picnoir
/
Systemd
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
Systemd/src/core/dbus-cgroup.c

1427 lines
59 KiB
C
Raw Normal View History

Add SPDX license identifiers to source files under the LGPL This follows what the kernel is doing, c.f. https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=5fd54ace4721fc5ce2bb5aef6318fcf17f421460.
2017-11-18 17:09:20 +01:00
/* SPDX-License-Identifier: LGPL-2.1+ */
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
#include <arpa/inet.h>
#include "af-list.h"
util-lib: split out allocation calls into alloc-util.[ch]
2015-10-27 03:01:06 +01:00
#include "alloc-util.h"
core: warn loudly if IP firewalling is configured but not in effect
2017-09-06 17:56:15 +02:00
#include "bpf-firewall.h"
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
#include "bus-util.h"
#include "cgroup-util.h"
#include "cgroup.h"
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
#include "dbus-cgroup.h"
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
#include "dbus-util.h"
util-lib: split out fd-related operations into fd-util.[ch] There are more than enough to deserve their own .c file, hence move them over.
2015-10-25 13:14:12 +01:00
#include "fd-util.h"
util-lib: move more file I/O related calls into fileio.[ch]
2015-10-26 18:05:03 +01:00
#include "fileio.h"
util: split out some stuff into a new file limits-util.[ch]
2019-03-13 11:35:47 +01:00
#include "limits-util.h"
util-lib: split out fd-related operations into fd-util.[ch] There are more than enough to deserve their own .c file, hence move them over.
2015-10-25 13:14:12 +01:00
#include "path-util.h"
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
static BUS_DEFINE_PROPERTY_GET_ENUM(property_get_cgroup_device_policy, cgroup_device_policy, CGroupDevicePolicy);
cgroup: Add DisableControllers= directive to disable controller in subtree Some controllers (like the CPU controller) have a performance cost that is non-trivial on certain workloads. While this can be mitigated and improved to an extent, there will for some controllers always be some overheads associated with the benefits gained from the controller. Inside Facebook, the fix applied has been to disable the CPU controller forcibly with `cgroup_disable=cpu` on the kernel command line. This presents a problem: to disable or reenable the controller, a reboot is required, but this is quite cumbersome and slow to do for many thousands of machines, especially machines where disabling/enabling a stateful service on a machine is a matter of several minutes. Currently systemd provides some configuration knobs for these in the form of `[Default]CPUAccounting`, `[Default]MemoryAccounting`, and the like. The limitation of these is that Default*Accounting is overrideable by individual services, of which any one could decide to reenable a controller within the hierarchy at any point just by using a controller feature implicitly (eg. `CPUWeight`), even if the use of that CPU feature could just be opportunistic. Since many services are provided by the distribution, or by upstream teams at a particular organisation, it's not a sustainable solution to simply try to find and remove offending directives from these units. This commit presents a more direct solution -- a DisableControllers= directive that forcibly disallows a controller from being enabled within a subtree.
2018-12-03 15:38:06 +01:00
static int property_get_cgroup_mask(
core: rework the Delegate= unit file setting to take a list of controller names Previously it was not possible to select which controllers to enable for a unit where Delegate=yes was set, as all controllers were enabled. With this change, this is made configurable, and thus delegation units can pick specifically what they want to manage themselves, and what they don't care about.
2017-11-09 15:29:34 +01:00
sd_bus *bus,
const char *path,
const char *interface,
const char *property,
sd_bus_message *reply,
void *userdata,
sd_bus_error *error) {
cgroup: Add DisableControllers= directive to disable controller in subtree Some controllers (like the CPU controller) have a performance cost that is non-trivial on certain workloads. While this can be mitigated and improved to an extent, there will for some controllers always be some overheads associated with the benefits gained from the controller. Inside Facebook, the fix applied has been to disable the CPU controller forcibly with `cgroup_disable=cpu` on the kernel command line. This presents a problem: to disable or reenable the controller, a reboot is required, but this is quite cumbersome and slow to do for many thousands of machines, especially machines where disabling/enabling a stateful service on a machine is a matter of several minutes. Currently systemd provides some configuration knobs for these in the form of `[Default]CPUAccounting`, `[Default]MemoryAccounting`, and the like. The limitation of these is that Default*Accounting is overrideable by individual services, of which any one could decide to reenable a controller within the hierarchy at any point just by using a controller feature implicitly (eg. `CPUWeight`), even if the use of that CPU feature could just be opportunistic. Since many services are provided by the distribution, or by upstream teams at a particular organisation, it's not a sustainable solution to simply try to find and remove offending directives from these units. This commit presents a more direct solution -- a DisableControllers= directive that forcibly disallows a controller from being enabled within a subtree.
2018-12-03 15:38:06 +01:00
CGroupMask *mask = userdata;
CGroupController ctrl;
core: rework the Delegate= unit file setting to take a list of controller names Previously it was not possible to select which controllers to enable for a unit where Delegate=yes was set, as all controllers were enabled. With this change, this is made configurable, and thus delegation units can pick specifically what they want to manage themselves, and what they don't care about.
2017-11-09 15:29:34 +01:00
int r;
assert(bus);
assert(reply);
r = sd_bus_message_open_container(reply, 'a', "s");
if (r < 0)
return r;
cgroup: Add DisableControllers= directive to disable controller in subtree Some controllers (like the CPU controller) have a performance cost that is non-trivial on certain workloads. While this can be mitigated and improved to an extent, there will for some controllers always be some overheads associated with the benefits gained from the controller. Inside Facebook, the fix applied has been to disable the CPU controller forcibly with `cgroup_disable=cpu` on the kernel command line. This presents a problem: to disable or reenable the controller, a reboot is required, but this is quite cumbersome and slow to do for many thousands of machines, especially machines where disabling/enabling a stateful service on a machine is a matter of several minutes. Currently systemd provides some configuration knobs for these in the form of `[Default]CPUAccounting`, `[Default]MemoryAccounting`, and the like. The limitation of these is that Default*Accounting is overrideable by individual services, of which any one could decide to reenable a controller within the hierarchy at any point just by using a controller feature implicitly (eg. `CPUWeight`), even if the use of that CPU feature could just be opportunistic. Since many services are provided by the distribution, or by upstream teams at a particular organisation, it's not a sustainable solution to simply try to find and remove offending directives from these units. This commit presents a more direct solution -- a DisableControllers= directive that forcibly disallows a controller from being enabled within a subtree.
2018-12-03 15:38:06 +01:00
for (ctrl = 0; ctrl < _CGROUP_CONTROLLER_MAX; ctrl++) {
if ((*mask & CGROUP_CONTROLLER_TO_MASK(ctrl)) == 0)
core: rework the Delegate= unit file setting to take a list of controller names Previously it was not possible to select which controllers to enable for a unit where Delegate=yes was set, as all controllers were enabled. With this change, this is made configurable, and thus delegation units can pick specifically what they want to manage themselves, and what they don't care about.
2017-11-09 15:29:34 +01:00
continue;
cgroup: Add DisableControllers= directive to disable controller in subtree Some controllers (like the CPU controller) have a performance cost that is non-trivial on certain workloads. While this can be mitigated and improved to an extent, there will for some controllers always be some overheads associated with the benefits gained from the controller. Inside Facebook, the fix applied has been to disable the CPU controller forcibly with `cgroup_disable=cpu` on the kernel command line. This presents a problem: to disable or reenable the controller, a reboot is required, but this is quite cumbersome and slow to do for many thousands of machines, especially machines where disabling/enabling a stateful service on a machine is a matter of several minutes. Currently systemd provides some configuration knobs for these in the form of `[Default]CPUAccounting`, `[Default]MemoryAccounting`, and the like. The limitation of these is that Default*Accounting is overrideable by individual services, of which any one could decide to reenable a controller within the hierarchy at any point just by using a controller feature implicitly (eg. `CPUWeight`), even if the use of that CPU feature could just be opportunistic. Since many services are provided by the distribution, or by upstream teams at a particular organisation, it's not a sustainable solution to simply try to find and remove offending directives from these units. This commit presents a more direct solution -- a DisableControllers= directive that forcibly disallows a controller from being enabled within a subtree.
2018-12-03 15:38:06 +01:00
r = sd_bus_message_append(reply, "s", cgroup_controller_to_string(ctrl));
core: rework the Delegate= unit file setting to take a list of controller names Previously it was not possible to select which controllers to enable for a unit where Delegate=yes was set, as all controllers were enabled. With this change, this is made configurable, and thus delegation units can pick specifically what they want to manage themselves, and what they don't care about.
2017-11-09 15:29:34 +01:00
if (r < 0)
return r;
}
return sd_bus_message_close_container(reply);
}
cgroup: Add DisableControllers= directive to disable controller in subtree Some controllers (like the CPU controller) have a performance cost that is non-trivial on certain workloads. While this can be mitigated and improved to an extent, there will for some controllers always be some overheads associated with the benefits gained from the controller. Inside Facebook, the fix applied has been to disable the CPU controller forcibly with `cgroup_disable=cpu` on the kernel command line. This presents a problem: to disable or reenable the controller, a reboot is required, but this is quite cumbersome and slow to do for many thousands of machines, especially machines where disabling/enabling a stateful service on a machine is a matter of several minutes. Currently systemd provides some configuration knobs for these in the form of `[Default]CPUAccounting`, `[Default]MemoryAccounting`, and the like. The limitation of these is that Default*Accounting is overrideable by individual services, of which any one could decide to reenable a controller within the hierarchy at any point just by using a controller feature implicitly (eg. `CPUWeight`), even if the use of that CPU feature could just be opportunistic. Since many services are provided by the distribution, or by upstream teams at a particular organisation, it's not a sustainable solution to simply try to find and remove offending directives from these units. This commit presents a more direct solution -- a DisableControllers= directive that forcibly disallows a controller from being enabled within a subtree.
2018-12-03 15:38:06 +01:00
static int property_get_delegate_controllers(
sd_bus *bus,
const char *path,
const char *interface,
const char *property,
sd_bus_message *reply,
void *userdata,
sd_bus_error *error) {
CGroupContext *c = userdata;
assert(bus);
assert(reply);
assert(c);
if (!c->delegate)
return sd_bus_message_append(reply, "as", 0);
return property_get_cgroup_mask(bus, path, interface, property, reply, &c->delegate_controllers, error);
}
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
static int property_get_io_device_weight(
sd_bus *bus,
const char *path,
const char *interface,
const char *property,
sd_bus_message *reply,
void *userdata,
sd_bus_error *error) {
CGroupContext *c = userdata;
CGroupIODeviceWeight *w;
int r;
assert(bus);
assert(reply);
assert(c);
r = sd_bus_message_open_container(reply, 'a', "(st)");
if (r < 0)
return r;
LIST_FOREACH(device_weights, w, c->io_device_weights) {
r = sd_bus_message_append(reply, "(st)", w->path, w->weight);
if (r < 0)
return r;
}
return sd_bus_message_close_container(reply);
}
static int property_get_io_device_limits(
sd_bus *bus,
const char *path,
const char *interface,
const char *property,
sd_bus_message *reply,
void *userdata,
sd_bus_error *error) {
CGroupContext *c = userdata;
CGroupIODeviceLimit *l;
int r;
assert(bus);
assert(reply);
assert(c);
r = sd_bus_message_open_container(reply, 'a', "(st)");
if (r < 0)
return r;
LIST_FOREACH(device_limits, l, c->io_device_limits) {
core: introduce CGroupIOLimitType enums Currently, there are two cgroup IO limits, bandwidth max for read and write, and they are hard-coded in various places. This is fine for two limits but IO is expected to grow more limits - low, high and max limits for bandwidth and IOPS - and hard-coding each limit won't make sense. This patch replaces hard-coded limits with an array indexed by CGroupIOLimitType and accompanying string and default value tables so that new limits can be added trivially.
2016-05-18 22:50:56 +02:00
CGroupIOLimitType type;
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
core: introduce CGroupIOLimitType enums Currently, there are two cgroup IO limits, bandwidth max for read and write, and they are hard-coded in various places. This is fine for two limits but IO is expected to grow more limits - low, high and max limits for bandwidth and IOPS - and hard-coding each limit won't make sense. This patch replaces hard-coded limits with an array indexed by CGroupIOLimitType and accompanying string and default value tables so that new limits can be added trivially.
2016-05-18 22:50:56 +02:00
type = cgroup_io_limit_type_from_string(property);
if (type < 0 || l->limits[type] == cgroup_io_limit_defaults[type])
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
continue;
core: introduce CGroupIOLimitType enums Currently, there are two cgroup IO limits, bandwidth max for read and write, and they are hard-coded in various places. This is fine for two limits but IO is expected to grow more limits - low, high and max limits for bandwidth and IOPS - and hard-coding each limit won't make sense. This patch replaces hard-coded limits with an array indexed by CGroupIOLimitType and accompanying string and default value tables so that new limits can be added trivially.
2016-05-18 22:50:56 +02:00
r = sd_bus_message_append(reply, "(st)", l->path, l->limits[type]);
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
if (r < 0)
return r;
}
return sd_bus_message_close_container(reply);
}
core: add IODeviceLatencyTargetSec This adds support for the following proposed latency based IO control mechanism. https://lkml.org/lkml/2018/6/5/428
2018-06-13 23:16:35 +02:00
static int property_get_io_device_latency(
sd_bus *bus,
const char *path,
const char *interface,
const char *property,
sd_bus_message *reply,
void *userdata,
sd_bus_error *error) {
CGroupContext *c = userdata;
CGroupIODeviceLatency *l;
int r;
assert(bus);
assert(reply);
assert(c);
r = sd_bus_message_open_container(reply, 'a', "(st)");
if (r < 0)
return r;
LIST_FOREACH(device_latencies, l, c->io_device_latencies) {
r = sd_bus_message_append(reply, "(st)", l->path, l->target_usec);
if (r < 0)
return r;
}
return sd_bus_message_close_container(reply);
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
static int property_get_blockio_device_weight(
sd_bus *bus,
const char *path,
const char *interface,
const char *property,
sd_bus_message *reply,
bus: rework message handlers to always take an error argument Message handler callbacks can be simplified drastically if the dispatcher automatically replies to method calls if errors are returned. Thus: add an sd_bus_error argument to all message handlers. When we dispatch a message handler and it returns negative or a set sd_bus_error we send this as message error back to the client. This means errors returned by handlers by default are given back to clients instead of rippling all the way up to the event loop, which is desirable to make things robust. As a side-effect we can now easily turn the SELinux checks into normal function calls, since the method call dispatcher will generate the right error replies automatically now. Also, make sure we always pass the error structure to all property and method handlers as last argument to follow the usual style of passing variables for return values as last argument.
2013-11-21 19:34:37 +01:00
void *userdata,
sd_bus_error *error) {
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
CGroupContext *c = userdata;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
CGroupBlockIODeviceWeight *w;
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
int r;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
assert(bus);
assert(reply);
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
assert(c);
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
r = sd_bus_message_open_container(reply, 'a', "(st)");
if (r < 0)
return r;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
LIST_FOREACH(device_weights, w, c->blockio_device_weights) {
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
r = sd_bus_message_append(reply, "(st)", w->path, w->weight);
if (r < 0)
return r;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
return sd_bus_message_close_container(reply);
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
static int property_get_blockio_device_bandwidths(
sd_bus *bus,
const char *path,
const char *interface,
const char *property,
sd_bus_message *reply,
bus: rework message handlers to always take an error argument Message handler callbacks can be simplified drastically if the dispatcher automatically replies to method calls if errors are returned. Thus: add an sd_bus_error argument to all message handlers. When we dispatch a message handler and it returns negative or a set sd_bus_error we send this as message error back to the client. This means errors returned by handlers by default are given back to clients instead of rippling all the way up to the event loop, which is desirable to make things robust. As a side-effect we can now easily turn the SELinux checks into normal function calls, since the method call dispatcher will generate the right error replies automatically now. Also, make sure we always pass the error structure to all property and method handlers as last argument to follow the usual style of passing variables for return values as last argument.
