr(group, VIR_CGROUP_CONTROLLER_BLKIO, + "io.weight", &value) < 0) { + return -1; + } + + if (!(tmp = strstr(value, "default "))) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("Cannot find default io weight.")); + return -1; + } + tmp += strlen("default "); + + if (virStrToLong_ui(tmp, NULL, 10, weight) < 0) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Unable to parse '%s' as an integer"), + tmp); + return -1; + } + + return 0; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -550,6 +596,9 @@ virCgroupBackend virCgroupV2Backend = { .hasEmptyTasks = virCgroupV2HasEmptyTasks, .bindMount = virCgroupV2BindMount, .setOwner = virCgroupV2SetOwner, + + .setBlkioWeight = virCgroupV2SetBlkioWeight, + .getBlkioWeight = virCgroupV2GetBlkioWeight, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 70 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 70 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index 83e2a1352f..a6f8799311 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -751,6 +751,74 @@ virCgroupV2GetBlkioDeviceWeight(virCgroupPtr group, } +static int +virCgroupV2SetBlkioDeviceReadIops(virCgroupPtr group, + const char *path, + unsigned int riops) +{ + VIR_AUTOFREE(char *) str = NULL; + VIR_AUTOFREE(char *) blkstr = NULL; + + if (!(blkstr = virCgroupGetBlockDevString(path))) + return -1; + + if (riops == 0) { + if (virAsprintf(&str, "%sriops=max", blkstr) < 0) + return -1; + } else { + if (virAsprintf(&str, "%sriops=%u", blkstr, riops) < 0) + return -1; + } + + return virCgroupSetValueStr(group, + VIR_CGROUP_CONTROLLER_BLKIO, + "io.max", + str); +} + + +static int +virCgroupV2GetBlkioDeviceReadIops(virCgroupPtr group, + const char *path, + unsigned int *riops) +{ + VIR_AUTOFREE(char *) str = NULL; + const char *name = "riops="; + char *tmp; + + if (virCgroupGetValueForBlkDev(group, + VIR_CGROUP_CONTROLLER_BLKIO, + "io.max", + path, + &str) < 0) { + return -1; + } + + if (!str) { + *riops = 0; + } else { + if (!(tmp = strstr(str, name))) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Unable to find '%s' limit for block device '%s'"), + name, path); + return -1; + } + tmp += strlen(name); + + if (STREQLEN(tmp, "max", 3)) { + *riops = 0; + } else if (virStrToLong_ui(tmp, NULL, 10, riops) < 0) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Unable to parse '%s' as an integer"), + str); + return -1; + } + } + + return 0; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -779,6 +847,8 @@ virCgroupBackend virCgroupV2Backend = { .getBlkioIoDeviceServiced = virCgroupV2GetBlkioIoDeviceServiced, .setBlkioDeviceWeight = virCgroupV2SetBlkioDeviceWeight, .getBlkioDeviceWeight = virCgroupV2GetBlkioDeviceWeight, + .setBlkioDeviceReadIops = virCgroupV2SetBlkioDeviceReadIops, + .getBlkioDeviceReadIops = virCgroupV2GetBlkioDeviceReadIops, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 70 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 70 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index 3017a9457e..6d54096882 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -887,6 +887,74 @@ virCgroupV2GetBlkioDeviceWriteIops(virCgroupPtr group, } +static int +virCgroupV2SetBlkioDeviceReadBps(virCgroupPtr group, + const char *path, + unsigned long long rbps) +{ + VIR_AUTOFREE(char *) str = NULL; + VIR_AUTOFREE(char *) blkstr = NULL; + + if (!(blkstr = virCgroupGetBlockDevString(path))) + return -1; + + if (rbps == 0) { + if (virAsprintf(&str, "%srbps=max", blkstr) < 0) + return -1; + } else { + if (virAsprintf(&str, "%srbps=%llu", blkstr, rbps) < 0) + return -1; + } + + return virCgroupSetValueStr(group, + VIR_CGROUP_CONTROLLER_BLKIO, + "io.max", + str); +} + + +static int +virCgroupV2GetBlkioDeviceReadBps(virCgroupPtr group, + const char *path, + unsigned long long *rbps) +{ + VIR_AUTOFREE(char *) str = NULL; + const char *name = "rbps="; + char *tmp; + + if (virCgroupGetValueForBlkDev(group, + VIR_CGROUP_CONTROLLER_BLKIO, + "io.max", + path, + &str) < 0) { + return -1; + } + + if (!str) { + *rbps = 0; + } else { + if (!(tmp = strstr(str, name))) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Unable to find '%s' limit for block device '%s'"), + name, path); + return -1; + } + tmp += strlen(name); + + if (STREQLEN(tmp, "max", 3)) { + *rbps = 0; + } else if (virStrToLong_ull(tmp, NULL, 10, rbps) < 0) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Unable to parse '%s' as an integer"), + str); + return -1; + } + } + + return 0; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -919,6 +987,8 @@ virCgroupBackend virCgroupV2Backend = { .getBlkioDeviceReadIops = virCgroupV2GetBlkioDeviceReadIops, .setBlkioDeviceWriteIops = virCgroupV2SetBlkioDeviceWriteIops, .getBlkioDeviceWriteIops = virCgroupV2GetBlkioDeviceWriteIops, + .setBlkioDeviceReadBps = virCgroupV2SetBlkioDeviceReadBps, + .getBlkioDeviceReadBps = virCgroupV2GetBlkioDeviceReadBps, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 29 +++++++++++++++++++++++++++++ 1 file changed, 29 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index cf0cac1a5b..2255bbbef5 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -1023,6 +1023,33 @@ virCgroupV2GetBlkioDeviceWriteBps(virCgroupPtr group, } +static int +virCgroupV2SetMemory(virCgroupPtr group, + unsigned long long kb) +{ + unsigned long long maxkb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED; + + if (kb > maxkb) { + virReportError(VIR_ERR_INVALID_ARG, + _("Memory '%llu' must be less than %llu"), + kb, maxkb); + return -1; + } + + if (kb == maxkb) { + return virCgroupSetValueStr(group, + VIR_CGROUP_CONTROLLER_MEMORY, + "memory.max", + "max"); + } else { + return virCgroupSetValueU64(group, + VIR_CGROUP_CONTROLLER_MEMORY, + "memory.max", + kb << 10); + } +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -1059,6 +1086,8 @@ virCgroupBackend virCgroupV2Backend = { .getBlkioDeviceReadBps = virCgroupV2GetBlkioDeviceReadBps, .setBlkioDeviceWriteBps = virCgroupV2SetBlkioDeviceWriteBps, .getBlkioDeviceWriteBps = virCgroupV2GetBlkioDeviceWriteBps, + + .setMemory = virCgroupV2SetMemory, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index fcc42e3df1..51e08fe74b 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -1124,6 +1124,20 @@ virCgroupV2GetMemoryStat(virCgroupPtr group, } +static int +virCgroupV2GetMemoryUsage(virCgroupPtr group, + unsigned long *kb) +{ + unsigned long long usage_in_bytes; + int ret = virCgroupGetValueU64(group, + VIR_CGROUP_CONTROLLER_MEMORY, + "memory.current", &usage_in_bytes); + if (ret == 0) + *kb = (unsigned long) usage_in_bytes >> 10; + return ret; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -1163,6 +1177,7 @@ virCgroupBackend virCgroupV2Backend = { .setMemory = virCgroupV2SetMemory, .getMemoryStat = virCgroupV2GetMemoryStat, + .getMemoryUsage = virCgroupV2GetMemoryUsage, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 61 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 61 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index 0904eaa548..3044bb192a 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -1179,6 +1179,65 @@ virCgroupV2GetMemoryHardLimit(virCgroupPtr group, } +static int +virCgroupV2SetMemorySoftLimit(virCgroupPtr group, + unsigned long long kb) +{ + unsigned long long maxkb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED; + + if (kb > maxkb) { + virReportError(VIR_ERR_INVALID_ARG, + _("Memory '%llu' must be less than %llu"), + kb, maxkb); + return -1; + } + + if (kb == maxkb) { + return virCgroupSetValueStr(group, + VIR_CGROUP_CONTROLLER_MEMORY, + "memory.high", + "max"); + } else { + return virCgroupSetValueU64(group, + VIR_CGROUP_CONTROLLER_MEMORY, + "memory.high", + kb << 10); + } +} + + +static int +virCgroupV2GetMemorySoftLimit(virCgroupPtr group, + unsigned long long *kb) +{ + VIR_AUTOFREE(char *) value = NULL; + unsigned long long high; + + if (virCgroupGetValueStr(group, + VIR_CGROUP_CONTROLLER_MEMORY, + "memory.high", &value) < 0) + return -1; + + if (STREQ(value, "max")) { + *kb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED; + return 0; + } + + if (virStrToLong_ull(value, NULL, 10, &high) < 0) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Failed to parse value '%s' as number."), + value); + return -1; + } + + *kb = high >> 10; + if (*kb >= VIR_DOMAIN_MEMORY_PARAM_UNLIMITED) + *kb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED; + + return 0; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -1221,6 +1280,8 @@ virCgroupBackend virCgroupV2Backend = { .getMemoryUsage = virCgroupV2GetMemoryUsage, .setMemoryHardLimit = virCgroupV2SetMemoryHardLimit, .getMemoryHardLimit = virCgroupV2GetMemoryHardLimit, + .setMemorySoftLimit = virCgroupV2SetMemorySoftLimit, + .getMemorySoftLimit = virCgroupV2GetMemorySoftLimit, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index fcadf9381d..991d32f4b7 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -1298,6 +1298,21 @@ virCgroupV2GetMemSwapHardLimit(virCgroupPtr group, } +static int +virCgroupV2GetMemSwapUsage(virCgroupPtr group, + unsigned long long *kb) +{ + unsigned long long usage_in_bytes; + int ret; + ret = virCgroupGetValueU64(group, + VIR_CGROUP_CONTROLLER_MEMORY, + "memory.swap.current", &usage_in_bytes); + if (ret == 0) + *kb = (unsigned long) usage_in_bytes >> 10; + return ret; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -1344,6 +1359,7 @@ virCgroupBackend virCgroupV2Backend = { .getMemorySoftLimit = virCgroupV2GetMemorySoftLimit, .setMemSwapHardLimit = virCgroupV2SetMemSwapHardLimit, .getMemSwapHardLimit = virCgroupV2GetMemSwapHardLimit, + .getMemSwapUsage = virCgroupV2GetMemSwapUsage, }; -- 2.17.1

In order to set CPU cfs period using cgroup v2 'cpu.max' interface we need to load the current value of CPU cfs quota first because format of 'cpu.max' interface is '$quota $period' and in order to change 'period' we need to write 'quota' as well. Writing only one number changes only 'quota'. Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 68 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 68 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index 986b2d07c1..24c5bb3e3d 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -1333,6 +1333,72 @@ virCgroupV2GetCpuShares(virCgroupPtr group, } +static int +virCgroupV2SetCpuCfsPeriod(virCgroupPtr group, + unsigned long long cfs_period) +{ + VIR_AUTOFREE(char *) value = NULL; + VIR_AUTOFREE(char *) str = NULL; + char *tmp; + + /* The cfs_period should be greater or equal than 1ms, and less or equal + * than 1s. + */ + if (cfs_period < 1000 || cfs_period > 1000000) { + virReportError(VIR_ERR_INVALID_ARG, + _("cfs_period '%llu' must be in range (1000, 1000000)"), + cfs_period); + return -1; + } + + if (virCgroupGetValueStr(group, VIR_CGROUP_CONTROLLER_CPU, + "cpu.max", &str) < 0) { + return -1; + } + + if (!(tmp = strchr(str, ' '))) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("Invalid 'cpu.max' data.")); + return -1; + } + *tmp = '\n'; + + if (virAsprintf(&value, "%s %llu", str, cfs_period) < 0) + return -1; + + return virCgroupSetValueStr(group, VIR_CGROUP_CONTROLLER_CPU, + "cpu.max", value); +} + + +static int +virCgroupV2GetCpuCfsPeriod(virCgroupPtr group, + unsigned long long *cfs_period) +{ + VIR_AUTOFREE(char *) str = NULL; + char *tmp; + + if (virCgroupGetValueStr(group, VIR_CGROUP_CONTROLLER_CPU, + "cpu.max", &str) < 0) { + return -1; + } + + if (!(tmp = strchr(str, ' '))) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("Invalid 'cpu.max' data.")); + return -1; + } + + if (virStrToLong_ull(tmp, NULL, 10, cfs_period) < 0) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Failed to parse value '%s' from cpu.max."), str); + return -1; + } + + return 0; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -1383,6 +1449,8 @@ virCgroupBackend virCgroupV2Backend = { .setCpuShares = virCgroupV2SetCpuShares, .getCpuShares = virCgroupV2GetCpuShares, + .setCpuCfsPeriod = virCgroupV2SetCpuCfsPeriod, + .getCpuCfsPeriod = virCgroupV2GetCpuCfsPeriod, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index 26fdb46098..45af617971 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -1449,6 +1449,21 @@ virCgroupV2GetCpuCfsQuota(virCgroupPtr group, } +static bool +virCgroupV2SupportsCpuBW(virCgroupPtr cgroup) +{ + VIR_AUTOFREE(char *) path = NULL; + + if (virCgroupV2PathOfController(cgroup, VIR_CGROUP_CONTROLLER_CPU, + "cpu.max", &path) < 0) { + virResetLastError(); + return false; + } + + return virFileExists(path); +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -1503,6 +1518,7 @@ virCgroupBackend virCgroupV2Backend = { .getCpuCfsPeriod = virCgroupV2GetCpuCfsPeriod, .setCpuCfsQuota = virCgroupV2SetCpuCfsQuota, .getCpuCfsQuota = virCgroupV2GetCpuCfsQuota, + .supportsCpuBW = virCgroupV2SupportsCpuBW, }; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroupv2.c | 49 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 49 insertions(+) diff --git a/src/util/vircgroupv2.c b/src/util/vircgroupv2.c index 93c1e39119..e02b438d95 100644 --- a/src/util/vircgroupv2.c +++ b/src/util/vircgroupv2.c @@ -1495,6 +1495,54 @@ virCgroupV2GetCpuacctUsage(virCgroupPtr group, } +static int +virCgroupV2GetCpuacctStat(virCgroupPtr group, + unsigned long long *user, + unsigned long long *sys) +{ + VIR_AUTOFREE(char *) str = NULL; + char *tmp; + unsigned long long userVal = 0; + unsigned long long sysVal = 0; + + if (virCgroupGetValueStr(group, VIR_CGROUP_CONTROLLER_CPUACCT, + "cpu.stat", &str) < 0) { + return -1; + } + + if (!(tmp = strstr(str, "user_usec "))) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("cannot parse cpu user stat '%s'"), str); + return -1; + } + tmp += strlen("user_usec "); + + if (virStrToLong_ull(tmp, &tmp, 10, &userVal) < 0) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Failed to parse value '%s' as number."), tmp); + return -1; + } + + if (!(tmp = strstr(str, "system_usec "))) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("cannot parse cpu sys stat '%s'"), str); + return -1; + } + tmp += strlen("system_usec "); + + if (virStrToLong_ull(tmp, &tmp, 10, &sysVal) < 0) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Failed to parse value '%s' as number."), tmp); + return -1; + } + + *user = userVal * 1000; + *sys = sysVal * 1000; + + return 0; +} + + virCgroupBackend virCgroupV2Backend = { .type = VIR_CGROUP_BACKEND_TYPE_V2, @@ -1552,6 +1600,7 @@ virCgroupBackend virCgroupV2Backend = { .supportsCpuBW = virCgroupV2SupportsCpuBW, .getCpuacctUsage = virCgroupV2GetCpuacctUsage, + .getCpuacctStat = virCgroupV2GetCpuacctStat, }; -- 2.17.1

This enables to use both cgroup v1 and v2 at the same time together with libvirt. It is supported by kernel and there is valid use-case, not all controllers are implemented in cgroup v2 so there might be configurations where administrator would enable these missing controllers in cgroup v1. Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- src/util/vircgroup.c | 351 ++++++++++++++++++++++++++---------- src/util/vircgroupbackend.c | 20 ++ src/util/vircgroupbackend.h | 16 +- src/util/vircgrouppriv.h | 2 +- 4 files changed, 291 insertions(+), 98 deletions(-) diff --git a/src/util/vircgroup.c b/src/util/vircgroup.c index dc249bfe33..4aec5f1bcf 100644 --- a/src/util/vircgroup.c +++ b/src/util/vircgroup.c @@ -228,6 +228,7 @@ virCgroupDetectMounts(virCgroupPtr group) struct mntent entry; char buf[CGROUP_MAX_VAL]; int ret = -1; + size_t i; mounts = fopen("/proc/mounts", "r"); if (mounts == NULL) { @@ -236,11 +237,14 @@ virCgroupDetectMounts(virCgroupPtr group) } while (getmntent_r(mounts, &entry, buf, sizeof(buf)) != NULL) { - if (group->backend->detectMounts(group, - entry.mnt_type, - entry.mnt_opts, - entry.mnt_dir) < 0) { - goto cleanup; + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (group->backends[i] && + group->backends[i]->detectMounts(group, + entry.mnt_type, + entry.mnt_opts, + entry.mnt_dir) < 0) { + goto cleanup; + } } } @@ -303,6 +307,7 @@ virCgroupDetectPlacement(virCgroupPtr group, } while (fgets(line, sizeof(line), mapping) != NULL) { + size_t i; char *controllers = strchr(line, ':'); char *selfpath = controllers ? strchr(controllers + 1, ':') : NULL; char *nl = selfpath ? strchr(selfpath, '\n') : NULL; @@ -317,9 +322,12 @@ virCgroupDetectPlacement(virCgroupPtr group, controllers++; selfpath++; - if (group->backend->detectPlacement(group, path, controllers, - selfpath) < 0) { - goto cleanup; + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (group->backends[i] && + group->backends[i]->detectPlacement(group, path, controllers, + selfpath) < 0) { + goto cleanup; + } } } @@ -338,8 +346,9 @@ virCgroupDetect(virCgroupPtr group, const char *path, virCgroupPtr parent) { - int rc; size_t