2013-11-21 19:34:37 +01:00
void *userdata,
sd_bus_error *error) {
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
CGroupContext *c = userdata;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
CGroupBlockIODeviceBandwidth *b;
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
int r;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
assert(bus);
assert(reply);
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
assert(c);
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
r = sd_bus_message_open_container(reply, 'a', "(st)");
if (r < 0)
return r;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
LIST_FOREACH(device_bandwidths, b, c->blockio_device_bandwidths) {
core: update CGroupBlockIODeviceBandwidth to record both rbps and wbps CGroupBlockIODeviceBandwith is used to keep track of IO bandwidth limits for legacy cgroup hierarchies. Unlike the unified hierarchy counterpart CGroupIODeviceLimit, a CGroupBlockIODeviceBandwiddth records either a read or write limit and has a couple issues. * There's no way to clear specific config entry. * When configs are cleared for an IO direction of a unit, the kernel settings aren't cleared accordingly creating discrepancies. This patch updates CGroupBlockIODeviceBandwidth so that it behaves similarly to CGroupIODeviceLimit - each entry records both rbps and wbps limits and is cleared if both are at default values after kernel settings are updated.
2016-05-18 22:51:46 +02:00
uint64_t v;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
core: update CGroupBlockIODeviceBandwidth to record both rbps and wbps CGroupBlockIODeviceBandwith is used to keep track of IO bandwidth limits for legacy cgroup hierarchies. Unlike the unified hierarchy counterpart CGroupIODeviceLimit, a CGroupBlockIODeviceBandwiddth records either a read or write limit and has a couple issues. * There's no way to clear specific config entry. * When configs are cleared for an IO direction of a unit, the kernel settings aren't cleared accordingly creating discrepancies. This patch updates CGroupBlockIODeviceBandwidth so that it behaves similarly to CGroupIODeviceLimit - each entry records both rbps and wbps limits and is cleared if both are at default values after kernel settings are updated.
2016-05-18 22:51:46 +02:00
if (streq(property, "BlockIOReadBandwidth"))
v = b->rbps;
else
v = b->wbps;
if (v == CGROUP_LIMIT_MAX)
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
continue;
core: update CGroupBlockIODeviceBandwidth to record both rbps and wbps CGroupBlockIODeviceBandwith is used to keep track of IO bandwidth limits for legacy cgroup hierarchies. Unlike the unified hierarchy counterpart CGroupIODeviceLimit, a CGroupBlockIODeviceBandwiddth records either a read or write limit and has a couple issues. * There's no way to clear specific config entry. * When configs are cleared for an IO direction of a unit, the kernel settings aren't cleared accordingly creating discrepancies. This patch updates CGroupBlockIODeviceBandwidth so that it behaves similarly to CGroupIODeviceLimit - each entry records both rbps and wbps limits and is cleared if both are at default values after kernel settings are updated.
2016-05-18 22:51:46 +02:00
r = sd_bus_message_append(reply, "(st)", b->path, v);
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
if (r < 0)
return r;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
return sd_bus_message_close_container(reply);
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
static int property_get_device_allow(
sd_bus *bus,
const char *path,
const char *interface,
const char *property,
sd_bus_message *reply,
bus: rework message handlers to always take an error argument Message handler callbacks can be simplified drastically if the dispatcher automatically replies to method calls if errors are returned. Thus: add an sd_bus_error argument to all message handlers. When we dispatch a message handler and it returns negative or a set sd_bus_error we send this as message error back to the client. This means errors returned by handlers by default are given back to clients instead of rippling all the way up to the event loop, which is desirable to make things robust. As a side-effect we can now easily turn the SELinux checks into normal function calls, since the method call dispatcher will generate the right error replies automatically now. Also, make sure we always pass the error structure to all property and method handlers as last argument to follow the usual style of passing variables for return values as last argument.
2013-11-21 19:34:37 +01:00
void *userdata,
sd_bus_error *error) {
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
CGroupContext *c = userdata;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
CGroupDeviceAllow *a;
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
int r;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
assert(bus);
assert(reply);
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
assert(c);
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
r = sd_bus_message_open_container(reply, 'a', "(ss)");
if (r < 0)
return r;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
LIST_FOREACH(device_allow, a, c->device_allow) {
unsigned k = 0;
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
char rwm[4];
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
if (a->r)
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
rwm[k++] = 'r';
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
if (a->w)
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
rwm[k++] = 'w';
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
if (a->m)
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
rwm[k++] = 'm';
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
rwm[k] = 0;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
r = sd_bus_message_append(reply, "(ss)", a->path, rwm);
if (r < 0)
return r;
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
return sd_bus_message_close_container(reply);
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
}
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
static int property_get_ip_address_access(
sd_bus *bus,
const char *path,
const char *interface,
const char *property,
sd_bus_message *reply,
void *userdata,
sd_bus_error *error) {
IPAddressAccessItem** items = userdata, *i;
int r;
r = sd_bus_message_open_container(reply, 'a', "(iayu)");
if (r < 0)
return r;
LIST_FOREACH(items, i, *items) {
r = sd_bus_message_open_container(reply, 'r', "iayu");
if (r < 0)
return r;
r = sd_bus_message_append(reply, "i", i->family);
if (r < 0)
return r;
r = sd_bus_message_append_array(reply, 'y', &i->address, FAMILY_ADDRESS_SIZE(i->family));
if (r < 0)
return r;
r = sd_bus_message_append(reply, "u", (uint32_t) i->prefixlen);
if (r < 0)
return r;
r = sd_bus_message_close_container(reply);
if (r < 0)
return r;
}
return sd_bus_message_close_container(reply);
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
const sd_bus_vtable bus_cgroup_vtable[] = {
SD_BUS_VTABLE_START(0),
core: introduce new Delegate=yes/no property controlling creation of cgroup subhierarchies For priviliged units this resource control property ensures that the processes have all controllers systemd manages enabled. For unpriviliged services (those with User= set) this ensures that access rights to the service cgroup is granted to the user in question, to create further subgroups. Note that this only applies to the name=systemd hierarchy though, as access to other controllers is not safe for unpriviliged processes. Delegate=yes should be set for container scopes where a systemd instance inside the container shall manage the hierarchies below its own cgroup and have access to all controllers. Delegate=yes should also be set for user@.service, so that systemd --user can run, controlling its own cgroup tree. This commit changes machined, systemd-nspawn@.service and user@.service to set this boolean, in order to ensure that container management will just work, and the user systemd instance can run fine.
2014-11-05 17:57:23 +01:00
SD_BUS_PROPERTY("Delegate", "b", bus_property_get_bool, offsetof(CGroupContext, delegate), 0),
core: rework the Delegate= unit file setting to take a list of controller names Previously it was not possible to select which controllers to enable for a unit where Delegate=yes was set, as all controllers were enabled. With this change, this is made configurable, and thus delegation units can pick specifically what they want to manage themselves, and what they don't care about.
2017-11-09 15:29:34 +01:00
SD_BUS_PROPERTY("DelegateControllers", "as", property_get_delegate_controllers, 0, 0),
core: the cgroup properties are not actually const
2013-12-22 19:44:04 +01:00
SD_BUS_PROPERTY("CPUAccounting", "b", bus_property_get_bool, offsetof(CGroupContext, cpu_accounting), 0),
core: add cgroup CPU controller support on the unified hierarchy Unfortunately, due to the disagreements in the kernel development community, CPU controller cgroup v2 support has not been merged and enabling it requires applying two small out-of-tree kernel patches. The situation is explained in the following documentation. https://git.kernel.org/cgit/linux/kernel/git/tj/cgroup.git/tree/Documentation/cgroup-v2-cpu.txt?h=cgroup-v2-cpu While it isn't clear what will happen with CPU controller cgroup v2 support, there are critical features which are possible only on cgroup v2 such as buffered write control making cgroup v2 essential for a lot of workloads. This commit implements systemd CPU controller support on the unified hierarchy so that users who choose to deploy CPU controller cgroup v2 support can easily take advantage of it. On the unified hierarchy, "cpu.weight" knob replaces "cpu.shares" and "cpu.max" replaces "cpu.cfs_period_us" and "cpu.cfs_quota_us". [Startup]CPUWeight config options are added with the usual compat translation. CPU quota settings remain unchanged and apply to both legacy and unified hierarchies. v2: - Error in man page corrected. - CPU config application in cgroup_context_apply() refactored. - CPU accounting now works on unified hierarchy.
2016-08-07 15:45:39 +02:00
SD_BUS_PROPERTY("CPUWeight", "t", NULL, offsetof(CGroupContext, cpu_weight), 0),
SD_BUS_PROPERTY("StartupCPUWeight", "t", NULL, offsetof(CGroupContext, startup_cpu_weight), 0),
core: refactor cpu shares/blockio weight cgroup logic Let's stop using the "unsigned long" type for weights/shares, and let's just use uint64_t for this, as that's what we expose on the bus. Unify parsers, and always validate the range for these fields. Correct the default blockio weight to 500, since that's what the kernel actually uses. When parsing the weight/shares settings from unit files accept the empty string as a way to reset the weight/shares value. When getting it via the bus, uniformly map (uint64_t) -1 to unset. Open up StartupCPUShares= and StartupBlockIOWeight= to transient units.
2015-09-11 16:48:24 +02:00
SD_BUS_PROPERTY("CPUShares", "t", NULL, offsetof(CGroupContext, cpu_shares), 0),
SD_BUS_PROPERTY("StartupCPUShares", "t", NULL, offsetof(CGroupContext, startup_cpu_shares), 0),
core/dbus: simplify handling of CPUQuotaPerSecUSec No functional change intended.
2014-09-29 05:42:33 +02:00
SD_BUS_PROPERTY("CPUQuotaPerSecUSec", "t", bus_property_get_usec, offsetof(CGroupContext, cpu_quota_per_sec_usec), 0),
core: add CPUQuotaPeriodSec= This new setting allows configuration of CFS period on the CPU cgroup, instead of using a hardcoded default of 100ms. Tested: - Legacy cgroup + Unified cgroup - systemctl set-property - systemctl show - Confirmed that the cgroup settings (such as cpu.cfs_period_ns) were set appropriately, including updating the CPU quota (cpu.cfs_quota_ns) when CPUQuotaPeriodSec= is updated. - Checked that clamping works properly when either period or (quota * period) are below the resolution of 1ms, or if period is above the max of 1s.
2018-11-02 17:21:57 +01:00
SD_BUS_PROPERTY("CPUQuotaPeriodUSec", "t", bus_property_get_usec, offsetof(CGroupContext, cpu_quota_period_usec), 0),
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
SD_BUS_PROPERTY("IOAccounting", "b", bus_property_get_bool, offsetof(CGroupContext, io_accounting), 0),
SD_BUS_PROPERTY("IOWeight", "t", NULL, offsetof(CGroupContext, io_weight), 0),
SD_BUS_PROPERTY("StartupIOWeight", "t", NULL, offsetof(CGroupContext, startup_io_weight), 0),
SD_BUS_PROPERTY("IODeviceWeight", "a(st)", property_get_io_device_weight, 0, 0),
SD_BUS_PROPERTY("IOReadBandwidthMax", "a(st)", property_get_io_device_limits, 0, 0),
SD_BUS_PROPERTY("IOWriteBandwidthMax", "a(st)", property_get_io_device_limits, 0, 0),
core: add support for IOReadIOPSMax and IOWriteIOPSMax cgroup IO controller supports maximum limits for both bandwidth and IOPS but systemd resource control currently only supports bandwidth limits. This patch adds support for IOReadIOPSMax and IOWriteIOPSMax when unified cgroup hierarchy is in use. It isn't difficult to also add BlockIOReadIOPS and BlockIOWriteIOPS for legacy hierarchies but IO control on legacy hierarchies is half-broken anyway, so let's leave it alone for now.
2016-05-18 22:50:56 +02:00
SD_BUS_PROPERTY("IOReadIOPSMax", "a(st)", property_get_io_device_limits, 0, 0),
SD_BUS_PROPERTY("IOWriteIOPSMax", "a(st)", property_get_io_device_limits, 0, 0),
core: add IODeviceLatencyTargetSec This adds support for the following proposed latency based IO control mechanism. https://lkml.org/lkml/2018/6/5/428
2018-06-13 23:16:35 +02:00
SD_BUS_PROPERTY("IODeviceLatencyTargetUSec", "a(st)", property_get_io_device_latency, 0, 0),
core: the cgroup properties are not actually const
2013-12-22 19:44:04 +01:00
SD_BUS_PROPERTY("BlockIOAccounting", "b", bus_property_get_bool, offsetof(CGroupContext, blockio_accounting), 0),
core: refactor cpu shares/blockio weight cgroup logic Let's stop using the "unsigned long" type for weights/shares, and let's just use uint64_t for this, as that's what we expose on the bus. Unify parsers, and always validate the range for these fields. Correct the default blockio weight to 500, since that's what the kernel actually uses. When parsing the weight/shares settings from unit files accept the empty string as a way to reset the weight/shares value. When getting it via the bus, uniformly map (uint64_t) -1 to unset. Open up StartupCPUShares= and StartupBlockIOWeight= to transient units.
2015-09-11 16:48:24 +02:00
SD_BUS_PROPERTY("BlockIOWeight", "t", NULL, offsetof(CGroupContext, blockio_weight), 0),
SD_BUS_PROPERTY("StartupBlockIOWeight", "t", NULL, offsetof(CGroupContext, startup_blockio_weight), 0),
core: the cgroup properties are not actually const
2013-12-22 19:44:04 +01:00
SD_BUS_PROPERTY("BlockIODeviceWeight", "a(st)", property_get_blockio_device_weight, 0, 0),
SD_BUS_PROPERTY("BlockIOReadBandwidth", "a(st)", property_get_blockio_device_bandwidths, 0, 0),
SD_BUS_PROPERTY("BlockIOWriteBandwidth", "a(st)", property_get_blockio_device_bandwidths, 0, 0),
SD_BUS_PROPERTY("MemoryAccounting", "b", bus_property_get_bool, offsetof(CGroupContext, memory_accounting), 0),
core: add MemoryMin The kernel added support for a new cgroup memory controller knob memory.min in bf8d5d52ffe8 ("memcg: introduce memory.min") which was merged during v4.18 merge window. Add MemoryMin to support memory.min.
2018-06-09 02:33:14 +02:00
SD_BUS_PROPERTY("MemoryMin", "t", NULL, offsetof(CGroupContext, memory_min), 0),
core: add cgroup memory controller support on the unified hierarchy (#3315) On the unified hierarchy, memory controller implements three control knobs - low, high and max which enables more useable and versatile control over memory usage. This patch implements support for the three control knobs. * MemoryLow, MemoryHigh and MemoryMax are added for memory.low, memory.high and memory.max, respectively. * As all absolute limits on the unified hierarchy use "max" for no limit, make memory limit parse functions accept "max" in addition to "infinity" and document "max" for the new knobs. * Implement compatibility translation between MemoryMax and MemoryLimit. v2: - Fixed missing else's in config_parse_memory_limit(). - Fixed missing newline when writing out drop-ins. - Coding style updates to use "val > 0" instead of "val". - Minor updates to documentation.
2016-05-27 18:10:18 +02:00
SD_BUS_PROPERTY("MemoryLow", "t", NULL, offsetof(CGroupContext, memory_low), 0),
SD_BUS_PROPERTY("MemoryHigh", "t", NULL, offsetof(CGroupContext, memory_high), 0),
SD_BUS_PROPERTY("MemoryMax", "t", NULL, offsetof(CGroupContext, memory_max), 0),
core: introduce MemorySwapMax= Similar to MemoryMax=, MemorySwapMax= limits swap usage. This controls controls "memory.swap.max" attribute in unified cgroup.
2016-07-04 09:03:54 +02:00
SD_BUS_PROPERTY("MemorySwapMax", "t", NULL, offsetof(CGroupContext, memory_swap_max), 0),
core: the cgroup properties are not actually const
2013-12-22 19:44:04 +01:00
SD_BUS_PROPERTY("MemoryLimit", "t", NULL, offsetof(CGroupContext, memory_limit), 0),
SD_BUS_PROPERTY("DevicePolicy", "s", property_get_cgroup_device_policy, offsetof(CGroupContext, device_policy), 0),
SD_BUS_PROPERTY("DeviceAllow", "a(ss)", property_get_device_allow, 0, 0),
core: add support for the "pids" cgroup controller This adds support for the new "pids" cgroup controller of 4.3 kernels. It allows accounting the number of tasks in a cgroup and enforcing limits on it. This adds two new setting TasksAccounting= and TasksMax= to each unit, as well as a gloabl option DefaultTasksAccounting=. This also updated "cgtop" to optionally make use of the new kernel-provided accounting. systemctl has been updated to show the number of tasks for each service if it is available. This patch also adds correct support for undoing memory limits for units using a MemoryLimit=infinity syntax. We do the same for TasksMax= now and hence keep things in sync here.
2015-09-10 12:32:16 +02:00
SD_BUS_PROPERTY("TasksAccounting", "b", bus_property_get_bool, offsetof(CGroupContext, tasks_accounting), 0),
SD_BUS_PROPERTY("TasksMax", "t", NULL, offsetof(CGroupContext, tasks_max), 0),
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
SD_BUS_PROPERTY("IPAccounting", "b", bus_property_get_bool, offsetof(CGroupContext, ip_accounting), 0),
SD_BUS_PROPERTY("IPAddressAllow", "a(iayu)", property_get_ip_address_access, offsetof(CGroupContext, ip_address_allow), 0),
SD_BUS_PROPERTY("IPAddressDeny", "a(iayu)", property_get_ip_address_access, offsetof(CGroupContext, ip_address_deny), 0),
cgroup: Add DisableControllers= directive to disable controller in subtree Some controllers (like the CPU controller) have a performance cost that is non-trivial on certain workloads. While this can be mitigated and improved to an extent, there will for some controllers always be some overheads associated with the benefits gained from the controller. Inside Facebook, the fix applied has been to disable the CPU controller forcibly with `cgroup_disable=cpu` on the kernel command line. This presents a problem: to disable or reenable the controller, a reboot is required, but this is quite cumbersome and slow to do for many thousands of machines, especially machines where disabling/enabling a stateful service on a machine is a matter of several minutes. Currently systemd provides some configuration knobs for these in the form of `[Default]CPUAccounting`, `[Default]MemoryAccounting`, and the like. The limitation of these is that Default*Accounting is overrideable by individual services, of which any one could decide to reenable a controller within the hierarchy at any point just by using a controller feature implicitly (eg. `CPUWeight`), even if the use of that CPU feature could just be opportunistic. Since many services are provided by the distribution, or by upstream teams at a particular organisation, it's not a sustainable solution to simply try to find and remove offending directives from these units. This commit presents a more direct solution -- a DisableControllers= directive that forcibly disallows a controller from being enabled within a subtree.
2018-12-03 15:38:06 +01:00
SD_BUS_PROPERTY("DisableControllers", "as", property_get_cgroup_mask, offsetof(CGroupContext, disable_controllers), 0),
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
SD_BUS_VTABLE_END
core: general cgroup rework Replace the very generic cgroup hookup with a much simpler one. With this change only the high-level cgroup settings remain, the ability to set arbitrary cgroup attributes is removed, so is support for adding units to arbitrary cgroup controllers or setting arbitrary paths for them (especially paths that are different for the various controllers). This also introduces a new -.slice root slice, that is the parent of system.slice and friends. This enables easy admin configuration of root-level cgrouo properties. This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in /dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is turned off by DevicePolicy=).
2013-06-27 04:14:27 +02:00
};
dbus: add infrastructure for changing multiple properties at once on units and hook some cgroup attributes up to it This introduces two bus calls to make runtime changes to selected bus properties, optionally with persistence. This currently hooks this up only for three cgroup atributes, but this brings the infrastructure to add more changable attributes. This allows setting multiple attributes at once, and takes an array rather than a dictionary of properties, in order to implement simple resetting of lists using the same approach as when they are sourced from unit files. This means, that list properties are appended to by this call, unless they are first reset via assigning the empty list.
2013-06-27 21:14:56 +02:00
core: introduce new Delegate=yes/no property controlling creation of cgroup subhierarchies For priviliged units this resource control property ensures that the processes have all controllers systemd manages enabled. For unpriviliged services (those with User= set) this ensures that access rights to the service cgroup is granted to the user in question, to create further subgroups. Note that this only applies to the name=systemd hierarchy though, as access to other controllers is not safe for unpriviliged processes. Delegate=yes should be set for container scopes where a systemd instance inside the container shall manage the hierarchies below its own cgroup and have access to all controllers. Delegate=yes should also be set for user@.service, so that systemd --user can run, controlling its own cgroup tree. This commit changes machined, systemd-nspawn@.service and user@.service to set this boolean, in order to ensure that container management will just work, and the user systemd instance can run fine.
2014-11-05 17:57:23 +01:00
static int bus_cgroup_set_transient_property(
Unit *u,
CGroupContext *c,
const char *name,
sd_bus_message *message,
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
UnitWriteFlags flags,
core: introduce new Delegate=yes/no property controlling creation of cgroup subhierarchies For priviliged units this resource control property ensures that the processes have all controllers systemd manages enabled. For unpriviliged services (those with User= set) this ensures that access rights to the service cgroup is granted to the user in question, to create further subgroups. Note that this only applies to the name=systemd hierarchy though, as access to other controllers is not safe for unpriviliged processes. Delegate=yes should be set for container scopes where a systemd instance inside the container shall manage the hierarchies below its own cgroup and have access to all controllers. Delegate=yes should also be set for user@.service, so that systemd --user can run, controlling its own cgroup tree. This commit changes machined, systemd-nspawn@.service and user@.service to set this boolean, in order to ensure that container management will just work, and the user systemd instance can run fine.
2014-11-05 17:57:23 +01:00
sd_bus_error *error) {
int r;
assert(u);
assert(c);
assert(name);
assert(message);
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
flags |= UNIT_PRIVATE;
core: introduce new Delegate=yes/no property controlling creation of cgroup subhierarchies For priviliged units this resource control property ensures that the processes have all controllers systemd manages enabled. For unpriviliged services (those with User= set) this ensures that access rights to the service cgroup is granted to the user in question, to create further subgroups. Note that this only applies to the name=systemd hierarchy though, as access to other controllers is not safe for unpriviliged processes. Delegate=yes should be set for container scopes where a systemd instance inside the container shall manage the hierarchies below its own cgroup and have access to all controllers. Delegate=yes should also be set for user@.service, so that systemd --user can run, controlling its own cgroup tree. This commit changes machined, systemd-nspawn@.service and user@.service to set this boolean, in order to ensure that container management will just work, and the user systemd instance can run fine.
2014-11-05 17:57:23 +01:00
if (streq(name, "Delegate")) {
int b;
cgroup: add a new "can_delegate" flag to the unit vtable, and set it for scope and service units only Currently we allowed delegation for alluntis with cgroup backing except for slices. Let's make this a bit more strict for now, and only allow this in service and scope units. Let's also add a generic accessor unit_cgroup_delegate() for checking whether a unit has delegation turned on that checks the new bool first. Also, when doing transient units, let's explcitly refuse turning on delegation for unit types that don#t support it. This is mostly cosmetical as we wouldn't act on the delegation request anyway, but certainly helpful for debugging.
2018-02-06 11:57:35 +01:00
if (!UNIT_VTABLE(u)->can_delegate)
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Delegation not available for unit type");
core: introduce new Delegate=yes/no property controlling creation of cgroup subhierarchies For priviliged units this resource control property ensures that the processes have all controllers systemd manages enabled. For unpriviliged services (those with User= set) this ensures that access rights to the service cgroup is granted to the user in question, to create further subgroups. Note that this only applies to the name=systemd hierarchy though, as access to other controllers is not safe for unpriviliged processes. Delegate=yes should be set for container scopes where a systemd instance inside the container shall manage the hierarchies below its own cgroup and have access to all controllers. Delegate=yes should also be set for user@.service, so that systemd --user can run, controlling its own cgroup tree. This commit changes machined, systemd-nspawn@.service and user@.service to set this boolean, in order to ensure that container management will just work, and the user systemd instance can run fine.
2014-11-05 17:57:23 +01:00
r = sd_bus_message_read(message, "b", &b);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: introduce new Delegate=yes/no property controlling creation of cgroup subhierarchies For priviliged units this resource control property ensures that the processes have all controllers systemd manages enabled. For unpriviliged services (those with User= set) this ensures that access rights to the service cgroup is granted to the user in question, to create further subgroups. Note that this only applies to the name=systemd hierarchy though, as access to other controllers is not safe for unpriviliged processes. Delegate=yes should be set for container scopes where a systemd instance inside the container shall manage the hierarchies below its own cgroup and have access to all controllers. Delegate=yes should also be set for user@.service, so that systemd --user can run, controlling its own cgroup tree. This commit changes machined, systemd-nspawn@.service and user@.service to set this boolean, in order to ensure that container management will just work, and the user systemd instance can run fine.
2014-11-05 17:57:23 +01:00
c->delegate = b;
core: rework the Delegate= unit file setting to take a list of controller names Previously it was not possible to select which controllers to enable for a unit where Delegate=yes was set, as all controllers were enabled. With this change, this is made configurable, and thus delegation units can pick specifically what they want to manage themselves, and what they don't care about.
2017-11-09 15:29:34 +01:00
c->delegate_controllers = b ? _CGROUP_MASK_ALL : 0;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
unit_write_settingf(u, flags, name, "Delegate=%s", yes_no(b));
core: introduce new Delegate=yes/no property controlling creation of cgroup subhierarchies For priviliged units this resource control property ensures that the processes have all controllers systemd manages enabled. For unpriviliged services (those with User= set) this ensures that access rights to the service cgroup is granted to the user in question, to create further subgroups. Note that this only applies to the name=systemd hierarchy though, as access to other controllers is not safe for unpriviliged processes. Delegate=yes should be set for container scopes where a systemd instance inside the container shall manage the hierarchies below its own cgroup and have access to all controllers. Delegate=yes should also be set for user@.service, so that systemd --user can run, controlling its own cgroup tree. This commit changes machined, systemd-nspawn@.service and user@.service to set this boolean, in order to ensure that container management will just work, and the user systemd instance can run fine.
2014-11-05 17:57:23 +01:00
}
core: rework the Delegate= unit file setting to take a list of controller names Previously it was not possible to select which controllers to enable for a unit where Delegate=yes was set, as all controllers were enabled. With this change, this is made configurable, and thus delegation units can pick specifically what they want to manage themselves, and what they don't care about.
2017-11-09 15:29:34 +01:00
return 1;
} else if (streq(name, "DelegateControllers")) {
CGroupMask mask = 0;
cgroup: add a new "can_delegate" flag to the unit vtable, and set it for scope and service units only Currently we allowed delegation for alluntis with cgroup backing except for slices. Let's make this a bit more strict for now, and only allow this in service and scope units. Let's also add a generic accessor unit_cgroup_delegate() for checking whether a unit has delegation turned on that checks the new bool first. Also, when doing transient units, let's explcitly refuse turning on delegation for unit types that don#t support it. This is mostly cosmetical as we wouldn't act on the delegation request anyway, but certainly helpful for debugging.
2018-02-06 11:57:35 +01:00
if (!UNIT_VTABLE(u)->can_delegate)
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Delegation not available for unit type");
core: rework the Delegate= unit file setting to take a list of controller names Previously it was not possible to select which controllers to enable for a unit where Delegate=yes was set, as all controllers were enabled. With this change, this is made configurable, and thus delegation units can pick specifically what they want to manage themselves, and what they don't care about.
2017-11-09 15:29:34 +01:00
r = sd_bus_message_enter_container(message, 'a', "s");
if (r < 0)
return r;
for (;;) {
CGroupController cc;
const char *t;
r = sd_bus_message_read(message, "s", &t);
if (r < 0)
return r;
if (r == 0)
break;
cc = cgroup_controller_from_string(t);
if (cc < 0)
src/core/dbus-cgroup.c: fix typo contoller -> controller (#8717) Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
2018-04-14 11:06:11 +02:00
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Unknown cgroup controller '%s'", t);
core: rework the Delegate= unit file setting to take a list of controller names Previously it was not possible to select which controllers to enable for a unit where Delegate=yes was set, as all controllers were enabled. With this change, this is made configurable, and thus delegation units can pick specifically what they want to manage themselves, and what they don't care about.
2017-11-09 15:29:34 +01:00
mask |= CGROUP_CONTROLLER_TO_MASK(cc);
}
r = sd_bus_message_exit_container(message);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: rework the Delegate= unit file setting to take a list of controller names Previously it was not possible to select which controllers to enable for a unit where Delegate=yes was set, as all controllers were enabled. With this change, this is made configurable, and thus delegation units can pick specifically what they want to manage themselves, and what they don't care about.
2017-11-09 15:29:34 +01:00
_cleanup_free_ char *t = NULL;
r = cg_mask_to_string(mask, &t);
if (r < 0)
return r;
c->delegate = true;
core/cgroup: assigning empty string to Delegate= resets list of controllers (#7336) Before this, assigning empty string to Delegate= makes no change to the controller list. This is inconsistent to the other options that take list of strings. After this, when empty string is assigned to Delegate=, the list of controllers is reset. Such behavior is consistent to other options and useful for drop-in configs. Closes #7334.
2017-11-17 10:04:25 +01:00
if (mask == 0)
c->delegate_controllers = 0;
else
c->delegate_controllers |= mask;
core: rework the Delegate= unit file setting to take a list of controller names Previously it was not possible to select which controllers to enable for a unit where Delegate=yes was set, as all controllers were enabled. With this change, this is made configurable, and thus delegation units can pick specifically what they want to manage themselves, and what they don't care about.
2017-11-09 15:29:34 +01:00
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
unit_write_settingf(u, flags, name, "Delegate=%s", strempty(t));
core: rework the Delegate= unit file setting to take a list of controller names Previously it was not possible to select which controllers to enable for a unit where Delegate=yes was set, as all controllers were enabled. With this change, this is made configurable, and thus delegation units can pick specifically what they want to manage themselves, and what they don't care about.
2017-11-09 15:29:34 +01:00
}
core: introduce new Delegate=yes/no property controlling creation of cgroup subhierarchies For priviliged units this resource control property ensures that the processes have all controllers systemd manages enabled. For unpriviliged services (those with User= set) this ensures that access rights to the service cgroup is granted to the user in question, to create further subgroups. Note that this only applies to the name=systemd hierarchy though, as access to other controllers is not safe for unpriviliged processes. Delegate=yes should be set for container scopes where a systemd instance inside the container shall manage the hierarchies below its own cgroup and have access to all controllers. Delegate=yes should also be set for user@.service, so that systemd --user can run, controlling its own cgroup tree. This commit changes machined, systemd-nspawn@.service and user@.service to set this boolean, in order to ensure that container management will just work, and the user systemd instance can run fine.