i; + bool backendAvailable = false; + int controllersAvailable = 0; virCgroupBackendPtr *backends = virCgroupBackendGetAll(); VIR_DEBUG("group=%p controllers=%d path=%s parent=%p", @@ -350,31 +359,40 @@ virCgroupDetect(virCgroupPtr group, for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { if (backends[i] && backends[i]->available()) { - group->backend = backends[i]; - break; + group->backends[i] = backends[i]; + backendAvailable = true; } } - if (!group->backend) { + if (!backendAvailable) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("no cgroup backend available")); return -1; } if (parent) { - if (group->backend->copyMounts(group, parent) < 0) - return -1; + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (group->backends[i] && + group->backends[i]->copyMounts(group, parent) < 0) { + return -1; + } + } } else { if (virCgroupDetectMounts(group) < 0) return -1; } - rc = group->backend->detectControllers(group, controllers); - if (rc < 0) - return -1; + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (group->backends[i]) { + int rc = group->backends[i]->detectControllers(group, controllers); + if (rc < 0) + return -1; + controllersAvailable |= rc; + } + } /* Check that at least 1 controller is available */ - if (rc == 0) { + if (controllersAvailable == 0) { virReportSystemError(ENXIO, "%s", _("At least one cgroup controller is required")); return -1; @@ -383,17 +401,26 @@ virCgroupDetect(virCgroupPtr group, /* In some cases we can copy part of the placement info * based on the parent cgroup... */ - if ((parent || path[0] == '/') && - group->backend->copyPlacement(group, path, parent) < 0) - return -1; + if (parent || path[0] == '/') { + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (group->backends[i] && + group->backends[i]->copyPlacement(group, path, parent) < 0) { + return -1; + } + } + } /* ... but use /proc/cgroups to fill in the rest */ if (virCgroupDetectPlacement(group, pid, path) < 0) return -1; /* Check that for every mounted controller, we found our placement */ - if (group->backend->validatePlacement(group, pid) < 0) - return -1; + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (group->backends[i] && + group->backends[i]->validatePlacement(group, pid) < 0) { + return -1; + } + } return 0; } @@ -599,9 +626,14 @@ virCgroupMakeGroup(virCgroupPtr parent, bool create, unsigned int flags) { - if (group->backend->makeGroup(parent, group, create, flags) < 0) { - virCgroupRemove(group); - return -1; + size_t i; + + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (group->backends[i] && + group->backends[i]->makeGroup(parent, group, create, flags) < 0) { + virCgroupRemove(group); + return -1; + } } return 0; @@ -662,6 +694,24 @@ virCgroupNew(pid_t pid, } +static int +virCgroupAddTaskInternal(virCgroupPtr group, + pid_t pid, + unsigned int flags) +{ + size_t i; + + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (group->backends[i] && + group->backends[i]->addTask(group, pid, flags) < 0) { + return -1; + } + } + + return 0; +} + + /** * virCgroupAddProcess: * @@ -676,7 +726,7 @@ virCgroupNew(pid_t pid, int virCgroupAddProcess(virCgroupPtr group, pid_t pid) { - return group->backend->addTask(group, pid, VIR_CGROUP_TASK_PROCESS); + return virCgroupAddTaskInternal(group, pid, VIR_CGROUP_TASK_PROCESS); } /** @@ -693,9 +743,9 @@ virCgroupAddProcess(virCgroupPtr group, pid_t pid) int virCgroupAddMachineProcess(virCgroupPtr group, pid_t pid) { - return group->backend->addTask(group, pid, - VIR_CGROUP_TASK_PROCESS | - VIR_CGROUP_TASK_SYSTEMD); + return virCgroupAddTaskInternal(group, pid, + VIR_CGROUP_TASK_PROCESS | + VIR_CGROUP_TASK_SYSTEMD); } /** @@ -713,7 +763,7 @@ int virCgroupAddThread(virCgroupPtr group, pid_t pid) { - return group->backend->addTask(group, pid, VIR_CGROUP_TASK_THREAD); + return virCgroupAddTaskInternal(group, pid, VIR_CGROUP_TASK_THREAD); } @@ -967,17 +1017,24 @@ virCgroupNewDetectMachine(const char *name, char *machinename, virCgroupPtr *group) { + size_t i; + if (virCgroupNewDetect(pid, controllers, group) < 0) { if (virCgroupNewIgnoreError()) return 0; return -1; } - if (!(*group)->backend->validateMachineGroup(*group, name, drivername, machinename)) { - VIR_DEBUG("Failed to validate machine name for '%s' driver '%s'", - name, drivername); - virCgroupFree(group); - return 0; + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if ((*group)->backends[i] && + !(*group)->backends[i]->validateMachineGroup(*group, name, + drivername, + machinename)) { + VIR_DEBUG("Failed to validate machine name for '%s' driver '%s'", + name, drivername); + virCgroupFree(group); + return 0; + } } return 0; @@ -1055,6 +1112,7 @@ virCgroupNewMachineSystemd(const char *name, int rv; virCgroupPtr init; VIR_AUTOFREE(char *) path = NULL; + size_t i; VIR_DEBUG("Trying to setup machine '%s' via systemd", name); if ((rv = virSystemdCreateMachine(name, @@ -1077,7 +1135,12 @@ virCgroupNewMachineSystemd(const char *name, &init) < 0) return -1; - path = init->backend->stealPlacement(init); + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (init->backends[i] && + (path = init->backends[i]->stealPlacement(init))) { + break; + } + } virCgroupFree(&init); if (!path || STREQ(path, "/") || path[0] != '/') { @@ -1256,12 +1319,21 @@ virCgroupFree(virCgroupPtr *group) bool virCgroupHasController(virCgroupPtr cgroup, int controller) { + size_t i; + if (!cgroup) return false; if (controller < 0 || controller >= VIR_CGROUP_CONTROLLER_LAST) return false; - return cgroup->backend->hasController(cgroup, controller); + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (cgroup->backends[i] && + cgroup->backends[i]->hasController(cgroup, controller)) { + return true; + } + } + + return false; } @@ -1277,7 +1349,8 @@ virCgroupPathOfController(virCgroupPtr group, return -1; } - return group->backend->pathOfController(group, controller, key, path); + VIR_CGROUP_BACKEND_CALL(group, controller, pathOfController, -1, + controller, key, path); } @@ -1299,7 +1372,8 @@ virCgroupGetBlkioIoServiced(virCgroupPtr group, long long *requests_read, long long *requests_write) { - VIR_CGROUP_BACKEND_CALL(group, getBlkioIoServiced, -1, + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + getBlkioIoServiced, -1, bytes_read, bytes_write, requests_read, requests_write); } @@ -1325,7 +1399,8 @@ virCgroupGetBlkioIoDeviceServiced(virCgroupPtr group, long long *requests_read, long long *requests_write) { - VIR_CGROUP_BACKEND_CALL(group, getBlkioIoDeviceServiced, -1, + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + getBlkioIoDeviceServiced, -1, path, bytes_read, bytes_write, requests_read, requests_write); } @@ -1342,7 +1417,8 @@ virCgroupGetBlkioIoDeviceServiced(virCgroupPtr group, int virCgroupSetBlkioWeight(virCgroupPtr group, unsigned int weight) { - VIR_CGROUP_BACKEND_CALL(group, setBlkioWeight, -1, weight); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + setBlkioWeight, -1, weight); } @@ -1357,7 +1433,8 @@ virCgroupSetBlkioWeight(virCgroupPtr group, unsigned int weight) int virCgroupGetBlkioWeight(virCgroupPtr group, unsigned int *weight) { - VIR_CGROUP_BACKEND_CALL(group, getBlkioWeight, -1, weight); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + getBlkioWeight, -1, weight); } /** @@ -1373,7 +1450,8 @@ virCgroupSetBlkioDeviceReadIops(virCgroupPtr group, const char *path, unsigned int riops) { - VIR_CGROUP_BACKEND_CALL(group, setBlkioDeviceReadIops, -1, path, riops); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + setBlkioDeviceReadIops, -1, path, riops); } @@ -1390,7 +1468,8 @@ virCgroupSetBlkioDeviceWriteIops(virCgroupPtr group, const char *path, unsigned int wiops) { - VIR_CGROUP_BACKEND_CALL(group, setBlkioDeviceWriteIops, -1, path, wiops); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + setBlkioDeviceWriteIops, -1, path, wiops); } @@ -1407,7 +1486,8 @@ virCgroupSetBlkioDeviceReadBps(virCgroupPtr group, const char *path, unsigned long long rbps) { - VIR_CGROUP_BACKEND_CALL(group, setBlkioDeviceReadBps, -1, path, rbps); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + setBlkioDeviceReadBps, -1, path, rbps); } /** @@ -1423,7 +1503,8 @@ virCgroupSetBlkioDeviceWriteBps(virCgroupPtr group, const char *path, unsigned long long wbps) { - VIR_CGROUP_BACKEND_CALL(group, setBlkioDeviceWriteBps, -1, path, wbps); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + setBlkioDeviceWriteBps, -1, path, wbps); } @@ -1441,7 +1522,8 @@ virCgroupSetBlkioDeviceWeight(virCgroupPtr group, const char *path, unsigned int weight) { - VIR_CGROUP_BACKEND_CALL(group, setBlkioDeviceWeight, -1, path, weight); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + setBlkioDeviceWeight, -1, path, weight); } /** @@ -1457,7 +1539,8 @@ virCgroupGetBlkioDeviceReadIops(virCgroupPtr group, const char *path, unsigned int *riops) { - VIR_CGROUP_BACKEND_CALL(group, getBlkioDeviceReadIops, -1, path, riops); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + getBlkioDeviceReadIops, -1, path, riops); } /** @@ -1473,7 +1556,8 @@ virCgroupGetBlkioDeviceWriteIops(virCgroupPtr group, const char *path, unsigned int *wiops) { - VIR_CGROUP_BACKEND_CALL(group, getBlkioDeviceWriteIops, -1, path, wiops); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + getBlkioDeviceWriteIops, -1, path, wiops); } /** @@ -1489,7 +1573,8 @@ virCgroupGetBlkioDeviceReadBps(virCgroupPtr group, const char *path, unsigned long long *rbps) { - VIR_CGROUP_BACKEND_CALL(group, getBlkioDeviceReadBps, -1, path, rbps); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + getBlkioDeviceReadBps, -1, path, rbps); } /** @@ -1505,7 +1590,8 @@ virCgroupGetBlkioDeviceWriteBps(virCgroupPtr group, const char *path, unsigned long long *wbps) { - VIR_CGROUP_BACKEND_CALL(group, getBlkioDeviceWriteBps, -1, path, wbps); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + getBlkioDeviceWriteBps, -1, path, wbps); } /** @@ -1521,7 +1607,8 @@ virCgroupGetBlkioDeviceWeight(virCgroupPtr group, const char *path, unsigned int *weight) { - VIR_CGROUP_BACKEND_CALL(group, getBlkioDeviceWeight, -1, path, weight); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO, + getBlkioDeviceWeight, -1, path, weight); } @@ -1536,7 +1623,8 @@ virCgroupGetBlkioDeviceWeight(virCgroupPtr group, int virCgroupSetMemory(virCgroupPtr group, unsigned long long kb) { - VIR_CGROUP_BACKEND_CALL(group, setMemory, -1, kb); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY, + setMemory, -1, kb); } @@ -1562,7 +1650,8 @@ virCgroupGetMemoryStat(virCgroupPtr group, unsigned long long *inactiveFile, unsigned long long *unevictable) { - VIR_CGROUP_BACKEND_CALL(group, getMemoryStat, -1, cache, + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY, + getMemoryStat, -1, cache, activeAnon, inactiveAnon, activeFile, inactiveFile, unevictable); @@ -1580,7 +1669,8 @@ virCgroupGetMemoryStat(virCgroupPtr group, int virCgroupGetMemoryUsage(virCgroupPtr group, unsigned long *kb) { - VIR_CGROUP_BACKEND_CALL(group, getMemoryUsage, -1, kb); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY, + getMemoryUsage, -1, kb); } @@ -1595,7 +1685,8 @@ virCgroupGetMemoryUsage(virCgroupPtr group, unsigned long *kb) int virCgroupSetMemoryHardLimit(virCgroupPtr group, unsigned long long kb) { - VIR_CGROUP_BACKEND_CALL(group, setMemoryHardLimit, -1, kb); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY, + setMemoryHardLimit, -1, kb); } @@ -1610,7 +1701,8 @@ virCgroupSetMemoryHardLimit(virCgroupPtr group, unsigned long long kb) int virCgroupGetMemoryHardLimit(virCgroupPtr group, unsigned long long *kb) { - VIR_CGROUP_BACKEND_CALL(group, getMemoryHardLimit, -1, kb); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY, + getMemoryHardLimit, -1, kb); } @@ -1625,7 +1717,8 @@ virCgroupGetMemoryHardLimit(virCgroupPtr group, unsigned long long *kb) int virCgroupSetMemorySoftLimit(virCgroupPtr group, unsigned long long kb) { - VIR_CGROUP_BACKEND_CALL(group, setMemorySoftLimit, -1, kb); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY, + setMemorySoftLimit, -1, kb); } @@ -1640,7 +1733,8 @@ virCgroupSetMemorySoftLimit(virCgroupPtr group, unsigned long long kb) int virCgroupGetMemorySoftLimit(virCgroupPtr group, unsigned long long *kb) { - VIR_CGROUP_BACKEND_CALL(group, getMemorySoftLimit, -1, kb); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY, + getMemorySoftLimit, -1, kb); } @@ -1655,7 +1749,8 @@ virCgroupGetMemorySoftLimit(virCgroupPtr group, unsigned long long *kb) int virCgroupSetMemSwapHardLimit(virCgroupPtr group, unsigned long long kb) { - VIR_CGROUP_BACKEND_CALL(group, setMemSwapHardLimit, -1, kb); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY, + setMemSwapHardLimit, -1, kb); } @@ -1670,7 +1765,8 @@ virCgroupSetMemSwapHardLimit(virCgroupPtr group, unsigned long long kb) int virCgroupGetMemSwapHardLimit(virCgroupPtr group, unsigned long long *kb) { - VIR_CGROUP_BACKEND_CALL(group, getMemSwapHardLimit, -1, kb); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY, + getMemSwapHardLimit, -1, kb); } @@ -1685,7 +1781,8 @@ virCgroupGetMemSwapHardLimit(virCgroupPtr group, unsigned long long *kb) int virCgroupGetMemSwapUsage(virCgroupPtr group, unsigned long long *kb) { - VIR_CGROUP_BACKEND_CALL(group, getMemSwapUsage, -1, kb); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY, + getMemSwapUsage, -1, kb); } @@ -1700,7 +1797,8 @@ virCgroupGetMemSwapUsage(virCgroupPtr group, unsigned long long *kb) int virCgroupSetCpusetMems(virCgroupPtr group, const char *mems) { - VIR_CGROUP_BACKEND_CALL(group, setCpusetMems, -1, mems); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUSET, + setCpusetMems, -1, mems); } @@ -1715,7 +1813,8 @@ virCgroupSetCpusetMems(virCgroupPtr group, const char *mems) int virCgroupGetCpusetMems(virCgroupPtr group, char **mems) { - VIR_CGROUP_BACKEND_CALL(group, getCpusetMems, -1, mems); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUSET, + getCpusetMems, -1, mems); } @@ -1730,7 +1829,8 @@ virCgroupGetCpusetMems(virCgroupPtr group, char **mems) int virCgroupSetCpusetMemoryMigrate(virCgroupPtr group, bool migrate) { - VIR_CGROUP_BACKEND_CALL(group, setCpusetMemoryMigrate, -1, migrate); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUSET, + setCpusetMemoryMigrate, -1, migrate); } @@ -1745,7 +1845,8 @@ virCgroupSetCpusetMemoryMigrate(virCgroupPtr group, bool migrate) int virCgroupGetCpusetMemoryMigrate(virCgroupPtr group, bool *migrate) { - VIR_CGROUP_BACKEND_CALL(group, getCpusetMemoryMigrate, -1, migrate); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUSET, + getCpusetMemoryMigrate, -1, migrate); } @@ -1760,7 +1861,8 @@ virCgroupGetCpusetMemoryMigrate(virCgroupPtr group, bool *migrate) int virCgroupSetCpusetCpus(virCgroupPtr group, const char *cpus) { - VIR_CGROUP_BACKEND_CALL(group, setCpusetCpus, -1, cpus); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUSET, + setCpusetCpus, -1, cpus); } @@ -1775,7 +1877,8 @@ virCgroupSetCpusetCpus(virCgroupPtr group, const char *cpus) int virCgroupGetCpusetCpus(virCgroupPtr group, char **cpus) { - VIR_CGROUP_BACKEND_CALL(group, getCpusetCpus, -1, cpus); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUSET, + getCpusetCpus, -1, cpus); } @@ -1789,7 +1892,8 @@ virCgroupGetCpusetCpus(virCgroupPtr group, char **cpus) int virCgroupDenyAllDevices(virCgroupPtr group) { - VIR_CGROUP_BACKEND_CALL(group, denyAllDevices, -1); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_DEVICES, + denyAllDevices, -1); } /** @@ -1809,7 +1913,8 @@ virCgroupDenyAllDevices(virCgroupPtr group) int virCgroupAllowAllDevices(virCgroupPtr group, int perms) { - VIR_CGROUP_BACKEND_CALL(group, allowAllDevices, -1, perms); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_DEVICES, + allowAllDevices, -1, perms); } @@ -1828,7 +1933,8 @@ int virCgroupAllowDevice(virCgroupPtr group, char type, int major, int minor, int perms) { - VIR_CGROUP_BACKEND_CALL(group, allowDevice, -1, type, major, minor, perms); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_DEVICES, + allowDevice, -1, type, major, minor, perms); } @@ -1867,7 +1973,8 @@ virCgroupAllowDevicePath(virCgroupPtr group, if (!S_ISCHR(sb.st_mode) && !S_ISBLK(sb.st_mode)) return 1; - VIR_CGROUP_BACKEND_CALL(group, allowDevice, -1, + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_DEVICES, + allowDevice, -1, S_ISCHR(sb.st_mode) ? 'c' : 'b', major(sb.st_rdev), minor(sb.st_rdev), @@ -1890,7 +1997,8 @@ int virCgroupDenyDevice(virCgroupPtr group, char type, int major, int minor, int perms) { - VIR_CGROUP_BACKEND_CALL(group, denyDevice, -1, type, major, minor, perms); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_DEVICES, + denyDevice, -1, type, major, minor, perms); } @@ -1929,7 +2037,8 @@ virCgroupDenyDevicePath(virCgroupPtr group, if (!S_ISCHR(sb.st_mode) && !S_ISBLK(sb.st_mode)) return 1; - VIR_CGROUP_BACKEND_CALL(group, denyDevice, -1, + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_DEVICES, + denyDevice, -1, S_ISCHR(sb.st_mode) ? 