2014-11-05 17:57:23 +01:00
return 1;
}
return 0;
}
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
static int bus_cgroup_set_boolean(
Unit *u,
const char *name,
bool *p,
CGroupMask mask,
sd_bus_message *message,
UnitWriteFlags flags,
sd_bus_error *error) {
int b, r;
assert(p);
r = sd_bus_message_read(message, "b", &b);
if (r < 0)
return r;
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
*p = b;
unit_invalidate_cgroup(u, mask);
unit_write_settingf(u, flags, name, "%s=%s", name, yes_no(b));
}
return 1;
}
core/cgroup: accepts MemorySwapMax=0 (#8366) Also, this moves two macros from dbus-util.h to dbus-cgroup.c, as they are only used in dbus-cgroup.c. Fixes #8363.
2018-03-09 11:34:50 +01:00
#define BUS_DEFINE_SET_CGROUP_WEIGHT(function, mask, check, val) \
static int bus_cgroup_set_##function( \
Unit *u, \
const char *name, \
uint64_t *p, \
sd_bus_message *message, \
UnitWriteFlags flags, \
sd_bus_error *error) { \
\
uint64_t v; \
int r; \
\
assert(p); \
\
r = sd_bus_message_read(message, "t", &v); \
if (r < 0) \
return r; \
\
if (!check(v)) \
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, \
"Value specified in %s is out of range", name); \
\
if (!UNIT_WRITE_FLAGS_NOOP(flags)) { \
*p = v; \
unit_invalidate_cgroup(u, (mask)); \
\
if (v == (val)) \
unit_write_settingf(u, flags, name, \
"%s=", name); \
else \
unit_write_settingf(u, flags, name, \
"%s=%" PRIu64, name, v); \
} \
\
return 1; \
}
#define BUS_DEFINE_SET_CGROUP_LIMIT(function, mask, scale, minimum) \
static int bus_cgroup_set_##function( \
Unit *u, \
const char *name, \
uint64_t *p, \
sd_bus_message *message, \
UnitWriteFlags flags, \
sd_bus_error *error) { \
\
uint64_t v; \
int r; \
\
assert(p); \
\
r = sd_bus_message_read(message, "t", &v); \
if (r < 0) \
return r; \
\
if (v < minimum) \
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, \
"Value specified in %s is out of range", name); \
\
if (!UNIT_WRITE_FLAGS_NOOP(flags)) { \
*p = v; \
unit_invalidate_cgroup(u, (mask)); \
\
if (v == CGROUP_LIMIT_MAX) \
unit_write_settingf(u, flags, name, \
"%s=infinity", name); \
else \
unit_write_settingf(u, flags, name, \
"%s=%" PRIu64, name, v); \
} \
\
return 1; \
} \
static int bus_cgroup_set_##function##_scale( \
Unit *u, \
const char *name, \
uint64_t *p, \
sd_bus_message *message, \
UnitWriteFlags flags, \
sd_bus_error *error) { \
\
uint64_t v; \
uint32_t raw; \
int r; \
\
assert(p); \
\
r = sd_bus_message_read(message, "u", &raw); \
if (r < 0) \
return r; \
\
v = scale(raw, UINT32_MAX); \
if (v < minimum || v >= UINT64_MAX) \
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, \
"Value specified in %s is out of range", name); \
\
if (!UNIT_WRITE_FLAGS_NOOP(flags)) { \
const char *e; \
\
*p = v; \
unit_invalidate_cgroup(u, (mask)); \
\
/* Chop off suffix */ \
assert_se(e = endswith(name, "Scale")); \
name = strndupa(name, e - name); \
\
unit_write_settingf(u, flags, name, "%s=%" PRIu32 "%%", name, \
(uint32_t) (DIV_ROUND_UP((uint64_t) raw * 100U, (uint64_t) UINT32_MAX))); \
} \
\
return 1; \
}
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wtype-limits"
BUS_DEFINE_SET_CGROUP_WEIGHT(cpu_weight, CGROUP_MASK_CPU, CGROUP_WEIGHT_IS_OK, CGROUP_WEIGHT_INVALID);
BUS_DEFINE_SET_CGROUP_WEIGHT(cpu_shares, CGROUP_MASK_CPU, CGROUP_CPU_SHARES_IS_OK, CGROUP_CPU_SHARES_INVALID);
BUS_DEFINE_SET_CGROUP_WEIGHT(io_weight, CGROUP_MASK_IO, CGROUP_WEIGHT_IS_OK, CGROUP_WEIGHT_INVALID);
BUS_DEFINE_SET_CGROUP_WEIGHT(blockio_weight, CGROUP_MASK_BLKIO, CGROUP_BLKIO_WEIGHT_IS_OK, CGROUP_BLKIO_WEIGHT_INVALID);
BUS_DEFINE_SET_CGROUP_LIMIT(memory, CGROUP_MASK_MEMORY, physical_memory_scale, 1);
BUS_DEFINE_SET_CGROUP_LIMIT(swap, CGROUP_MASK_MEMORY, physical_memory_scale, 0);
BUS_DEFINE_SET_CGROUP_LIMIT(tasks_max, CGROUP_MASK_PIDS, system_tasks_max_scale, 1);
#pragma GCC diagnostic pop
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
dbus: add infrastructure for changing multiple properties at once on units and hook some cgroup attributes up to it This introduces two bus calls to make runtime changes to selected bus properties, optionally with persistence. This currently hooks this up only for three cgroup atributes, but this brings the infrastructure to add more changable attributes. This allows setting multiple attributes at once, and takes an array rather than a dictionary of properties, in order to implement simple resetting of lists using the same approach as when they are sourced from unit files. This means, that list properties are appended to by this call, unless they are first reset via assigning the empty list.
2013-06-27 21:14:56 +02:00
int bus_cgroup_set_property(
Unit *u,
CGroupContext *c,
const char *name,
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
sd_bus_message *message,
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
UnitWriteFlags flags,
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
sd_bus_error *error) {
core: introduce CGroupIOLimitType enums Currently, there are two cgroup IO limits, bandwidth max for read and write, and they are hard-coded in various places. This is fine for two limits but IO is expected to grow more limits - low, high and max limits for bandwidth and IOPS - and hard-coding each limit won't make sense. This patch replaces hard-coded limits with an array indexed by CGroupIOLimitType and accompanying string and default value tables so that new limits can be added trivially.
2016-05-18 22:50:56 +02:00
CGroupIOLimitType iol_type;
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
int r;
dbus: add infrastructure for changing multiple properties at once on units and hook some cgroup attributes up to it This introduces two bus calls to make runtime changes to selected bus properties, optionally with persistence. This currently hooks this up only for three cgroup atributes, but this brings the infrastructure to add more changable attributes. This allows setting multiple attributes at once, and takes an array rather than a dictionary of properties, in order to implement simple resetting of lists using the same approach as when they are sourced from unit files. This means, that list properties are appended to by this call, unless they are first reset via assigning the empty list.
2013-06-27 21:14:56 +02:00
assert(u);
assert(c);
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
assert(name);
assert(message);
dbus: add infrastructure for changing multiple properties at once on units and hook some cgroup attributes up to it This introduces two bus calls to make runtime changes to selected bus properties, optionally with persistence. This currently hooks this up only for three cgroup atributes, but this brings the infrastructure to add more changable attributes. This allows setting multiple attributes at once, and takes an array rather than a dictionary of properties, in order to implement simple resetting of lists using the same approach as when they are sourced from unit files. This means, that list properties are appended to by this call, unless they are first reset via assigning the empty list.
2013-06-27 21:14:56 +02:00
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
flags |= UNIT_PRIVATE;
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "CPUAccounting"))
cgroup v2: Don't require CPU controller for CPU accounting in 4.15+ systemd only uses functions that are as of Linux 4.15+ provided externally to the CPU controller (currently usage_usec), so if we have a new enough kernel, we don't need to set CGROUP_MASK_CPU for CPUAccounting=true as the CPU controller does not need to necessarily be enabled in this case. Part of this patch is modelled on an earlier patch by Ryutaroh Matsumoto (see PR #9665).
2018-11-17 12:19:07 +01:00
return bus_cgroup_set_boolean(u, name, &c->cpu_accounting, get_cpu_accounting_mask(), message, flags, error);
dbus: add infrastructure for changing multiple properties at once on units and hook some cgroup attributes up to it This introduces two bus calls to make runtime changes to selected bus properties, optionally with persistence. This currently hooks this up only for three cgroup atributes, but this brings the infrastructure to add more changable attributes. This allows setting multiple attributes at once, and takes an array rather than a dictionary of properties, in order to implement simple resetting of lists using the same approach as when they are sourced from unit files. This means, that list properties are appended to by this call, unless they are first reset via assigning the empty list.
2013-06-27 21:14:56 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "CPUWeight"))
return bus_cgroup_set_cpu_weight(u, name, &c->cpu_weight, message, flags, error);
dbus: add infrastructure for changing multiple properties at once on units and hook some cgroup attributes up to it This introduces two bus calls to make runtime changes to selected bus properties, optionally with persistence. This currently hooks this up only for three cgroup atributes, but this brings the infrastructure to add more changable attributes. This allows setting multiple attributes at once, and takes an array rather than a dictionary of properties, in order to implement simple resetting of lists using the same approach as when they are sourced from unit files. This means, that list properties are appended to by this call, unless they are first reset via assigning the empty list.
2013-06-27 21:14:56 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "StartupCPUWeight"))
return bus_cgroup_set_cpu_weight(u, name, &c->startup_cpu_weight, message, flags, error);
dbus: make more cgroup attributes runtime settable
2013-06-27 21:50:35 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "CPUShares"))
return bus_cgroup_set_cpu_shares(u, name, &c->cpu_shares, message, flags, error);
dbus: make more cgroup attributes runtime settable
2013-06-27 21:50:35 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "StartupCPUShares"))
return bus_cgroup_set_cpu_shares(u, name, &c->startup_cpu_shares, message, flags, error);
core: add cgroup CPU controller support on the unified hierarchy Unfortunately, due to the disagreements in the kernel development community, CPU controller cgroup v2 support has not been merged and enabling it requires applying two small out-of-tree kernel patches. The situation is explained in the following documentation. https://git.kernel.org/cgit/linux/kernel/git/tj/cgroup.git/tree/Documentation/cgroup-v2-cpu.txt?h=cgroup-v2-cpu While it isn't clear what will happen with CPU controller cgroup v2 support, there are critical features which are possible only on cgroup v2 such as buffered write control making cgroup v2 essential for a lot of workloads. This commit implements systemd CPU controller support on the unified hierarchy so that users who choose to deploy CPU controller cgroup v2 support can easily take advantage of it. On the unified hierarchy, "cpu.weight" knob replaces "cpu.shares" and "cpu.max" replaces "cpu.cfs_period_us" and "cpu.cfs_quota_us". [Startup]CPUWeight config options are added with the usual compat translation. CPU quota settings remain unchanged and apply to both legacy and unified hierarchies. v2: - Error in man page corrected. - CPU config application in cgroup_context_apply() refactored. - CPU accounting now works on unified hierarchy.
2016-08-07 15:45:39 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "IOAccounting"))
return bus_cgroup_set_boolean(u, name, &c->io_accounting, CGROUP_MASK_IO, message, flags, error);
core: add cgroup CPU controller support on the unified hierarchy Unfortunately, due to the disagreements in the kernel development community, CPU controller cgroup v2 support has not been merged and enabling it requires applying two small out-of-tree kernel patches. The situation is explained in the following documentation. https://git.kernel.org/cgit/linux/kernel/git/tj/cgroup.git/tree/Documentation/cgroup-v2-cpu.txt?h=cgroup-v2-cpu While it isn't clear what will happen with CPU controller cgroup v2 support, there are critical features which are possible only on cgroup v2 such as buffered write control making cgroup v2 essential for a lot of workloads. This commit implements systemd CPU controller support on the unified hierarchy so that users who choose to deploy CPU controller cgroup v2 support can easily take advantage of it. On the unified hierarchy, "cpu.weight" knob replaces "cpu.shares" and "cpu.max" replaces "cpu.cfs_period_us" and "cpu.cfs_quota_us". [Startup]CPUWeight config options are added with the usual compat translation. CPU quota settings remain unchanged and apply to both legacy and unified hierarchies. v2: - Error in man page corrected. - CPU config application in cgroup_context_apply() refactored. - CPU accounting now works on unified hierarchy.
2016-08-07 15:45:39 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "IOWeight"))
return bus_cgroup_set_io_weight(u, name, &c->io_weight, message, flags, error);
core: add cgroup CPU controller support on the unified hierarchy Unfortunately, due to the disagreements in the kernel development community, CPU controller cgroup v2 support has not been merged and enabling it requires applying two small out-of-tree kernel patches. The situation is explained in the following documentation. https://git.kernel.org/cgit/linux/kernel/git/tj/cgroup.git/tree/Documentation/cgroup-v2-cpu.txt?h=cgroup-v2-cpu While it isn't clear what will happen with CPU controller cgroup v2 support, there are critical features which are possible only on cgroup v2 such as buffered write control making cgroup v2 essential for a lot of workloads. This commit implements systemd CPU controller support on the unified hierarchy so that users who choose to deploy CPU controller cgroup v2 support can easily take advantage of it. On the unified hierarchy, "cpu.weight" knob replaces "cpu.shares" and "cpu.max" replaces "cpu.cfs_period_us" and "cpu.cfs_quota_us". [Startup]CPUWeight config options are added with the usual compat translation. CPU quota settings remain unchanged and apply to both legacy and unified hierarchies. v2: - Error in man page corrected. - CPU config application in cgroup_context_apply() refactored. - CPU accounting now works on unified hierarchy.
2016-08-07 15:45:39 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "StartupIOWeight"))
return bus_cgroup_set_io_weight(u, name, &c->startup_io_weight, message, flags, error);
core: add cgroup CPU controller support on the unified hierarchy Unfortunately, due to the disagreements in the kernel development community, CPU controller cgroup v2 support has not been merged and enabling it requires applying two small out-of-tree kernel patches. The situation is explained in the following documentation. https://git.kernel.org/cgit/linux/kernel/git/tj/cgroup.git/tree/Documentation/cgroup-v2-cpu.txt?h=cgroup-v2-cpu While it isn't clear what will happen with CPU controller cgroup v2 support, there are critical features which are possible only on cgroup v2 such as buffered write control making cgroup v2 essential for a lot of workloads. This commit implements systemd CPU controller support on the unified hierarchy so that users who choose to deploy CPU controller cgroup v2 support can easily take advantage of it. On the unified hierarchy, "cpu.weight" knob replaces "cpu.shares" and "cpu.max" replaces "cpu.cfs_period_us" and "cpu.cfs_quota_us". [Startup]CPUWeight config options are added with the usual compat translation. CPU quota settings remain unchanged and apply to both legacy and unified hierarchies. v2: - Error in man page corrected. - CPU config application in cgroup_context_apply() refactored. - CPU accounting now works on unified hierarchy.
2016-08-07 15:45:39 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "BlockIOAccounting"))
return bus_cgroup_set_boolean(u, name, &c->blockio_accounting, CGROUP_MASK_BLKIO, message, flags, error);
core: add cgroup CPU controller support on the unified hierarchy Unfortunately, due to the disagreements in the kernel development community, CPU controller cgroup v2 support has not been merged and enabling it requires applying two small out-of-tree kernel patches. The situation is explained in the following documentation. https://git.kernel.org/cgit/linux/kernel/git/tj/cgroup.git/tree/Documentation/cgroup-v2-cpu.txt?h=cgroup-v2-cpu While it isn't clear what will happen with CPU controller cgroup v2 support, there are critical features which are possible only on cgroup v2 such as buffered write control making cgroup v2 essential for a lot of workloads. This commit implements systemd CPU controller support on the unified hierarchy so that users who choose to deploy CPU controller cgroup v2 support can easily take advantage of it. On the unified hierarchy, "cpu.weight" knob replaces "cpu.shares" and "cpu.max" replaces "cpu.cfs_period_us" and "cpu.cfs_quota_us". [Startup]CPUWeight config options are added with the usual compat translation. CPU quota settings remain unchanged and apply to both legacy and unified hierarchies. v2: - Error in man page corrected. - CPU config application in cgroup_context_apply() refactored. - CPU accounting now works on unified hierarchy.
2016-08-07 15:45:39 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "BlockIOWeight"))
return bus_cgroup_set_blockio_weight(u, name, &c->blockio_weight, message, flags, error);
core: add cgroup CPU controller support on the unified hierarchy Unfortunately, due to the disagreements in the kernel development community, CPU controller cgroup v2 support has not been merged and enabling it requires applying two small out-of-tree kernel patches. The situation is explained in the following documentation. https://git.kernel.org/cgit/linux/kernel/git/tj/cgroup.git/tree/Documentation/cgroup-v2-cpu.txt?h=cgroup-v2-cpu While it isn't clear what will happen with CPU controller cgroup v2 support, there are critical features which are possible only on cgroup v2 such as buffered write control making cgroup v2 essential for a lot of workloads. This commit implements systemd CPU controller support on the unified hierarchy so that users who choose to deploy CPU controller cgroup v2 support can easily take advantage of it. On the unified hierarchy, "cpu.weight" knob replaces "cpu.shares" and "cpu.max" replaces "cpu.cfs_period_us" and "cpu.cfs_quota_us". [Startup]CPUWeight config options are added with the usual compat translation. CPU quota settings remain unchanged and apply to both legacy and unified hierarchies. v2: - Error in man page corrected. - CPU config application in cgroup_context_apply() refactored. - CPU accounting now works on unified hierarchy.
2016-08-07 15:45:39 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "StartupBlockIOWeight"))
return bus_cgroup_set_blockio_weight(u, name, &c->startup_blockio_weight, message, flags, error);
core: add cgroup CPU controller support on the unified hierarchy Unfortunately, due to the disagreements in the kernel development community, CPU controller cgroup v2 support has not been merged and enabling it requires applying two small out-of-tree kernel patches. The situation is explained in the following documentation. https://git.kernel.org/cgit/linux/kernel/git/tj/cgroup.git/tree/Documentation/cgroup-v2-cpu.txt?h=cgroup-v2-cpu While it isn't clear what will happen with CPU controller cgroup v2 support, there are critical features which are possible only on cgroup v2 such as buffered write control making cgroup v2 essential for a lot of workloads. This commit implements systemd CPU controller support on the unified hierarchy so that users who choose to deploy CPU controller cgroup v2 support can easily take advantage of it. On the unified hierarchy, "cpu.weight" knob replaces "cpu.shares" and "cpu.max" replaces "cpu.cfs_period_us" and "cpu.cfs_quota_us". [Startup]CPUWeight config options are added with the usual compat translation. CPU quota settings remain unchanged and apply to both legacy and unified hierarchies. v2: - Error in man page corrected. - CPU config application in cgroup_context_apply() refactored. - CPU accounting now works on unified hierarchy.
2016-08-07 15:45:39 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "MemoryAccounting"))
return bus_cgroup_set_boolean(u, name, &c->memory_accounting, CGROUP_MASK_MEMORY, message, flags, error);
dbus: make more cgroup attributes runtime settable
2013-06-27 21:50:35 +02:00
core: add MemoryMin The kernel added support for a new cgroup memory controller knob memory.min in bf8d5d52ffe8 ("memcg: introduce memory.min") which was merged during v4.18 merge window. Add MemoryMin to support memory.min.
2018-06-09 02:33:14 +02:00
if (streq(name, "MemoryMin"))
return bus_cgroup_set_memory(u, name, &c->memory_min, message, flags, error);
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "MemoryLow"))
return bus_cgroup_set_memory(u, name, &c->memory_low, message, flags, error);
dbus: make more cgroup attributes runtime settable
2013-06-27 21:50:35 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "MemoryHigh"))
return bus_cgroup_set_memory(u, name, &c->memory_high, message, flags, error);
dbus: make more cgroup attributes runtime settable
2013-06-27 21:50:35 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "MemorySwapMax"))
core/cgroup: accepts MemorySwapMax=0 (#8366) Also, this moves two macros from dbus-util.h to dbus-cgroup.c, as they are only used in dbus-cgroup.c. Fixes #8363.
2018-03-09 11:34:50 +01:00
return bus_cgroup_set_swap(u, name, &c->memory_swap_max, message, flags, error);
core: add startup resource control option Similar to CPUShares= and BlockIOWeight= respectively. However only assign the specified weight during startup. Each control group attribute is re-assigned as weight by CPUShares=weight and BlockIOWeight=weight after startup. If not CPUShares= or BlockIOWeight= be specified, then the attribute is re-assigned to each default attribute value. (default cpu.shares=1024, blkio.weight=1000) If only CPUShares=weight or BlockIOWeight=weight be specified, then that implies StartupCPUShares=weight and StartupBlockIOWeight=weight.
2014-05-15 17:09:34 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "MemoryMax"))
return bus_cgroup_set_memory(u, name, &c->memory_max, message, flags, error);
core: refactor cpu shares/blockio weight cgroup logic Let's stop using the "unsigned long" type for weights/shares, and let's just use uint64_t for this, as that's what we expose on the bus. Unify parsers, and always validate the range for these fields. Correct the default blockio weight to 500, since that's what the kernel actually uses. When parsing the weight/shares settings from unit files accept the empty string as a way to reset the weight/shares value. When getting it via the bus, uniformly map (uint64_t) -1 to unset. Open up StartupCPUShares= and StartupBlockIOWeight= to transient units.
2015-09-11 16:48:24 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "MemoryLimit"))
return bus_cgroup_set_memory(u, name, &c->memory_limit, message, flags, error);
core: add startup resource control option Similar to CPUShares= and BlockIOWeight= respectively. However only assign the specified weight during startup. Each control group attribute is re-assigned as weight by CPUShares=weight and BlockIOWeight=weight after startup. If not CPUShares= or BlockIOWeight= be specified, then the attribute is re-assigned to each default attribute value. (default cpu.shares=1024, blkio.weight=1000) If only CPUShares=weight or BlockIOWeight=weight be specified, then that implies StartupCPUShares=weight and StartupBlockIOWeight=weight.
2014-05-15 17:09:34 +02:00
core: add MemoryMin The kernel added support for a new cgroup memory controller knob memory.min in bf8d5d52ffe8 ("memcg: introduce memory.min") which was merged during v4.18 merge window. Add MemoryMin to support memory.min.
2018-06-09 02:33:14 +02:00
if (streq(name, "MemoryMinScale"))
return bus_cgroup_set_memory_scale(u, name, &c->memory_min, message, flags, error);
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "MemoryLowScale"))
return bus_cgroup_set_memory_scale(u, name, &c->memory_low, message, flags, error);
if (streq(name, "MemoryHighScale"))
return bus_cgroup_set_memory_scale(u, name, &c->memory_high, message, flags, error);
if (streq(name, "MemorySwapMaxScale"))
core/cgroup: accepts MemorySwapMax=0 (#8366) Also, this moves two macros from dbus-util.h to dbus-cgroup.c, as they are only used in dbus-cgroup.c. Fixes #8363.
2018-03-09 11:34:50 +01:00
return bus_cgroup_set_swap_scale(u, name, &c->memory_swap_max, message, flags, error);
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "MemoryMaxScale"))
return bus_cgroup_set_memory_scale(u, name, &c->memory_max, message, flags, error);
core: add startup resource control option Similar to CPUShares= and BlockIOWeight= respectively. However only assign the specified weight during startup. Each control group attribute is re-assigned as weight by CPUShares=weight and BlockIOWeight=weight after startup. If not CPUShares= or BlockIOWeight= be specified, then the attribute is re-assigned to each default attribute value. (default cpu.shares=1024, blkio.weight=1000) If only CPUShares=weight or BlockIOWeight=weight be specified, then that implies StartupCPUShares=weight and StartupBlockIOWeight=weight.
2014-05-15 17:09:34 +02:00
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
if (streq(name, "MemoryLimitScale"))
return bus_cgroup_set_memory_scale(u, name, &c->memory_limit, message, flags, error);
cgroup: fix handling of TasksAccounting= property
2018-01-15 11:33:20 +01:00
if (streq(name, "TasksAccounting"))
dbus-cgroup: simplify bus_cgroup_set_property()
2018-01-02 18:30:16 +01:00
return bus_cgroup_set_boolean(u, name, &c->tasks_accounting, CGROUP_MASK_PIDS, message, flags, error);
if (streq(name, "TasksMax"))
return bus_cgroup_set_tasks_max(u, name, &c->tasks_max, message, flags, error);
if (streq(name, "TasksMaxScale"))
return bus_cgroup_set_tasks_max_scale(u, name, &c->tasks_max, message, flags, error);
if (streq(name, "CPUQuotaPerSecUSec")) {
core: expose CFS CPU time quota as high-level unit properties
2014-04-25 13:27:25 +02:00
uint64_t u64;
r = sd_bus_message_read(message, "t", &u64);
if (r < 0)
return r;
if (u64 <= 0)
tree-wide: generate SD_BUS_ERROR_INVALID_ARGS when we get invalid arguments on bus calls Let's make sure that when we return a D-Bus error, we return a native one, if we generate it ourselves, and use errno-based error synthetization only if we received an errno ourselves. Yes, this makes things slightly longer, but is highly misleading as we propagate D-Bus errors, and not errnos to the client.
2017-11-22 15:15:11 +01:00
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "CPUQuotaPerSecUSec= value out of range");
core: expose CFS CPU time quota as high-level unit properties
2014-04-25 13:27:25 +02:00
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: expose CFS CPU time quota as high-level unit properties
2014-04-25 13:27:25 +02:00
c->cpu_quota_per_sec_usec = u64;
core: downgrade CPUQuotaPeriodSec= clamping logs to debug After the first warning log, further messages are downgraded to LOG_DEBUG.
2019-01-24 05:19:44 +01:00
u->warned_clamping_cpu_quota_period = false;
cgroup: unify how we invalidate cgroup controller settings Let's make sure that we follow the same codepaths when adjusting a cgroup property via the dbus SetProperty() call, and when we execute the StartupCPUShares= effect.
2015-09-11 18:21:53 +02:00
unit_invalidate_cgroup(u, CGROUP_MASK_CPU);
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
pid1: properly encode infinity when writing CPUQuota snippet (#6141) We would write [Slice] CPUQuota=1844674407370955% which is (numerically) correct, but it seems better to just write [Slice] CPUQuota= which is interpreted as USEC_INFINITY by the parser in config_parse_cpu_quota(). Fixes #5965.
2017-06-18 11:18:41 +02:00
if (c->cpu_quota_per_sec_usec == USEC_INFINITY)
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
unit_write_setting(u, flags, "CPUQuota", "CPUQuota=");
pid1: properly encode infinity when writing CPUQuota snippet (#6141) We would write [Slice] CPUQuota=1844674407370955% which is (numerically) correct, but it seems better to just write [Slice] CPUQuota= which is interpreted as USEC_INFINITY by the parser in config_parse_cpu_quota(). Fixes #5965.
2017-06-18 11:18:41 +02:00
else
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
/* config_parse_cpu_quota() requires an integer, so truncating division is used on
* purpose here. */
unit_write_settingf(u, flags, "CPUQuota",
"CPUQuota=%0.f%%",
(double) (c->cpu_quota_per_sec_usec / 10000));
core: expose CFS CPU time quota as high-level unit properties
2014-04-25 13:27:25 +02:00
}
return 1;
core: add CPUQuotaPeriodSec= This new setting allows configuration of CFS period on the CPU cgroup, instead of using a hardcoded default of 100ms. Tested: - Legacy cgroup + Unified cgroup - systemctl set-property - systemctl show - Confirmed that the cgroup settings (such as cpu.cfs_period_ns) were set appropriately, including updating the CPU quota (cpu.cfs_quota_ns) when CPUQuotaPeriodSec= is updated. - Checked that clamping works properly when either period or (quota * period) are below the resolution of 1ms, or if period is above the max of 1s.
2018-11-02 17:21:57 +01:00
} else if (streq(name, "CPUQuotaPeriodUSec")) {
uint64_t u64;
r = sd_bus_message_read(message, "t", &u64);
if (r < 0)
return r;
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
c->cpu_quota_period_usec = u64;
core: downgrade CPUQuotaPeriodSec= clamping logs to debug After the first warning log, further messages are downgraded to LOG_DEBUG.
2019-01-24 05:19:44 +01:00
u->warned_clamping_cpu_quota_period = false;
core: add CPUQuotaPeriodSec= This new setting allows configuration of CFS period on the CPU cgroup, instead of using a hardcoded default of 100ms. Tested: - Legacy cgroup + Unified cgroup - systemctl set-property - systemctl show - Confirmed that the cgroup settings (such as cpu.cfs_period_ns) were set appropriately, including updating the CPU quota (cpu.cfs_quota_ns) when CPUQuotaPeriodSec= is updated. - Checked that clamping works properly when either period or (quota * period) are below the resolution of 1ms, or if period is above the max of 1s.
2018-11-02 17:21:57 +01:00
unit_invalidate_cgroup(u, CGROUP_MASK_CPU);
if (c->cpu_quota_period_usec == USEC_INFINITY)
unit_write_setting(u, flags, "CPUQuotaPeriodSec", "CPUQuotaPeriodSec=");
else {
char v[FORMAT_TIMESPAN_MAX];
unit_write_settingf(u, flags, "CPUQuotaPeriodSec",
"CPUQuotaPeriodSec=%s",
format_timespan(v, sizeof(v), c->cpu_quota_period_usec, 1));
}
}
return 1;
core: introduce CGroupIOLimitType enums Currently, there are two cgroup IO limits, bandwidth max for read and write, and they are hard-coded in various places. This is fine for two limits but IO is expected to grow more limits - low, high and max limits for bandwidth and IOPS - and hard-coding each limit won't make sense. This patch replaces hard-coded limits with an array indexed by CGroupIOLimitType and accompanying string and default value tables so that new limits can be added trivially.
2016-05-18 22:50:56 +02:00
} else if ((iol_type = cgroup_io_limit_type_from_string(name)) >= 0) {
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
const char *path;
unsigned n = 0;
uint64_t u64;
r = sd_bus_message_enter_container(message, 'a', "(st)");
if (r < 0)
return r;
while ((r = sd_bus_message_read(message, "(st)", &path, &u64)) > 0) {
cgroup: relax checks for block device cgroup settings This drops needless safety checks that ensure we only reference block devices for blockio/io settings. The backing code was already able to accept regular file system paths too, in which case the backing device node of that file system would be used. Hence, let's drop the artificial restrictions and open up this underlying functionality.
2018-06-11 12:19:01 +02:00
if (!path_is_normalized(path))
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Path '%s' specified in %s= is not normalized.", name, path);
cgroup: move path checking logic to dbus-cgroup.c
2017-12-19 07:34:04 +01:00
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
CGroupIODeviceLimit *a = NULL, *b;
LIST_FOREACH(device_limits, b, c->io_device_limits) {
if (path_equal(path, b->path)) {
a = b;
break;
}
}
if (!a) {
core: introduce CGroupIOLimitType enums Currently, there are two cgroup IO limits, bandwidth max for read and write, and they are hard-coded in various places. This is fine for two limits but IO is expected to grow more limits - low, high and max limits for bandwidth and IOPS - and hard-coding each limit won't make sense. This patch replaces hard-coded limits with an array indexed by CGroupIOLimitType and accompanying string and default value tables so that new limits can be added trivially.