'c' : 'b', major(sb.st_rdev), minor(sb.st_rdev), @@ -2172,14 +2281,16 @@ virCgroupGetDomainTotalCpuStats(virCgroupPtr group, int virCgroupSetCpuShares(virCgroupPtr group, unsigned long long shares) { - VIR_CGROUP_BACKEND_CALL(group, setCpuShares, -1, shares); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPU, + setCpuShares, -1, shares); } int virCgroupGetCpuShares(virCgroupPtr group, unsigned long long *shares) { - VIR_CGROUP_BACKEND_CALL(group, getCpuShares, -1, shares); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPU, + getCpuShares, -1, shares); } @@ -2194,7 +2305,8 @@ virCgroupGetCpuShares(virCgroupPtr group, unsigned long long *shares) int virCgroupSetCpuCfsPeriod(virCgroupPtr group, unsigned long long cfs_period) { - VIR_CGROUP_BACKEND_CALL(group, setCpuCfsPeriod, -1, cfs_period); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPU, + setCpuCfsPeriod, -1, cfs_period); } @@ -2209,7 +2321,8 @@ virCgroupSetCpuCfsPeriod(virCgroupPtr group, unsigned long long cfs_period) int virCgroupGetCpuCfsPeriod(virCgroupPtr group, unsigned long long *cfs_period) { - VIR_CGROUP_BACKEND_CALL(group, getCpuCfsPeriod, -1, cfs_period); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPU, + getCpuCfsPeriod, -1, cfs_period); } @@ -2225,14 +2338,16 @@ virCgroupGetCpuCfsPeriod(virCgroupPtr group, unsigned long long *cfs_period) int virCgroupSetCpuCfsQuota(virCgroupPtr group, long long cfs_quota) { - VIR_CGROUP_BACKEND_CALL(group, setCpuCfsQuota, -1, cfs_quota); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPU, + setCpuCfsQuota, -1, cfs_quota); } int virCgroupGetCpuacctPercpuUsage(virCgroupPtr group, char **usage) { - VIR_CGROUP_BACKEND_CALL(group, getCpuacctPercpuUsage, -1, usage); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUACCT, + getCpuacctPercpuUsage, -1, usage); } @@ -2299,7 +2414,16 @@ virCgroupRemoveRecursively(char *grppath) int virCgroupRemove(virCgroupPtr group) { - return group->backend->remove(group); + size_t i; + + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (group->backends[i] && + group->backends[i]->remove(group) < 0) { + return -1; + } + } + + return 0; } @@ -2308,11 +2432,16 @@ virCgroupPathOfAnyController(virCgroupPtr group, const char *name, char **keypath) { + size_t i; int controller; - controller = group->backend->getAnyController(group); - if (controller >= 0) - return virCgroupPathOfController(group, controller, name, keypath); + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (group->backends[i]) { + controller = group->backends[i]->getAnyController(group); + if (controller >= 0) + return virCgroupPathOfController(group, controller, name, keypath); + } + } virReportSystemError(ENOSYS, "%s", _("No controllers are mounted")); @@ -2548,14 +2677,16 @@ virCgroupKillPainfully(virCgroupPtr group) int virCgroupGetCpuCfsQuota(virCgroupPtr group, long long *cfs_quota) { - VIR_CGROUP_BACKEND_CALL(group, getCpuCfsQuota, -1, cfs_quota); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPU, + getCpuCfsQuota, -1, cfs_quota); } int virCgroupGetCpuacctUsage(virCgroupPtr group, unsigned long long *usage) { - VIR_CGROUP_BACKEND_CALL(group, getCpuacctUsage, -1, usage); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUACCT, + getCpuacctUsage, -1, usage); } @@ -2563,21 +2694,24 @@ int virCgroupGetCpuacctStat(virCgroupPtr group, unsigned long long *user, unsigned long long *sys) { - VIR_CGROUP_BACKEND_CALL(group, getCpuacctStat, -1, user, sys); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUACCT, + getCpuacctStat, -1, user, sys); } int virCgroupSetFreezerState(virCgroupPtr group, const char *state) { - VIR_CGROUP_BACKEND_CALL(group, setFreezerState, -1, state); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_FREEZER, + setFreezerState, -1, state); } int virCgroupGetFreezerState(virCgroupPtr group, char **state) { - VIR_CGROUP_BACKEND_CALL(group, getFreezerState, -1, state); + VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_FREEZER, + getFreezerState, -1, state); } @@ -2585,7 +2719,16 @@ int virCgroupBindMount(virCgroupPtr group, const char *oldroot, const char *mountopts) { - return group->backend->bindMount(group, oldroot, mountopts); + size_t i; + + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (group->backends[i] && + group->backends[i]->bindMount(group, oldroot, mountopts) < 0) { + return -1; + } + } + + return 0; } @@ -2594,7 +2737,16 @@ int virCgroupSetOwner(virCgroupPtr cgroup, gid_t gid, int controllers) { - return cgroup->backend->setOwner(cgroup, uid, gid, controllers); + size_t i; + + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (cgroup->backends[i] && + cgroup->backends[i]->setOwner(cgroup, uid, gid, controllers) < 0) { + return -1; + } + } + + return 0; } @@ -2608,13 +2760,24 @@ int virCgroupSetOwner(virCgroupPtr cgroup, bool virCgroupSupportsCpuBW(virCgroupPtr cgroup) { - VIR_CGROUP_BACKEND_CALL(cgroup, supportsCpuBW, false); + VIR_CGROUP_BACKEND_CALL(cgroup, VIR_CGROUP_CONTROLLER_CPU, + supportsCpuBW, false); } int virCgroupHasEmptyTasks(virCgroupPtr cgroup, int controller) { - return cgroup->backend->hasEmptyTasks(cgroup, controller); + size_t i; + + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (cgroup->backends[i]) { + int rc = cgroup->backends[i]->hasEmptyTasks(cgroup, controller); + if (rc <= 0) + return rc; + } + } + + return 1; } bool diff --git a/src/util/vircgroupbackend.c b/src/util/vircgroupbackend.c index 7ee39ac8ca..2e90781dc3 100644 --- a/src/util/vircgroupbackend.c +++ b/src/util/vircgroupbackend.c @@ -20,6 +20,9 @@ #include <config.h> #include "vircgroupbackend.h" +#define __VIR_CGROUP_ALLOW_INCLUDE_PRIV_H__ +#include "vircgrouppriv.h" +#undef __VIR_CGROUP_ALLOW_INCLUDE_PRIV_H__ #include "vircgroupv1.h" #include "vircgroupv2.h" #include "virerror.h" @@ -67,3 +70,20 @@ virCgroupBackendGetAll(void) } return virCgroupBackends; } + + +virCgroupBackendPtr +virCgroupBackendForController(virCgroupPtr group, + unsigned int controller) +{ + size_t i; + + for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) { + if (group->backends[i] && + group->backends[i]->hasController(group, controller)) { + return group->backends[i]; + } + } + + return NULL; +} diff --git a/src/util/vircgroupbackend.h b/src/util/vircgroupbackend.h index 86d1539e07..bc60b44643 100644 --- a/src/util/vircgroupbackend.h +++ b/src/util/vircgroupbackend.h @@ -436,12 +436,22 @@ virCgroupBackendRegister(virCgroupBackendPtr backend); virCgroupBackendPtr * virCgroupBackendGetAll(void); -# define VIR_CGROUP_BACKEND_CALL(group, func, ret, ...) \ - if (!group->backend->func) { \ +virCgroupBackendPtr +virCgroupBackendForController(virCgroupPtr group, + unsigned int controller); + +# define VIR_CGROUP_BACKEND_CALL(group, controller, func, ret, ...) \ + virCgroupBackendPtr backend = virCgroupBackendForController(group, controller); \ + if (!backend) { \ + virReportError(VIR_ERR_INTERNAL_ERROR, \ + _("failed to get cgroup backend for '%s'"), #func); \ + return ret; \ + } \ + if (!backend->func) { \ virReportError(VIR_ERR_OPERATION_UNSUPPORTED, \ _("operation '%s' not supported"), #func); \ return ret; \ } \ - return group->backend->func(group, ##__VA_ARGS__); + return backend->func(group, ##__VA_ARGS__); #endif /* __VIR_CGROUP_BACKEND_H__ */ diff --git a/src/util/vircgrouppriv.h b/src/util/vircgrouppriv.h index 4a0d75ddbc..8f24b0891e 100644 --- a/src/util/vircgrouppriv.h +++ b/src/util/vircgrouppriv.h @@ -56,7 +56,7 @@ typedef virCgroupV2Controller *virCgroupV2ControllerPtr; struct _virCgroup { char *path; - virCgroupBackendPtr backend; + virCgroupBackendPtr backends[VIR_CGROUP_BACKEND_TYPE_LAST]; virCgroupV1Controller legacy[VIR_CGROUP_CONTROLLER_LAST]; virCgroupV2Controller unified; -- 2.17.1

We need to create the cgroup v2 sysfs the same way as we do for cgroup v1. This introduces new VIR_CGROUP_MOCK_MODE env variable which will configure which cgroup mode each test requires. There are three different modes: - legacy: only cgroup v1 is available and it's the default mode - hybrid: both cgroup v1 and cgroup v2 are available and have some controllers - unified: only cgroup v2 is available Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- tests/vircgroupmock.c | 161 ++++++++++++++++++++++++++++++++++++++---- 1 file changed, 146 insertions(+), 15 deletions(-) diff --git a/tests/vircgroupmock.c b/tests/vircgroupmock.c index 5c48ef012a..b0911d4df2 100644 --- a/tests/vircgroupmock.c +++ b/tests/vircgroupmock.c @@ -105,7 +105,8 @@ static int make_file(const char *path, return ret; } -static int make_controller(const char *path, mode_t mode) + +static int make_controller_v1(const char *path, mode_t mode) { int ret = -1; const char *controller; @@ -231,11 +232,86 @@ static int make_controller(const char *path, mode_t mode) goto cleanup; } +# undef MAKE_FILE + ret = 0; cleanup: return ret; } + +static int make_controller_v2(const char *path, mode_t mode) +{ + if (!STRPREFIX(path, fakesysfscgroupdir)) { + errno = EINVAL; + return -1; + } + + if (real_mkdir(path, mode) < 0 && errno != EEXIST) + return -1; + +# define MAKE_FILE(name, value) \ + do { \ + if (make_file(path, name, value) < 0) \ + return -1; \ + } while (0) + + MAKE_FILE("cgroup.controllers", "cpu io memory\n"); + MAKE_FILE("cgroup.subtree_control", ""); + MAKE_FILE("cgroup.type", "domain\n"); + MAKE_FILE("cpu.max", "max 100000\n"); + MAKE_FILE("cpu.stat", + "usage_usec 0\n" + "user_usec 0\n" + "system_usec 0\n" + "nr_periods 0\n" + "nr_throttled 0\n" + "throttled_usec 0\n"); + MAKE_FILE("cpu.weight", "100\n"); + MAKE_FILE("memory.current", "1455321088\n"); + MAKE_FILE("memory.high", "max\n"); + MAKE_FILE("memory.max", "max\n"); + MAKE_FILE("memory.stat", + "anon 0\n" + "file 0\n" + "kernel_stack 0\n" + "slab 0\n" + "sock 0\n" + "shmem 0\n" + "file_mapped 0\n" + "file_dirty 0\n" + "file_writeback 0\n" + "inactive_anon 0\n" + "active_anon 0\n" + "inactive_file 0\n" + "active_file 0\n" + "unevictable 0\n" + "slab_reclaimable 0\n" + "slab_unreclaimable 0\n" + "pgfault 0\n" + "pgmajfault 0\n" + "pgrefill 0\n" + "pgscan 0\n" + "pgsteal 0\n" + "pgactivate 0\n" + "pgdeactivate 0\n" + "pglazyfree 0\n" + "pglazyfreed 0\n" + "workingset_refault 0\n" + "workingset_activate 0\n" + "workingset_nodereclaim 0\n"); + MAKE_FILE("memory.swap.current", "0\n"); + MAKE_FILE("memory.swap.max", "max\n"); + MAKE_FILE("io.stat", "8:0 rbytes=26828800 wbytes=77062144 rios=2256 wios=7849 dbytes=0 dios=0\n"); + MAKE_FILE("io.max", ""); + MAKE_FILE("io.weight", "default 100\n"); + +# undef MAKE_FILE + + return 0; +} + + static void init_syms(void) { if (real_fopen) @@ -249,16 +325,64 @@ static void init_syms(void) VIR_MOCK_REAL_INIT(open); } + +static int make_controller(const char *path, mode_t mode) +{ + const char *mock; + bool unified = false; + bool hybrid = false; + + mock = getenv("VIR_CGROUP_MOCK_MODE"); + if (mock) { + if (STREQ(mock, "unified")) { + unified = true; + } else if (STREQ(mock, "hybrid")) { + hybrid = true; + } else { + fprintf(stderr, "invalid mode '%s'\n", mock); + abort(); + } + } + + if (unified || (hybrid && strstr(path, "unified"))) { + return make_controller_v2(path, mode); + } else { + return make_controller_v1(path, mode); + } +} + + static void init_sysfs(void) { - if (fakerootdir && fakesysfscgroupdir) - return; + const char *mock; + char *newfakerootdir; + bool unified = false; + bool hybrid = false; - if (!(fakerootdir = getenv("LIBVIRT_FAKE_ROOT_DIR"))) { + if (!(newfakerootdir = getenv("LIBVIRT_FAKE_ROOT_DIR"))) { fprintf(stderr, "Missing LIBVIRT_FAKE_ROOT_DIR env variable\n"); abort(); } + if (fakerootdir && STREQ(fakerootdir, newfakerootdir)) + return; + + fakerootdir = newfakerootdir; + + mock = getenv("VIR_CGROUP_MOCK_MODE"); + if (mock) { + if (STREQ(mock, "unified")) { + unified = true; + } else if (STREQ(mock, "hybrid")) { + hybrid = true; + } else { + fprintf(stderr, "invalid mode '%s'\n", mock); + abort(); + } + } + + VIR_FREE(fakesysfscgroupdir); + if (virAsprintfQuiet(&fakesysfscgroupdir, "%s%s", fakerootdir, SYSFS_CGROUP_PREFIX) < 0) abort(); @@ -281,18 +405,25 @@ static void init_sysfs(void) free(path); \ } 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"); + if (unified) { + MAKE_CONTROLLER(""); + } else if (hybrid) { + MAKE_CONTROLLER("unified"); + MAKE_CONTROLLER("cpuset"); + MAKE_CONTROLLER("freezer"); + } else { + MAKE_CONTROLLER("cpu"); + MAKE_CONTROLLER("cpuacct"); + MAKE_CONTROLLER("cpu,cpuacct"); + MAKE_CONTROLLER("cpuset"); + MAKE_CONTROLLER("blkio"); + MAKE_CONTROLLER("memory"); + MAKE_CONTROLLER("freezer"); - if (make_file(fakesysfscgroupdir, - SYSFS_CPU_PRESENT_MOCKED, "8-23,48-159\n") < 0) - abort(); + if (make_file(fakesysfscgroupdir, + SYSFS_CPU_PRESENT_MOCKED, "8-23,48-159\n") < 0) + abort(); + } } -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- tests/vircgroupdata/all-in-one.parsed | 1 + tests/vircgroupdata/cgroups1.parsed | 1 + tests/vircgroupdata/cgroups2.parsed | 1 + tests/vircgroupdata/cgroups3.parsed | 1 + tests/vircgroupdata/fedora-18.parsed | 1 + tests/vircgroupdata/fedora-21.parsed | 1 + tests/vircgroupdata/kubevirt.parsed | 1 + tests/vircgroupdata/ovirt-node-6.6.parsed | 1 + tests/vircgroupdata/ovirt-node-7.1.parsed | 1 + tests/vircgroupdata/rhel-7.1.parsed | 1 + tests/vircgrouptest.c | 2 ++ 11 files changed, 12 insertions(+) diff --git a/tests/vircgroupdata/all-in-one.parsed b/tests/vircgroupdata/all-in-one.parsed index d703d08fb9..97c96e8ecd 100644 --- a/tests/vircgroupdata/all-in-one.parsed +++ b/tests/vircgroupdata/all-in-one.parsed @@ -8,3 +8,4 @@ blkio /not/really/sys/fs/cgroup net_cls <null> perf_event <null> name=systemd <null> +unified <null> diff --git a/tests/vircgroupdata/cgroups1.parsed b/tests/vircgroupdata/cgroups1.parsed index b6916f17a1..16431aa006 100644 --- a/tests/vircgroupdata/cgroups1.parsed +++ b/tests/vircgroupdata/cgroups1.parsed @@ -8,3 +8,4 @@ blkio /sys/fs/cgroup/blkio net_cls /sys/fs/cgroup/net_cls perf_event <null> name=systemd <null> +unified <null> diff --git a/tests/vircgroupdata/cgroups2.parsed b/tests/vircgroupdata/cgroups2.parsed index 5eb2bc7bb2..d1bb0dcb7c 100644 --- a/tests/vircgroupdata/cgroups2.parsed +++ b/tests/vircgroupdata/cgroups2.parsed @@ -8,3 +8,4 @@ blkio /sys/fs/cgroup/blkio net_cls <null> perf_event /sys/fs/cgroup/perf_event name=systemd <null> +unified <null> diff --git a/tests/vircgroupdata/cgroups3.parsed b/tests/vircgroupdata/cgroups3.parsed index 2b1f3825c1..44e475c9d2 100644 --- a/tests/vircgroupdata/cgroups3.parsed +++ b/tests/vircgroupdata/cgroups3.parsed @@ -8,3 +8,4 @@ blkio /sys/fs/cgroup/blkio net_cls /sys/fs/cgroup/net_cls perf_event /sys/fs/cgroup/perf_event name=systemd <null> +unified <null> diff --git a/tests/vircgroupdata/fedora-18.parsed b/tests/vircgroupdata/fedora-18.parsed index 8d5ba75c7e..662a38a9e8 100644 --- a/tests/vircgroupdata/fedora-18.parsed +++ b/tests/vircgroupdata/fedora-18.parsed @@ -8,3 +8,4 @@ blkio /sys/fs/cgroup/blkio net_cls /sys/fs/cgroup/net_cls perf_event /sys/fs/cgroup/perf_event name=systemd /sys/fs/cgroup/systemd +unified <null> diff --git a/tests/vircgroupdata/fedora-21.parsed b/tests/vircgroupdata/fedora-21.parsed index 3377af0382..4e447fd7bd 100644 --- a/tests/vircgroupdata/fedora-21.parsed +++ b/tests/vircgroupdata/fedora-21.parsed @@ -8,3 +8,4 @@ blkio /sys/fs/cgroup/blkio net_cls /sys/fs/cgroup/net_cls,net_prio perf_event /sys/fs/cgroup/perf_event name=systemd /sys/fs/cgroup/systemd +unified <null> diff --git a/tests/vircgroupdata/kubevirt.parsed b/tests/vircgroupdata/kubevirt.parsed index 6948707238..bf977f8363 100644 --- a/tests/vircgroupdata/kubevirt.parsed +++ b/tests/vircgroupdata/kubevirt.parsed @@ -8,3 +8,4 @@ blkio /sys/fs/cgroup/blkio net_cls <null> perf_event /sys/fs/cgroup/perf_event name=systemd <null> +unified <null> diff --git a/tests/vircgroupdata/ovirt-node-6.6.parsed b/tests/vircgroupdata/ovirt-node-6.6.parsed index 01bf466be6..9d10813d8c 100644 --- a/tests/vircgroupdata/ovirt-node-6.6.parsed +++ b/tests/vircgroupdata/ovirt-node-6.6.parsed @@ -8,3 +8,4 @@ blkio /cgroup/blkio net_cls /cgroup/net_cls perf_event <null> name=systemd <null> +unified <null> diff --git a/tests/vircgroupdata/ovirt-node-7.1.parsed b/tests/vircgroupdata/ovirt-node-7.1.parsed index 8d5ba75c7e..662a38a9e8 100644 --- a/tests/vircgroupdata/ovirt-node-7.1.parsed +++ b/tests/vircgroupdata/ovirt-node-7.1.parsed @@ -8,3 +8,4 @@ blkio /sys/fs/cgroup/blkio net_cls /sys/fs/cgroup/net_cls perf_event /sys/fs/cgroup/perf_event name=systemd /sys/fs/cgroup/systemd +unified <null> diff --git a/tests/vircgroupdata/rhel-7.1.parsed b/tests/vircgroupdata/rhel-7.1.parsed index 8d5ba75c7e..662a38a9e8 100644 --- a/tests/vircgroupdata/rhel-7.1.parsed +++ b/tests/vircgroupdata/rhel-7.1.parsed @@ -8,3 +8,4 @@ blkio /sys/fs/cgroup/blkio net_cls /sys/fs/cgroup/net_cls perf_event /sys/fs/cgroup/perf_event name=systemd /sys/fs/cgroup/systemd +unified <null> diff --git a/tests/vircgrouptest.c b/tests/vircgrouptest.c index 3e8793a6c3..75187216d9 100644 --- a/tests/vircgrouptest.c +++ b/tests/vircgrouptest.c @@ -175,6 +175,8 @@ testCgroupDetectMounts(const void *args) virCgroupControllerTypeToString(i), NULLSTR(group->legacy[i].mountPoint)); } + virBufferAsprintf(&buf, "%-12s %s\n", + "unified", NULLSTR(group->unified.mountPoint)); if (virBufferCheckError(&buf) < 0) goto cleanup; -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- tests/vircgroupdata/hybrid.cgroups | 12 ++++++++++++ tests/vircgroupdata/hybrid.mounts | 23 +++++++++++++++++++++++ tests/vircgroupdata/hybrid.parsed | 11 +++++++++++ tests/vircgroupdata/hybrid.self.cgroup | 9 +++++++++ tests/vircgrouptest.c | 3 +++ 5 files changed, 58 insertions(+) create mode 100644 tests/vircgroupdata/hybrid.cgroups create mode 100644 tests/vircgroupdata/hybrid.mounts create mode 100644 tests/vircgroupdata/hybrid.parsed create mode 100644 tests/vircgroupdata/hybrid.self.cgroup diff --git a/tests/vircgroupdata/hybrid.cgroups b/tests/vircgroupdata/hybrid.cgroups new file mode 100644 index 0000000000..7f3bc7b8cb --- /dev/null +++ b/tests/vircgroupdata/hybrid.cgroups @@ -0,0 +1,12 @@ +#subsys_name hierarchy num_cgroups enabled +cpuset 2 1 1 +cpu 0 1 1 +cpuacct 0 1 1 +blkio 0 1 1 +memory 0 1 1 +devices 3 1 1 +freezer 5 1 1 +net_cls 4 1 1 +perf_event 6 1 1 +hugetlb 7 1 1 +pids 8 1 1 diff --git a/tests/vircgroupdata/hybrid.mounts b/tests/vircgroupdata/hybrid.mounts new file mode 100644 index 0000000000..d6f5f82115 --- /dev/null +++ b/tests/vircgroupdata/hybrid.mounts @@ -0,0 +1,23 @@ +proc /proc proc rw,nosuid,nodev,noexec,relatime 0 0 +sysfs /sys sysfs rw,nosuid,nodev,noexec,relatime 0 0 +devtmpfs /dev devtmpfs rw,nosuid,size=10240k,nr_inodes=502705,mode=755 0 0 +devpts /dev/pts devpts rw,nosuid,noexec,relatime,gid=5,mode=620,ptmxmode=000 0 0 +tmpfs /dev/shm tmpfs rw,nosuid,nodev,noexec 0 0 +tmpfs /run tmpfs rw,nosuid,nodev,noexec,mode=755 0 0 +mqueue /dev/mqueue mqueue rw,nosuid,nodev,noexec,relatime 0 0 +securityfs /sys/kernel/security securityfs rw,nosuid,nodev,noexec,relatime 0 0 +debugfs /sys/kernel/debug debugfs rw,nosuid,nodev,noexec,relatime 0 0 +configfs /sys/kernel/config configfs rw,nosuid,nodev,noexec,relatime 0 0 +fusectl /sys/fs/fuse/connections fusectl rw,nosuid,nodev,noexec,relatime 0 0 +selinuxfs /sys/fs/selinux selinuxfs rw,relatime 0 0 +pstore /sys/fs/pstore pstore rw,nosuid,nodev,noexec,relatime 0 0 +cgroup_root /not/really/sys/fs/cgroup tmpfs rw,nosuid,nodev,noexec,relatime,size=10240k,mode=755 0 0 +openrc /not/really/sys/fs/cgroup/openrc cgroup rw,nosuid,nodev,noexec,relatime,release_agent=/lib64/rc/sh/cgroup-release-agent.sh,name=openrc 0 0 +none /not/really/sys/fs/cgroup/unified cgroup2 rw,nosuid,nodev,noexec,relatime,nsdelegate 0 0 +cpuset /not/really/sys/fs/cgroup/cpuset cgroup rw,nosuid,nodev,noexec,relatime,cpuset 0 0 +devices /not/really/sys/fs/cgroup/devices cgroup rw,nosuid,nodev,noexec,relatime,devices 0 0 +net_cls /not/really/sys/fs/cgroup/net_cls cgroup rw,nosuid,nodev,noexec,relatime,net_cls 0 0 +freezer /not/really/sys/fs/cgroup/freezer cgroup rw,nosuid,nodev,noexec,relatime,freezer 0 0 +perf_event /not/really/sys/fs/cgroup/perf_event cgroup rw,nosuid,nodev,noexec,relatime,perf_event 0 0 +hugetlb /not/really/sys/fs/cgroup/hugetlb cgroup rw,nosuid,nodev,noexec,relatime,hugetlb 0 0 +pids /not/really/sys/fs/cgroup/pids cgroup rw,nosuid,nodev,noexec,relatime,pids 0 0 diff --git a/tests/vircgroupdata/hybrid.parsed b/tests/vircgroupdata/hybrid.parsed new file mode 100644 index 0000000000..7600de5f45 --- /dev/null +++ b/tests/vircgroupdata/hybrid.parsed @@ -0,0 +1,11 @@ +cpu <null> +cpuacct <null> +cpuset /not/really/sys/fs/cgroup/cpuset +memory <null> +devices /not/really/sys/fs/cgroup/devices +freezer /not/really/sys/fs/cgroup/freezer +blkio <null> +net_cls /not/really/sys/fs/cgroup/net_cls +perf_event /not/really/sys/fs/cgroup/perf_event +name=systemd <null> +unified /not/really/sys/fs/cgroup/unified diff --git a/tests/vircgroupdata/hybrid.self.cgroup b/tests/vircgroupdata/hybrid.self.cgroup new file mode 100644 index 0000000000..2a08905c91 --- /dev/null +++ b/tests/vircgroupdata/hybrid.self.cgroup @@ -0,0 +1,9 @@ +8:pids:/ +7:hugetlb:/ +6:perf_event:/ +5:freezer:/ +4:net_cls:/ +3:devices:/ +2:cpuset:/ +1:name=openrc:/ +0::/ diff --git a/tests/vircgrouptest.c b/tests/vircgrouptest.c index 800522e311..e766ad7cbe 100644 --- a/tests/vircgrouptest.c +++ b/tests/vircgrouptest.c @@ -902,6 +902,9 @@ mymain(void) fakerootdir = initFakeFS("unified", NULL); DETECT_MOUNTS("unified"); cleanupFakeFS(fakerootdir); + fakerootdir = initFakeFS("hybrid", NULL); + DETECT_MOUNTS("hybrid"); + cleanupFakeFS(fakerootdir); fakerootdir = initFakeFS(NULL, "systemd"); if (virTestRun("New cgroup for self", testCgroupNewForSelf, NULL) < 0) -- 2.17.1

Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- tests/vircgrouptest.c | 34 +++++++++++++++++++++++++++++++--- 1 file changed, 31 insertions(+), 3 deletions(-) diff --git a/tests/vircgrouptest.c b/tests/vircgrouptest.c index ead5046600..fe0cb9585a 100644 --- a/tests/vircgrouptest.c +++ b/tests/vircgrouptest.c @@ -567,16 +567,37 @@ static int testCgroupNewForSelfAllInOne(const void *args ATTRIBUTE_UNUSED) static int testCgroupNewForSelfLogind(const void *args ATTRIBUTE_UNUSED) +{ + virCgroupPtr cgroup = NULL; + + if (virCgroupNewSelf(&cgroup) >= 0) { + fprintf(stderr, "Expected to fail, only systemd cgroup available.\n"); + virCgroupFree(&cgroup); + return -1; + } + + return 0; +} + + +static int testCgroupNewForSelfUnified(const void *args ATTRIBUTE_UNUSED) { virCgroupPtr cgroup = NULL; int ret = -1; + const char *empty[VIR_CGROUP_CONTROLLER_LAST] = { 0 }; + unsigned int controllers = + (1 << VIR_CGROUP_CONTROLLER_CPU) | + (1 << VIR_CGROUP_CONTROLLER_CPUACCT) | + (1 << VIR_CGROUP_CONTROLLER_MEMORY) | + (1 << VIR_CGROUP_CONTROLLER_BLKIO); - if (virCgroupNewSelf(&cgroup) == 0) { - fprintf(stderr, "Expected cgroup creation to fail.\n"); + if (virCgroupNewSelf(&cgroup) < 0) { + fprintf(stderr, "Cannot create cgroup for self\n"); goto cleanup; } - ret = 0; + ret = validateCgroup(cgroup, "", empty, empty, empty, + "/not/really/sys/fs/cgroup", "/", controllers); cleanup: virCgroupFree(&cgroup); return ret; @@ -993,7 +1014,14 @@ mymain(void) ret = -1; cleanupFakeFS(fakerootdir); + /* cgroup unified */ + fakerootdir = initFakeFS("unified", "unified"); + if (virTestRun("New cgroup for self (unified)", testCgroupNewForSelfUnified, NULL) < 0) + ret = -1; + if (virTestRun("Cgroup available (unified)", testCgroupAvailable, (void*)0x1) < 0) + ret = -1; + cleanupFakeFS(fakerootdir); return ret == 0 ? EXIT_SUCCESS : EXIT_FAILURE; } -- 2.17.1