2016-05-18 22:50:56 +02:00
CGroupIOLimitType type;
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
a = new0(CGroupIODeviceLimit, 1);
if (!a)
return -ENOMEM;
a->path = strdup(path);
if (!a->path) {
free(a);
return -ENOMEM;
}
core: introduce CGroupIOLimitType enums Currently, there are two cgroup IO limits, bandwidth max for read and write, and they are hard-coded in various places. This is fine for two limits but IO is expected to grow more limits - low, high and max limits for bandwidth and IOPS - and hard-coding each limit won't make sense. This patch replaces hard-coded limits with an array indexed by CGroupIOLimitType and accompanying string and default value tables so that new limits can be added trivially.
2016-05-18 22:50:56 +02:00
for (type = 0; type < _CGROUP_IO_LIMIT_TYPE_MAX; type++)
a->limits[type] = cgroup_io_limit_defaults[type];
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
LIST_PREPEND(device_limits, c->io_device_limits, a);
}
core: introduce CGroupIOLimitType enums Currently, there are two cgroup IO limits, bandwidth max for read and write, and they are hard-coded in various places. This is fine for two limits but IO is expected to grow more limits - low, high and max limits for bandwidth and IOPS - and hard-coding each limit won't make sense. This patch replaces hard-coded limits with an array indexed by CGroupIOLimitType and accompanying string and default value tables so that new limits can be added trivially.
2016-05-18 22:50:56 +02:00
a->limits[iol_type] = u64;
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
}
n++;
}
if (r < 0)
return r;
r = sd_bus_message_exit_container(message);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
CGroupIODeviceLimit *a;
_cleanup_free_ char *buf = NULL;
_cleanup_fclose_ FILE *f = NULL;
size_t size = 0;
if (n == 0) {
LIST_FOREACH(device_limits, a, c->io_device_limits)
core: introduce CGroupIOLimitType enums Currently, there are two cgroup IO limits, bandwidth max for read and write, and they are hard-coded in various places. This is fine for two limits but IO is expected to grow more limits - low, high and max limits for bandwidth and IOPS - and hard-coding each limit won't make sense. This patch replaces hard-coded limits with an array indexed by CGroupIOLimitType and accompanying string and default value tables so that new limits can be added trivially.
2016-05-18 22:50:56 +02:00
a->limits[iol_type] = cgroup_io_limit_defaults[iol_type];
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
}
unit_invalidate_cgroup(u, CGROUP_MASK_IO);
Add open_memstream_unlocked() wrapper
2019-04-04 11:46:44 +02:00
f = open_memstream_unlocked(&buf, &size);
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
if (!f)
return -ENOMEM;
core: introduce CGroupIOLimitType enums Currently, there are two cgroup IO limits, bandwidth max for read and write, and they are hard-coded in various places. This is fine for two limits but IO is expected to grow more limits - low, high and max limits for bandwidth and IOPS - and hard-coding each limit won't make sense. This patch replaces hard-coded limits with an array indexed by CGroupIOLimitType and accompanying string and default value tables so that new limits can be added trivially.
2016-05-18 22:50:56 +02:00
fprintf(f, "%s=\n", name);
LIST_FOREACH(device_limits, a, c->io_device_limits)
if (a->limits[iol_type] != cgroup_io_limit_defaults[iol_type])
fprintf(f, "%s=%s %" PRIu64 "\n", name, a->path, a->limits[iol_type]);
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
r = fflush_and_check(f);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
unit_write_setting(u, flags, name, buf);
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
}
return 1;
} else if (streq(name, "IODeviceWeight")) {
const char *path;
uint64_t weight;
unsigned n = 0;
r = sd_bus_message_enter_container(message, 'a', "(st)");
if (r < 0)
return r;
while ((r = sd_bus_message_read(message, "(st)", &path, &weight)) > 0) {
cgroup: relax checks for block device cgroup settings This drops needless safety checks that ensure we only reference block devices for blockio/io settings. The backing code was already able to accept regular file system paths too, in which case the backing device node of that file system would be used. Hence, let's drop the artificial restrictions and open up this underlying functionality.
2018-06-11 12:19:01 +02:00
if (!path_is_normalized(path))
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Path '%s' specified in %s= is not normalized.", name, path);
cgroup: move path checking logic to dbus-cgroup.c
2017-12-19 07:34:04 +01:00
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
if (!CGROUP_WEIGHT_IS_OK(weight) || weight == CGROUP_WEIGHT_INVALID)
tree-wide: generate SD_BUS_ERROR_INVALID_ARGS when we get invalid arguments on bus calls Let's make sure that when we return a D-Bus error, we return a native one, if we generate it ourselves, and use errno-based error synthetization only if we received an errno ourselves. Yes, this makes things slightly longer, but is highly misleading as we propagate D-Bus errors, and not errnos to the client.
2017-11-22 15:15:11 +01:00
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "IODeviceWeight= value out of range");
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
CGroupIODeviceWeight *a = NULL, *b;
LIST_FOREACH(device_weights, b, c->io_device_weights) {
if (path_equal(b->path, path)) {
a = b;
break;
}
}
if (!a) {
a = new0(CGroupIODeviceWeight, 1);
if (!a)
return -ENOMEM;
a->path = strdup(path);
if (!a->path) {
free(a);
return -ENOMEM;
}
dbus-cgroup: add missing space
2018-01-02 18:28:17 +01:00
LIST_PREPEND(device_weights, c->io_device_weights, a);
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
}
a->weight = weight;
}
n++;
}
r = sd_bus_message_exit_container(message);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
_cleanup_free_ char *buf = NULL;
_cleanup_fclose_ FILE *f = NULL;
CGroupIODeviceWeight *a;
size_t size = 0;
if (n == 0) {
while (c->io_device_weights)
cgroup_context_free_io_device_weight(c, c->io_device_weights);
}
unit_invalidate_cgroup(u, CGROUP_MASK_IO);
Add open_memstream_unlocked() wrapper
2019-04-04 11:46:44 +02:00
f = open_memstream_unlocked(&buf, &size);
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
if (!f)
return -ENOMEM;
tree-wide: use __fsetlocking() instead of fxyz_unlocked() Let's replace usage of fputc_unlocked() and friends by __fsetlocking(f, FSETLOCKING_BYCALLER). This turns off locking for the entire FILE*, instead of doing individual per-call decision whether to use normal calls or _unlocked() calls. This has various benefits: 1. It's easier to read and easier not to forget 2. It's more comprehensive, as fprintf() and friends are covered too (as these functions have no _unlocked() counterpart) 3. Philosophically, it's a bit more correct, because it's more a property of the file handle really whether we ever pass it on to another thread, not of the operations we then apply to it. This patch reworks all pieces of codes that so far used fxyz_unlocked() calls to use __fsetlocking() instead. It also reworks all places that use open_memstream(), i.e. use stdio FILE* for string manipulations. Note that this in some way a revert of 4b61c8751135c58be043d86b9fef4c8ec7aadf18.
2017-12-11 19:50:30 +01:00
fputs("IODeviceWeight=\n", f);
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
LIST_FOREACH(device_weights, a, c->io_device_weights)
fprintf(f, "IODeviceWeight=%s %" PRIu64 "\n", a->path, a->weight);
r = fflush_and_check(f);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
unit_write_setting(u, flags, name, buf);
core: add IODeviceLatencyTargetSec This adds support for the following proposed latency based IO control mechanism. https://lkml.org/lkml/2018/6/5/428
2018-06-13 23:16:35 +02:00
}
return 1;
} else if (streq(name, "IODeviceLatencyTargetUSec")) {
const char *path;
uint64_t target;
unsigned n = 0;
r = sd_bus_message_enter_container(message, 'a', "(st)");
if (r < 0)
return r;
while ((r = sd_bus_message_read(message, "(st)", &path, &target)) > 0) {
if (!path_is_normalized(path))
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Path '%s' specified in %s= is not normalized.", name, path);
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
CGroupIODeviceLatency *a = NULL, *b;
LIST_FOREACH(device_latencies, b, c->io_device_latencies) {
if (path_equal(b->path, path)) {
a = b;
break;
}
}
if (!a) {
a = new0(CGroupIODeviceLatency, 1);
if (!a)
return -ENOMEM;
a->path = strdup(path);
if (!a->path) {
free(a);
return -ENOMEM;
}
LIST_PREPEND(device_latencies, c->io_device_latencies, a);
}
a->target_usec = target;
}
n++;
}
r = sd_bus_message_exit_container(message);
if (r < 0)
return r;
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
_cleanup_free_ char *buf = NULL;
_cleanup_fclose_ FILE *f = NULL;
char ts[FORMAT_TIMESPAN_MAX];
CGroupIODeviceLatency *a;
size_t size = 0;
if (n == 0) {
while (c->io_device_latencies)
cgroup_context_free_io_device_latency(c, c->io_device_latencies);
}
unit_invalidate_cgroup(u, CGROUP_MASK_IO);
Add open_memstream_unlocked() wrapper
2019-04-04 11:46:44 +02:00
f = open_memstream_unlocked(&buf, &size);
core: add IODeviceLatencyTargetSec This adds support for the following proposed latency based IO control mechanism. https://lkml.org/lkml/2018/6/5/428
2018-06-13 23:16:35 +02:00
if (!f)
return -ENOMEM;
fputs("IODeviceLatencyTargetSec=\n", f);
LIST_FOREACH(device_latencies, a, c->io_device_latencies)
fprintf(f, "IODeviceLatencyTargetSec=%s %s\n",
a->path, format_timespan(ts, sizeof(ts), a->target_usec, 1));
r = fflush_and_check(f);
if (r < 0)
return r;
unit_write_setting(u, flags, name, buf);
core: add io controller support on the unified hierarchy On the unified hierarchy, blkio controller is renamed to io and the interface is changed significantly. * blkio.weight and blkio.weight_device are consolidated into io.weight which uses the standardized weight range [1, 10000] with 100 as the default value. * blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max. Expansion of throttling features is being worked on to support work-conserving absolute limits (io.low and io.high). * All stats are consolidated into io.stats. This patchset adds support for the new interface. As the interface has been revamped and new features are expected to be added, it seems best to treat it as a separate controller rather than trying to expand the blkio settings although we might add automatic translation if only blkio settings are specified. * io.weight handling is mostly identical to blkio.weight[_device] handling except that the weight range is different. * Both read and write bandwidth settings are consolidated into CGroupIODeviceLimit which describes all limits applicable to the device. This makes it less painful to add new limits. * "max" can be used to specify the maximum limit which is equivalent to no config for max limits and treated as such. If a given CGroupIODeviceLimit doesn't contain any non-default configs, the config struct is discarded once the no limit config is applied to cgroup. * lookup_blkio_device() is renamed to lookup_block_device(). Signed-off-by: Tejun Heo <htejun@fb.com>
2016-05-05 22:42:55 +02:00
}
return 1;
core: when receiving a memory limit via the bus, refuse 0 When parsing unit files we already refuse unit memory limits of zero, let's also refuse it when the value is set via the bus.
2016-06-08 20:04:22 +02:00
} else if (STR_IN_SET(name, "BlockIOReadBandwidth", "BlockIOWriteBandwidth")) {
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
const char *path;
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
bool read = true;
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
unsigned n = 0;
uint64_t u64;
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
if (streq(name, "BlockIOWriteBandwidth"))
read = false;
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
r = sd_bus_message_enter_container(message, 'a', "(st)");
if (r < 0)
return r;
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
while ((r = sd_bus_message_read(message, "(st)", &path, &u64)) > 0) {
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
cgroup: relax checks for block device cgroup settings This drops needless safety checks that ensure we only reference block devices for blockio/io settings. The backing code was already able to accept regular file system paths too, in which case the backing device node of that file system would be used. Hence, let's drop the artificial restrictions and open up this underlying functionality.
2018-06-11 12:19:01 +02:00
if (!path_is_normalized(path))
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Path '%s' specified in %s= is not normalized.", name, path);
cgroup: move path checking logic to dbus-cgroup.c
2017-12-19 07:34:04 +01:00
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
CGroupBlockIODeviceBandwidth *a = NULL, *b;
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
LIST_FOREACH(device_bandwidths, b, c->blockio_device_bandwidths) {
core: update CGroupBlockIODeviceBandwidth to record both rbps and wbps CGroupBlockIODeviceBandwith is used to keep track of IO bandwidth limits for legacy cgroup hierarchies. Unlike the unified hierarchy counterpart CGroupIODeviceLimit, a CGroupBlockIODeviceBandwiddth records either a read or write limit and has a couple issues. * There's no way to clear specific config entry. * When configs are cleared for an IO direction of a unit, the kernel settings aren't cleared accordingly creating discrepancies. This patch updates CGroupBlockIODeviceBandwidth so that it behaves similarly to CGroupIODeviceLimit - each entry records both rbps and wbps limits and is cleared if both are at default values after kernel settings are updated.
2016-05-18 22:51:46 +02:00
if (path_equal(path, b->path)) {
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
a = b;
break;
}
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
if (!a) {
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
a = new0(CGroupBlockIODeviceBandwidth, 1);
if (!a)
return -ENOMEM;
core: update CGroupBlockIODeviceBandwidth to record both rbps and wbps CGroupBlockIODeviceBandwith is used to keep track of IO bandwidth limits for legacy cgroup hierarchies. Unlike the unified hierarchy counterpart CGroupIODeviceLimit, a CGroupBlockIODeviceBandwiddth records either a read or write limit and has a couple issues. * There's no way to clear specific config entry. * When configs are cleared for an IO direction of a unit, the kernel settings aren't cleared accordingly creating discrepancies. This patch updates CGroupBlockIODeviceBandwidth so that it behaves similarly to CGroupIODeviceLimit - each entry records both rbps and wbps limits and is cleared if both are at default values after kernel settings are updated.
2016-05-18 22:51:46 +02:00
a->rbps = CGROUP_LIMIT_MAX;
a->wbps = CGROUP_LIMIT_MAX;
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
a->path = strdup(path);
if (!a->path) {
free(a);
return -ENOMEM;
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
LIST_PREPEND(device_bandwidths, c->blockio_device_bandwidths, a);
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
}
core: update CGroupBlockIODeviceBandwidth to record both rbps and wbps CGroupBlockIODeviceBandwith is used to keep track of IO bandwidth limits for legacy cgroup hierarchies. Unlike the unified hierarchy counterpart CGroupIODeviceLimit, a CGroupBlockIODeviceBandwiddth records either a read or write limit and has a couple issues. * There's no way to clear specific config entry. * When configs are cleared for an IO direction of a unit, the kernel settings aren't cleared accordingly creating discrepancies. This patch updates CGroupBlockIODeviceBandwidth so that it behaves similarly to CGroupIODeviceLimit - each entry records both rbps and wbps limits and is cleared if both are at default values after kernel settings are updated.
2016-05-18 22:51:46 +02:00
if (read)
a->rbps = u64;
else
a->wbps = u64;
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
}
n++;
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
if (r < 0)
return r;
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
cgroup: parse array cgroup properties correctly when they aren't at the end of the message
2014-02-24 02:59:57 +01:00
r = sd_bus_message_exit_container(message);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: update CGroupBlockIODeviceBandwidth to record both rbps and wbps CGroupBlockIODeviceBandwith is used to keep track of IO bandwidth limits for legacy cgroup hierarchies. Unlike the unified hierarchy counterpart CGroupIODeviceLimit, a CGroupBlockIODeviceBandwiddth records either a read or write limit and has a couple issues. * There's no way to clear specific config entry. * When configs are cleared for an IO direction of a unit, the kernel settings aren't cleared accordingly creating discrepancies. This patch updates CGroupBlockIODeviceBandwidth so that it behaves similarly to CGroupIODeviceLimit - each entry records both rbps and wbps limits and is cleared if both are at default values after kernel settings are updated.