This removes code duplication and simplifies cgroup detection. As a drawback we will not have separate messages to enable cgroup controller in kernel or to mount it. On the other side the rewrite adds support for cgroup v2. The kernel config support was wrong because it was parsing '/proc/self/cgroup' instead of '/proc/cgroups/' file. The mount suggestion is removed as well because it will not work with cgroup v2. Signed-off-by: Pavel Hrdina <phrdina@redhat.com> --- tools/virt-host-validate-common.c | 162 ++++++------------------------ tools/virt-host-validate-common.h | 7 +- tools/virt-host-validate-lxc.c | 40 ++------ tools/virt-host-validate-qemu.c | 40 ++------ 4 files changed, 54 insertions(+), 195 deletions(-) diff --git a/tools/virt-host-validate-common.c b/tools/virt-host-validate-common.c index ccbd764c84..4e70fe9e9c 100644 --- a/tools/virt-host-validate-common.c +++ b/tools/virt-host-validate-common.c @@ -24,12 +24,10 @@ #include <stdarg.h> #include <unistd.h> #include <sys/utsname.h> -#ifdef HAVE_MNTENT_H -# include <mntent.h> -#endif /* HAVE_MNTENT_H */ #include <sys/stat.h> #include "viralloc.h" +#include "vircgroup.h" #include "virfile.h" #include "virt-host-validate-common.h" #include "virstring.h" @@ -288,152 +286,50 @@ int virHostValidateLinuxKernel(const char *hvname, } } - -static int virHostValidateCGroupSupport(const char *hvname, - const char *cg_name, - virHostValidateLevel level, - const char *config_name) -{ - virHostMsgCheck(hvname, "for cgroup '%s' controller support", cg_name); - FILE *fp = fopen("/proc/self/cgroup", "r"); - size_t len = 0; - char *line = NULL; - ssize_t ret; - bool matched = false; - - if (!fp) - goto error; - - while ((ret = getline(&line, &len, fp)) >= 0 && !matched) { - char **cgroups; - char *start; - char *end; - size_t ncgroups; - size_t i; - - /* Each line in this file looks like - * - * 4:cpu,cpuacct:/machine.slice/machine-qemu\x2dtest.scope/emulator - * - * Since multiple cgroups can be part of the same line and some cgroup - * names can appear as part of other cgroup names (eg. 'cpu' is a - * prefix for both 'cpuacct' and 'cpuset'), it's not enough to simply - * check whether the cgroup name is present somewhere inside the file. - * - * Moreover, there's nothing stopping the cgroup name from appearing - * in an unrelated mount point name as well */ - - /* Look for the first colon. - * The part we're interested in starts right after it */ - if (!(start = strchr(line, ':'))) - continue; - start++; - - /* Look for the second colon. - * The part we're interested in ends exactly there */ - if (!(end = strchr(start, ':'))) - continue; - *end = '\0'; - - if (!(cgroups = virStringSplitCount(start, ",", 0, &ncgroups))) - continue; - - /* Look for the matching cgroup */ - for (i = 0; i < ncgroups; i++) { - if (STREQ(cgroups[i], cg_name)) - matched = true; - } - - virStringListFreeCount(cgroups, ncgroups); - } - - VIR_FREE(line); - VIR_FORCE_FCLOSE(fp); - if (!matched) - goto error; - - virHostMsgPass(); - return 0; - - error: - VIR_FREE(line); - virHostMsgFail(level, "Enable CONFIG_%s in kernel Kconfig file", config_name); - return -1; -} - -#ifdef HAVE_MNTENT_H -static int virHostValidateCGroupMount(const char *hvname, - const char *cg_name, - virHostValidateLevel level) +#ifdef __linux__ +int virHostValidateCGroupControllers(const char *hvname, + int controllers, + virHostValidateLevel level) { - virHostMsgCheck(hvname, "for cgroup '%s' controller mount-point", cg_name); - FILE *fp = setmntent("/proc/mounts", "r"); - struct mntent ent; - char mntbuf[1024]; - bool matched = false; + virCgroupPtr group = NULL; + int ret = 0; + size_t i; - if (!fp) - goto error; + if (virCgroupNewSelf(&group) < 0) + return -1; - while (getmntent_r(fp, &ent, mntbuf, sizeof(mntbuf)) && !matched) { - char **opts; - size_t nopts; - size_t i; + for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) { + int flag = 1 << i; + const char *cg_name = virCgroupControllerTypeToString(i); - /* Ignore non-cgroup mounts */ - if (STRNEQ(ent.mnt_type, "cgroup")) + if (!(controllers & flag)) continue; - if (!(opts = virStringSplitCount(ent.mnt_opts, ",", 0, &nopts))) - continue; + virHostMsgCheck(hvname, "for cgroup '%s' controller support", cg_name); - /* Look for a mount option matching the cgroup name */ - for (i = 0; i < nopts; i++) { - if (STREQ(opts[i], cg_name)) - matched = true; + if (!virCgroupHasController(group, i)) { + ret = -1; + virHostMsgFail(level, "Enable '%s' in kernel Kconfig file or " + "mount/enable cgroup controller in your system", + cg_name); + } else { + virHostMsgPass(); } - - virStringListFreeCount(opts, nopts); } - endmntent(fp); - if (!matched) - goto error; - virHostMsgPass(); - return 0; + virCgroupFree(&group); - error: - virHostMsgFail(level, "Mount '%s' cgroup controller (suggested at /sys/fs/cgroup/%s)", - cg_name, cg_name); - return -1; + return ret; } -#else /* ! HAVE_MNTENT_H */ -static int virHostValidateCGroupMount(const char *hvname, - const char *cg_name, - virHostValidateLevel level) +#else /* !__linux__ */ +int virHostValidateCGroupControllers(const char *hvname, + int controllers, + virHostValidateLevel level) { - virHostMsgCheck(hvname, "for cgroup '%s' controller mount-point", cg_name); virHostMsgFail(level, "%s", "This platform does not support cgroups"); return -1; } -#endif /* ! HAVE_MNTENT_H */ - -int virHostValidateCGroupController(const char *hvname, - const char *cg_name, - virHostValidateLevel level, - const char *config_name) -{ - if (virHostValidateCGroupSupport(hvname, - cg_name, - level, - config_name) < 0) - return -1; - if (virHostValidateCGroupMount(hvname, - cg_name, - level) < 0) - return -1; - return 0; -} +#endif /* !__linux__ */ int virHostValidateIOMMU(const char *hvname, virHostValidateLevel level) diff --git a/tools/virt-host-validate-common.h b/tools/virt-host-validate-common.h index b6fe17daa7..b23dd7cdbe 100644 --- a/tools/virt-host-validate-common.h +++ b/tools/virt-host-validate-common.h @@ -77,10 +77,9 @@ int virHostValidateNamespace(const char *hvname, virHostValidateLevel level, const char *hint); -int virHostValidateCGroupController(const char *hvname, - const char *cg_name, - virHostValidateLevel level, - const char *config_name); +int virHostValidateCGroupControllers(const char *hvname, + int controllers, + virHostValidateLevel level); int virHostValidateIOMMU(const char *hvname, virHostValidateLevel level); diff --git a/tools/virt-host-validate-lxc.c b/tools/virt-host-validate-lxc.c index 64d9279c30..3c55b1b26d 100644 --- a/tools/virt-host-validate-lxc.c +++ b/tools/virt-host-validate-lxc.c @@ -23,6 +23,7 @@ #include "virt-host-validate-lxc.h" #include "virt-host-validate-common.h" +#include "vircgroup.h" int virHostValidateLXC(void) { @@ -63,35 +64,16 @@ int virHostValidateLXC(void) _("User namespace support is recommended")) < 0) ret = -1; - if (virHostValidateCGroupController("LXC", "memory", - VIR_HOST_VALIDATE_FAIL, - "MEMCG") < 0) - ret = -1; - - if (virHostValidateCGroupController("LXC", "cpu", - VIR_HOST_VALIDATE_FAIL, - "CGROUP_CPU") < 0) - ret = -1; - - if (virHostValidateCGroupController("LXC", "cpuacct", - VIR_HOST_VALIDATE_FAIL, - "CGROUP_CPUACCT") < 0) - ret = -1; - - if (virHostValidateCGroupController("LXC", "cpuset", - VIR_HOST_VALIDATE_FAIL, - "CPUSETS") < 0) - ret = -1; - - if (virHostValidateCGroupController("LXC", "devices", - VIR_HOST_VALIDATE_FAIL, - "CGROUP_DEVICE") < 0) - ret = -1; - - if (virHostValidateCGroupController("LXC", "blkio", - VIR_HOST_VALIDATE_FAIL, - "BLK_CGROUP") < 0) - ret = -1; + if (virHostValidateCGroupControllers("LXC", + (1 << VIR_CGROUP_CONTROLLER_MEMORY) | + (1 << VIR_CGROUP_CONTROLLER_CPU) | + (1 << VIR_CGROUP_CONTROLLER_CPUACCT) | + (1 << VIR_CGROUP_CONTROLLER_CPUSET) | + (1 << VIR_CGROUP_CONTROLLER_DEVICES) | + (1 << VIR_CGROUP_CONTROLLER_BLKIO), + VIR_HOST_VALIDATE_FAIL) < 0) { + ret = -1; + } #if WITH_FUSE if (virHostValidateDeviceExists("LXC", "/sys/fs/fuse/connections", diff --git a/tools/virt-host-validate-qemu.c b/tools/virt-host-validate-qemu.c index d7573ea8b3..ff3c1f0231 100644 --- a/tools/virt-host-validate-qemu.c +++ b/tools/virt-host-validate-qemu.c @@ -26,6 +26,7 @@ #include "virt-host-validate-common.h" #include "virarch.h" #include "virbitmap.h" +#include "vircgroup.h" int virHostValidateQEMU(void) { @@ -96,35 +97,16 @@ int virHostValidateQEMU(void) _("Load the 'tun' module to enable networking for QEMU guests")) < 0) ret = -1; - if (virHostValidateCGroupController("QEMU", "memory", - VIR_HOST_VALIDATE_WARN, - "MEMCG") < 0) - ret = -1; - - if (virHostValidateCGroupController("QEMU", "cpu", - VIR_HOST_VALIDATE_WARN, - "CGROUP_CPU") < 0) - ret = -1; - - if (virHostValidateCGroupController("QEMU", "cpuacct", - VIR_HOST_VALIDATE_WARN, - "CGROUP_CPUACCT") < 0) - ret = -1; - - if (virHostValidateCGroupController("QEMU", "cpuset", - VIR_HOST_VALIDATE_WARN, - "CPUSETS") < 0) - ret = -1; - - if (virHostValidateCGroupController("QEMU", "devices", - VIR_HOST_VALIDATE_WARN, - "CGROUP_DEVICES") < 0) - ret = -1; - - if (virHostValidateCGroupController("QEMU", "blkio", - VIR_HOST_VALIDATE_WARN, - "BLK_CGROUP") < 0) - ret = -1; + if (virHostValidateCGroupControllers("QEMU", + (1 << VIR_CGROUP_CONTROLLER_MEMORY) | + (1 << VIR_CGROUP_CONTROLLER_CPU) | + (1 << VIR_CGROUP_CONTROLLER_CPUACCT) | + (1 << VIR_CGROUP_CONTROLLER_CPUSET) | + (1 << VIR_CGROUP_CONTROLLER_DEVICES) | + (1 << VIR_CGROUP_CONTROLLER_BLKIO), + VIR_HOST_VALIDATE_WARN) < 0) { + ret = -1; + } if (virHostValidateIOMMU("QEMU", VIR_HOST_VALIDATE_WARN) < 0) -- 2.17.1