2016-05-18 22:51:46 +02:00
CGroupBlockIODeviceBandwidth *a;
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
_cleanup_free_ char *buf = NULL;
_cleanup_fclose_ FILE *f = NULL;
size_t size = 0;
if (n == 0) {
core: update CGroupBlockIODeviceBandwidth to record both rbps and wbps CGroupBlockIODeviceBandwith is used to keep track of IO bandwidth limits for legacy cgroup hierarchies. Unlike the unified hierarchy counterpart CGroupIODeviceLimit, a CGroupBlockIODeviceBandwiddth records either a read or write limit and has a couple issues. * There's no way to clear specific config entry. * When configs are cleared for an IO direction of a unit, the kernel settings aren't cleared accordingly creating discrepancies. This patch updates CGroupBlockIODeviceBandwidth so that it behaves similarly to CGroupIODeviceLimit - each entry records both rbps and wbps limits and is cleared if both are at default values after kernel settings are updated.
2016-05-18 22:51:46 +02:00
LIST_FOREACH(device_bandwidths, a, c->blockio_device_bandwidths) {
if (read)
a->rbps = CGROUP_LIMIT_MAX;
else
a->wbps = CGROUP_LIMIT_MAX;
}
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
}
cgroup: unify how we invalidate cgroup controller settings Let's make sure that we follow the same codepaths when adjusting a cgroup property via the dbus SetProperty() call, and when we execute the StartupCPUShares= effect.
2015-09-11 18:21:53 +02:00
unit_invalidate_cgroup(u, CGROUP_MASK_BLKIO);
core: make sure we always write changed cgroup attributes to the cgroupfs
2014-04-25 13:26:37 +02:00
Add open_memstream_unlocked() wrapper
2019-04-04 11:46:44 +02:00
f = open_memstream_unlocked(&buf, &size);
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
if (!f)
return -ENOMEM;
tree-wide: fix indentation
2015-08-06 00:31:09 +02:00
if (read) {
tree-wide: use __fsetlocking() instead of fxyz_unlocked() Let's replace usage of fputc_unlocked() and friends by __fsetlocking(f, FSETLOCKING_BYCALLER). This turns off locking for the entire FILE*, instead of doing individual per-call decision whether to use normal calls or _unlocked() calls. This has various benefits: 1. It's easier to read and easier not to forget 2. It's more comprehensive, as fprintf() and friends are covered too (as these functions have no _unlocked() counterpart) 3. Philosophically, it's a bit more correct, because it's more a property of the file handle really whether we ever pass it on to another thread, not of the operations we then apply to it. This patch reworks all pieces of codes that so far used fxyz_unlocked() calls to use __fsetlocking() instead. It also reworks all places that use open_memstream(), i.e. use stdio FILE* for string manipulations. Note that this in some way a revert of 4b61c8751135c58be043d86b9fef4c8ec7aadf18.
2017-12-11 19:50:30 +01:00
fputs("BlockIOReadBandwidth=\n", f);
tree-wide: fix indentation
2015-08-06 00:31:09 +02:00
LIST_FOREACH(device_bandwidths, a, c->blockio_device_bandwidths)
core: update CGroupBlockIODeviceBandwidth to record both rbps and wbps CGroupBlockIODeviceBandwith is used to keep track of IO bandwidth limits for legacy cgroup hierarchies. Unlike the unified hierarchy counterpart CGroupIODeviceLimit, a CGroupBlockIODeviceBandwiddth records either a read or write limit and has a couple issues. * There's no way to clear specific config entry. * When configs are cleared for an IO direction of a unit, the kernel settings aren't cleared accordingly creating discrepancies. This patch updates CGroupBlockIODeviceBandwidth so that it behaves similarly to CGroupIODeviceLimit - each entry records both rbps and wbps limits and is cleared if both are at default values after kernel settings are updated.
2016-05-18 22:51:46 +02:00
if (a->rbps != CGROUP_LIMIT_MAX)
fprintf(f, "BlockIOReadBandwidth=%s %" PRIu64 "\n", a->path, a->rbps);
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
} else {
tree-wide: use __fsetlocking() instead of fxyz_unlocked() Let's replace usage of fputc_unlocked() and friends by __fsetlocking(f, FSETLOCKING_BYCALLER). This turns off locking for the entire FILE*, instead of doing individual per-call decision whether to use normal calls or _unlocked() calls. This has various benefits: 1. It's easier to read and easier not to forget 2. It's more comprehensive, as fprintf() and friends are covered too (as these functions have no _unlocked() counterpart) 3. Philosophically, it's a bit more correct, because it's more a property of the file handle really whether we ever pass it on to another thread, not of the operations we then apply to it. This patch reworks all pieces of codes that so far used fxyz_unlocked() calls to use __fsetlocking() instead. It also reworks all places that use open_memstream(), i.e. use stdio FILE* for string manipulations. Note that this in some way a revert of 4b61c8751135c58be043d86b9fef4c8ec7aadf18.
2017-12-11 19:50:30 +01:00
fputs("BlockIOWriteBandwidth=\n", f);
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
LIST_FOREACH(device_bandwidths, a, c->blockio_device_bandwidths)
core: update CGroupBlockIODeviceBandwidth to record both rbps and wbps CGroupBlockIODeviceBandwith is used to keep track of IO bandwidth limits for legacy cgroup hierarchies. Unlike the unified hierarchy counterpart CGroupIODeviceLimit, a CGroupBlockIODeviceBandwiddth records either a read or write limit and has a couple issues. * There's no way to clear specific config entry. * When configs are cleared for an IO direction of a unit, the kernel settings aren't cleared accordingly creating discrepancies. This patch updates CGroupBlockIODeviceBandwidth so that it behaves similarly to CGroupIODeviceLimit - each entry records both rbps and wbps limits and is cleared if both are at default values after kernel settings are updated.
2016-05-18 22:51:46 +02:00
if (a->wbps != CGROUP_LIMIT_MAX)
fprintf(f, "BlockIOWriteBandwidth=%s %" PRIu64 "\n", a->path, a->wbps);
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
}
core dbus: Check that flush works with memstream
2015-10-21 18:17:12 +02:00
r = fflush_and_check(f);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
unit_write_setting(u, flags, name, buf);
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
}
return 1;
} else if (streq(name, "BlockIODeviceWeight")) {
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
const char *path;
core: refactor cpu shares/blockio weight cgroup logic Let's stop using the "unsigned long" type for weights/shares, and let's just use uint64_t for this, as that's what we expose on the bus. Unify parsers, and always validate the range for these fields. Correct the default blockio weight to 500, since that's what the kernel actually uses. When parsing the weight/shares settings from unit files accept the empty string as a way to reset the weight/shares value. When getting it via the bus, uniformly map (uint64_t) -1 to unset. Open up StartupCPUShares= and StartupBlockIOWeight= to transient units.
2015-09-11 16:48:24 +02:00
uint64_t weight;
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
unsigned n = 0;
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
r = sd_bus_message_enter_container(message, 'a', "(st)");
if (r < 0)
return r;
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
core: refactor cpu shares/blockio weight cgroup logic Let's stop using the "unsigned long" type for weights/shares, and let's just use uint64_t for this, as that's what we expose on the bus. Unify parsers, and always validate the range for these fields. Correct the default blockio weight to 500, since that's what the kernel actually uses. When parsing the weight/shares settings from unit files accept the empty string as a way to reset the weight/shares value. When getting it via the bus, uniformly map (uint64_t) -1 to unset. Open up StartupCPUShares= and StartupBlockIOWeight= to transient units.
2015-09-11 16:48:24 +02:00
while ((r = sd_bus_message_read(message, "(st)", &path, &weight)) > 0) {
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
cgroup: relax checks for block device cgroup settings This drops needless safety checks that ensure we only reference block devices for blockio/io settings. The backing code was already able to accept regular file system paths too, in which case the backing device node of that file system would be used. Hence, let's drop the artificial restrictions and open up this underlying functionality.
2018-06-11 12:19:01 +02:00
if (!path_is_normalized(path))
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Path '%s' specified in %s= is not normalized.", name, path);
cgroup: move path checking logic to dbus-cgroup.c
2017-12-19 07:34:04 +01:00
core: refactor cpu shares/blockio weight cgroup logic Let's stop using the "unsigned long" type for weights/shares, and let's just use uint64_t for this, as that's what we expose on the bus. Unify parsers, and always validate the range for these fields. Correct the default blockio weight to 500, since that's what the kernel actually uses. When parsing the weight/shares settings from unit files accept the empty string as a way to reset the weight/shares value. When getting it via the bus, uniformly map (uint64_t) -1 to unset. Open up StartupCPUShares= and StartupBlockIOWeight= to transient units.
2015-09-11 16:48:24 +02:00
if (!CGROUP_BLKIO_WEIGHT_IS_OK(weight) || weight == CGROUP_BLKIO_WEIGHT_INVALID)
tree-wide: generate SD_BUS_ERROR_INVALID_ARGS when we get invalid arguments on bus calls Let's make sure that when we return a D-Bus error, we return a native one, if we generate it ourselves, and use errno-based error synthetization only if we received an errno ourselves. Yes, this makes things slightly longer, but is highly misleading as we propagate D-Bus errors, and not errnos to the client.
2017-11-22 15:15:11 +01:00
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "BlockIODeviceWeight= out of range");
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
CGroupBlockIODeviceWeight *a = NULL, *b;
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
LIST_FOREACH(device_weights, b, c->blockio_device_weights) {
if (path_equal(b->path, path)) {
a = b;
break;
}
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
if (!a) {
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
a = new0(CGroupBlockIODeviceWeight, 1);
if (!a)
return -ENOMEM;
a->path = strdup(path);
if (!a->path) {
free(a);
return -ENOMEM;
}
dbus-cgroup: add missing space
2018-01-02 18:28:17 +01:00
LIST_PREPEND(device_weights, c->blockio_device_weights, a);
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
}
core: refactor cpu shares/blockio weight cgroup logic Let's stop using the "unsigned long" type for weights/shares, and let's just use uint64_t for this, as that's what we expose on the bus. Unify parsers, and always validate the range for these fields. Correct the default blockio weight to 500, since that's what the kernel actually uses. When parsing the weight/shares settings from unit files accept the empty string as a way to reset the weight/shares value. When getting it via the bus, uniformly map (uint64_t) -1 to unset. Open up StartupCPUShares= and StartupBlockIOWeight= to transient units.
2015-09-11 16:48:24 +02:00
a->weight = weight;
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
}
n++;
}
cgroup: parse array cgroup properties correctly when they aren't at the end of the message
2014-02-24 02:59:57 +01:00
r = sd_bus_message_exit_container(message);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
_cleanup_free_ char *buf = NULL;
_cleanup_fclose_ FILE *f = NULL;
CGroupBlockIODeviceWeight *a;
size_t size = 0;
if (n == 0) {
while (c->blockio_device_weights)
cgroup_context_free_blockio_device_weight(c, c->blockio_device_weights);
}
cgroup: unify how we invalidate cgroup controller settings Let's make sure that we follow the same codepaths when adjusting a cgroup property via the dbus SetProperty() call, and when we execute the StartupCPUShares= effect.
2015-09-11 18:21:53 +02:00
unit_invalidate_cgroup(u, CGROUP_MASK_BLKIO);
core: make sure we always write changed cgroup attributes to the cgroupfs
2014-04-25 13:26:37 +02:00
Add open_memstream_unlocked() wrapper
2019-04-04 11:46:44 +02:00
f = open_memstream_unlocked(&buf, &size);
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
if (!f)
return -ENOMEM;
tree-wide: use __fsetlocking() instead of fxyz_unlocked() Let's replace usage of fputc_unlocked() and friends by __fsetlocking(f, FSETLOCKING_BYCALLER). This turns off locking for the entire FILE*, instead of doing individual per-call decision whether to use normal calls or _unlocked() calls. This has various benefits: 1. It's easier to read and easier not to forget 2. It's more comprehensive, as fprintf() and friends are covered too (as these functions have no _unlocked() counterpart) 3. Philosophically, it's a bit more correct, because it's more a property of the file handle really whether we ever pass it on to another thread, not of the operations we then apply to it. This patch reworks all pieces of codes that so far used fxyz_unlocked() calls to use __fsetlocking() instead. It also reworks all places that use open_memstream(), i.e. use stdio FILE* for string manipulations. Note that this in some way a revert of 4b61c8751135c58be043d86b9fef4c8ec7aadf18.
2017-12-11 19:50:30 +01:00
fputs("BlockIODeviceWeight=\n", f);
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
LIST_FOREACH(device_weights, a, c->blockio_device_weights)
core: refactor cpu shares/blockio weight cgroup logic Let's stop using the "unsigned long" type for weights/shares, and let's just use uint64_t for this, as that's what we expose on the bus. Unify parsers, and always validate the range for these fields. Correct the default blockio weight to 500, since that's what the kernel actually uses. When parsing the weight/shares settings from unit files accept the empty string as a way to reset the weight/shares value. When getting it via the bus, uniformly map (uint64_t) -1 to unset. Open up StartupCPUShares= and StartupBlockIOWeight= to transient units.
2015-09-11 16:48:24 +02:00
fprintf(f, "BlockIODeviceWeight=%s %" PRIu64 "\n", a->path, a->weight);
cgroup: setup BlockIODeviceWeight in bus_cgroup_set_property This patch adds the support for setting up BlockIODeviceWeight in bus_cgroup_set_property. most of the codes are copied from the case that sets up DeviceAllow.
2013-08-27 07:36:53 +02:00
core dbus: Check that flush works with memstream
2015-10-21 18:17:12 +02:00
r = fflush_and_check(f);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
unit_write_setting(u, flags, name, buf);
cgroup: setup BlockIORead/WriteBandwidth in bus_cgroup_set_property This patch adds the support for setting up BlockIORead/WriteBandwidth in bus_cgroup_set_property.
2013-09-10 17:23:07 +02:00
}
return 1;
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
} else if (streq(name, "DevicePolicy")) {
const char *policy;
CGroupDevicePolicy p;
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
r = sd_bus_message_read(message, "s", &policy);
if (r < 0)
return r;
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
p = cgroup_device_policy_from_string(policy);
if (p < 0)
return -EINVAL;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
c->device_policy = p;
cgroup: unify how we invalidate cgroup controller settings Let's make sure that we follow the same codepaths when adjusting a cgroup property via the dbus SetProperty() call, and when we execute the StartupCPUShares= effect.
2015-09-11 18:21:53 +02:00
unit_invalidate_cgroup(u, CGROUP_MASK_DEVICES);
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
unit_write_settingf(u, flags, name, "DevicePolicy=%s", policy);
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
}
return 1;
} else if (streq(name, "DeviceAllow")) {
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
const char *path, *rwm;
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
unsigned n = 0;
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
r = sd_bus_message_enter_container(message, 'a', "(ss)");
if (r < 0)
return r;
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
while ((r = sd_bus_message_read(message, "(ss)", &path, &rwm)) > 0) {
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
core: rework how we validate DeviceAllow= settings Let's make sure we don't validate "char-*" and "block-*" expressions as paths.
2018-06-11 12:22:58 +02:00
if (!valid_device_allow_pattern(path) || strpbrk(path, WHITESPACE))
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "DeviceAllow= requires device node or pattern");
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
if (isempty(rwm))
rwm = "rwm";
core: rework how we validate DeviceAllow= settings Let's make sure we don't validate "char-*" and "block-*" expressions as paths.
2018-06-11 12:22:58 +02:00
else if (!in_charset(rwm, "rwm"))
tree-wide: generate SD_BUS_ERROR_INVALID_ARGS when we get invalid arguments on bus calls Let's make sure that when we return a D-Bus error, we return a native one, if we generate it ourselves, and use errno-based error synthetization only if we received an errno ourselves. Yes, this makes things slightly longer, but is highly misleading as we propagate D-Bus errors, and not errnos to the client.
2017-11-22 15:15:11 +01:00
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "DeviceAllow= requires combination of rwm flags");
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
CGroupDeviceAllow *a = NULL, *b;
device cgroup: don't create a new CGroupDeviceAllow when it already in the list If a device node is already in the device_allow list of CGroupContext, we should replace it instead of create a new one and append this new one to the end of device_allow list. change from v1: use streq to replace !strcmp
2013-08-28 03:49:11 +02:00
LIST_FOREACH(device_allow, b, c->device_allow) {
cgroup: compare fs paths with path_equal() rather than streq()
2013-09-10 18:21:10 +02:00
if (path_equal(b->path, path)) {
device cgroup: don't create a new CGroupDeviceAllow when it already in the list If a device node is already in the device_allow list of CGroupContext, we should replace it instead of create a new one and append this new one to the end of device_allow list. change from v1: use streq to replace !strcmp
2013-08-28 03:49:11 +02:00
a = b;
break;
}
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
if (!a) {
device cgroup: don't create a new CGroupDeviceAllow when it already in the list If a device node is already in the device_allow list of CGroupContext, we should replace it instead of create a new one and append this new one to the end of device_allow list. change from v1: use streq to replace !strcmp
2013-08-28 03:49:11 +02:00
a = new0(CGroupDeviceAllow, 1);
if (!a)
return -ENOMEM;
a->path = strdup(path);
if (!a->path) {
free(a);
return -ENOMEM;
}
core: convert PID 1 to libsystemd-bus This patch converts PID 1 to libsystemd-bus and thus drops the dependency on libdbus. The only remaining code using libdbus is a test case that validates our bus marshalling against libdbus' marshalling, and this dependency can be turned off. This patch also adds a couple of things to libsystem-bus, that are necessary to make the port work: - Synthesizing of "Disconnected" messages when bus connections are severed. - Support for attaching multiple vtables for the same interface on the same path. This patch also fixes the SetDefaultTarget() and GetDefaultTarget() bus calls which used an inappropriate signature. As a side effect we will now generate PropertiesChanged messages which carry property contents, rather than just invalidation information.
2013-11-19 21:12:59 +01:00
LIST_PREPEND(device_allow, c->device_allow, a);
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
}
a->r = !!strchr(rwm, 'r');
a->w = !!strchr(rwm, 'w');
a->m = !!strchr(rwm, 'm');
}
core: add transient units Transient units can be created via the bus API. They are configured via the method call parameters rather than on-disk files. They are subject to normal GC. Transient units currently may only be created for services (however, we will extend this), and currently only ExecStart= and the cgroup parameters can be configured (also to be extended). Transient units require a unique name, that previously had no configuration file on disk. A tool systemd-run is added that makes use of this functionality to run arbitrary command lines as transient services: $ systemd-run /bin/ping www.heise.de Will cause systemd to create a new transient service and run ping in it.
2013-06-28 04:12:58 +02:00
n++;
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
}
build-sys: minor fixes found with cppcheck
2013-12-25 19:00:12 +01:00
if (r < 0)
return r;
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
cgroup: parse array cgroup properties correctly when they aren't at the end of the message
2014-02-24 02:59:57 +01:00
r = sd_bus_message_exit_container(message);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
_cleanup_free_ char *buf = NULL;
_cleanup_fclose_ FILE *f = NULL;
CGroupDeviceAllow *a;
size_t size = 0;
if (n == 0) {
while (c->device_allow)
cgroup_context_free_device_allow(c, c->device_allow);
}
cgroup: unify how we invalidate cgroup controller settings Let's make sure that we follow the same codepaths when adjusting a cgroup property via the dbus SetProperty() call, and when we execute the StartupCPUShares= effect.
2015-09-11 18:21:53 +02:00
unit_invalidate_cgroup(u, CGROUP_MASK_DEVICES);
core: make sure we always write changed cgroup attributes to the cgroupfs
2014-04-25 13:26:37 +02:00
Add open_memstream_unlocked() wrapper
2019-04-04 11:46:44 +02:00
f = open_memstream_unlocked(&buf, &size);
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
if (!f)
return -ENOMEM;
tree-wide: use __fsetlocking() instead of fxyz_unlocked() Let's replace usage of fputc_unlocked() and friends by __fsetlocking(f, FSETLOCKING_BYCALLER). This turns off locking for the entire FILE*, instead of doing individual per-call decision whether to use normal calls or _unlocked() calls. This has various benefits: 1. It's easier to read and easier not to forget 2. It's more comprehensive, as fprintf() and friends are covered too (as these functions have no _unlocked() counterpart) 3. Philosophically, it's a bit more correct, because it's more a property of the file handle really whether we ever pass it on to another thread, not of the operations we then apply to it. This patch reworks all pieces of codes that so far used fxyz_unlocked() calls to use __fsetlocking() instead. It also reworks all places that use open_memstream(), i.e. use stdio FILE* for string manipulations. Note that this in some way a revert of 4b61c8751135c58be043d86b9fef4c8ec7aadf18.
2017-12-11 19:50:30 +01:00
fputs("DeviceAllow=\n", f);
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
LIST_FOREACH(device_allow, a, c->device_allow)
fprintf(f, "DeviceAllow=%s %s%s%s\n", a->path, a->r ? "r" : "", a->w ? "w" : "", a->m ? "m" : "");
core dbus: Check that flush works with memstream
2015-10-21 18:17:12 +02:00
r = fflush_and_check(f);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
unit_write_setting(u, flags, name, buf);
dbus: make DeviceAllow=/DevicePolicy= writable
2013-06-27 23:13:17 +02:00
}
dbus: make more cgroup attributes runtime settable
2013-06-27 21:50:35 +02:00
return 1;
core: introduce new Delegate=yes/no property controlling creation of cgroup subhierarchies For priviliged units this resource control property ensures that the processes have all controllers systemd manages enabled. For unpriviliged services (those with User= set) this ensures that access rights to the service cgroup is granted to the user in question, to create further subgroups. Note that this only applies to the name=systemd hierarchy though, as access to other controllers is not safe for unpriviliged processes. Delegate=yes should be set for container scopes where a systemd instance inside the container shall manage the hierarchies below its own cgroup and have access to all controllers. Delegate=yes should also be set for user@.service, so that systemd --user can run, controlling its own cgroup tree. This commit changes machined, systemd-nspawn@.service and user@.service to set this boolean, in order to ensure that container management will just work, and the user systemd instance can run fine.
2014-11-05 17:57:23 +01:00
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
} else if (streq(name, "IPAccounting")) {
int b;
r = sd_bus_message_read(message, "b", &b);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
c->ip_accounting = b;
unit_invalidate_cgroup_bpf(u);
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
unit_write_settingf(u, flags, name, "IPAccounting=%s", yes_no(b));
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
}
return 1;
} else if (STR_IN_SET(name, "IPAddressAllow", "IPAddressDeny")) {
IPAddressAccessItem **list;
size_t n = 0;
list = streq(name, "IPAddressAllow") ? &c->ip_address_allow : &c->ip_address_deny;
r = sd_bus_message_enter_container(message, 'a', "(iayu)");
if (r < 0)
return r;
for (;;) {
const void *ap;
int32_t family;
uint32_t prefixlen;
size_t an;
r = sd_bus_message_enter_container(message, 'r', "iayu");
if (r < 0)
return r;
if (r == 0)
break;
r = sd_bus_message_read(message, "i", &family);
if (r < 0)
return r;
if (!IN_SET(family, AF_INET, AF_INET6))
tree-wide: generate SD_BUS_ERROR_INVALID_ARGS when we get invalid arguments on bus calls Let's make sure that when we return a D-Bus error, we return a native one, if we generate it ourselves, and use errno-based error synthetization only if we received an errno ourselves. Yes, this makes things slightly longer, but is highly misleading as we propagate D-Bus errors, and not errnos to the client.
2017-11-22 15:15:11 +01:00
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "%s= expects IPv4 or IPv6 addresses only.", name);
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
r = sd_bus_message_read_array(message, 'y', &ap, &an);
if (r < 0)
return r;
if (an != FAMILY_ADDRESS_SIZE(family))
tree-wide: generate SD_BUS_ERROR_INVALID_ARGS when we get invalid arguments on bus calls Let's make sure that when we return a D-Bus error, we return a native one, if we generate it ourselves, and use errno-based error synthetization only if we received an errno ourselves. Yes, this makes things slightly longer, but is highly misleading as we propagate D-Bus errors, and not errnos to the client.
2017-11-22 15:15:11 +01:00
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "IP address has wrong size for family (%s, expected %zu, got %zu)",
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
af_to_name(family), FAMILY_ADDRESS_SIZE(family), an);
r = sd_bus_message_read(message, "u", &prefixlen);
if (r < 0)
return r;
if (prefixlen > FAMILY_ADDRESS_SIZE(family)*8)
tree-wide: generate SD_BUS_ERROR_INVALID_ARGS when we get invalid arguments on bus calls Let's make sure that when we return a D-Bus error, we return a native one, if we generate it ourselves, and use errno-based error synthetization only if we received an errno ourselves. Yes, this makes things slightly longer, but is highly misleading as we propagate D-Bus errors, and not errnos to the client.
2017-11-22 15:15:11 +01:00
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Prefix length %" PRIu32 " too large for address family %s.", prefixlen, af_to_name(family));
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
IPAddressAccessItem *item;
item = new0(IPAddressAccessItem, 1);
if (!item)
return -ENOMEM;
item->family = family;
item->prefixlen = prefixlen;
memcpy(&item->address, ap, an);
LIST_PREPEND(items, *list, item);
}
r = sd_bus_message_exit_container(message);
if (r < 0)
return r;
n++;
}
r = sd_bus_message_exit_container(message);
if (r < 0)
return r;
ip-address-access: minimize IP address lists Let's drop redundant items from the IP address list after parsing. Let's also mask out redundant bits hidden by the prefixlength.
2017-09-05 17:41:34 +02:00
*list = ip_address_access_reduce(*list);
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
_cleanup_free_ char *buf = NULL;
_cleanup_fclose_ FILE *f = NULL;
IPAddressAccessItem *item;
size_t size = 0;
if (n == 0)
*list = ip_address_access_free_all(*list);
unit_invalidate_cgroup_bpf(u);
Add open_memstream_unlocked() wrapper
2019-04-04 11:46:44 +02:00
f = open_memstream_unlocked(&buf, &size);
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
if (!f)
return -ENOMEM;
tree-wide: use __fsetlocking() instead of fxyz_unlocked() Let's replace usage of fputc_unlocked() and friends by __fsetlocking(f, FSETLOCKING_BYCALLER). This turns off locking for the entire FILE*, instead of doing individual per-call decision whether to use normal calls or _unlocked() calls. This has various benefits: 1. It's easier to read and easier not to forget 2. It's more comprehensive, as fprintf() and friends are covered too (as these functions have no _unlocked() counterpart) 3. Philosophically, it's a bit more correct, because it's more a property of the file handle really whether we ever pass it on to another thread, not of the operations we then apply to it. This patch reworks all pieces of codes that so far used fxyz_unlocked() calls to use __fsetlocking() instead. It also reworks all places that use open_memstream(), i.e. use stdio FILE* for string manipulations. Note that this in some way a revert of 4b61c8751135c58be043d86b9fef4c8ec7aadf18.
2017-12-11 19:50:30 +01:00
fputs(name, f);
fputs("=\n", f);
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
LIST_FOREACH(items, item, *list) {
char buffer[CONST_MAX(INET_ADDRSTRLEN, INET6_ADDRSTRLEN)];
errno = 0;
if (!inet_ntop(item->family, &item->address, buffer, sizeof(buffer)))
return errno > 0 ? -errno : -EINVAL;
fprintf(f, "%s=%s/%u\n", name, buffer, item->prefixlen);
}
r = fflush_and_check(f);
if (r < 0)
return r;
core: add proper escaping to writing of drop-ins/transient unit files This majorly refactors the transient unit file and drop-in writing logic, so that we properly C-escape and specifier-escape (% → %%) everything we write out, so that when we read it back again, specifiers are parsed that aren't supposed to be parsed. This renames unit_write_drop_in() and friends by unit_write_setting(). The name change is supposed to clarify that the functions are not only used to write drop-in files, but also transient unit files. The previous "mode" parameter to this function is replaced by a more generic "flags", which knows additional flags for implicit C-style and specifier escaping before writing things out. This can cover most properties where either form of escaping is defined. For the cases where this isn't sufficient, we add helpers unit_escape_setting() and unit_concat_strv() for escaping individual strings or strvs properly. While we are at it, we also prettify generation of transient unit files: we try to reduce the number of section headers written out: previously we'd write the right section header our for each setting. With this change we do so only if the setting lives in a different section than the one before. (This should also be considered preparation for when we add proper APIs to systemd to write normal, persistant unit files through the bus API)
2017-11-22 15:03:51 +01:00
unit_write_setting(u, flags, name, buf);
core: warn loudly if IP firewalling is configured but not in effect
2017-09-06 17:56:15 +02:00
if (*list) {
r = bpf_firewall_supported();
if (r < 0)
return r;
bpf: beef up bpf detection, check if BPF_F_ALLOW_MULTI is supported This improves the BPF/cgroup detection logic, and looks whether BPF_ALLOW_MULTI is supported. This flag allows execution of multiple BPF filters in a recursive fashion for a whole cgroup tree. It enables us to properly report IP accounting for slice units, as well as delegation of BPF support to units without breaking our own IP accounting.
2018-02-16 11:55:33 +01:00
if (r == BPF_FIREWALL_UNSUPPORTED) {
core: only warn about BPF/cgroup missing once per runtime (#7319) Let's reduce the amount of noise a bit, there's little point in complaining loudly about every single unit like this, let's complain only about the first one, and then downgrade the log level to LOG_DEBUG for the other cases. Fixes: #7188
2017-11-13 22:02:51 +01:00
static bool warned = false;
log_full(warned ? LOG_DEBUG : LOG_WARNING,
"Transient unit %s configures an IP firewall, but the local system does not support BPF/cgroup firewalling.\n"
"Proceeding WITHOUT firewalling in effect! (This warning is only shown for the first started transient unit using IP firewalling.)", u->id);
warned = true;
}
core: warn loudly if IP firewalling is configured but not in effect
2017-09-06 17:56:15 +02:00
}
core: support IP firewalling to be configured for transient units
2017-09-05 11:16:35 +02:00
}
core: add support for the "pids" cgroup controller This adds support for the new "pids" cgroup controller of 4.3 kernels. It allows accounting the number of tasks in a cgroup and enforcing limits on it. This adds two new setting TasksAccounting= and TasksMax= to each unit, as well as a gloabl option DefaultTasksAccounting=. This also updated "cgtop" to optionally make use of the new kernel-provided accounting. systemctl has been updated to show the number of tasks for each service if it is available. This patch also adds correct support for undoing memory limits for units using a MemoryLimit=infinity syntax. We do the same for TasksMax= now and hence keep things in sync here.
2015-09-10 12:32:16 +02:00
return 1;
core: introduce new Delegate=yes/no property controlling creation of cgroup subhierarchies For priviliged units this resource control property ensures that the processes have all controllers systemd manages enabled. For unpriviliged services (those with User= set) this ensures that access rights to the service cgroup is granted to the user in question, to create further subgroups. Note that this only applies to the name=systemd hierarchy though, as access to other controllers is not safe for unpriviliged processes. Delegate=yes should be set for container scopes where a systemd instance inside the container shall manage the hierarchies below its own cgroup and have access to all controllers. Delegate=yes should also be set for user@.service, so that systemd --user can run, controlling its own cgroup tree. This commit changes machined, systemd-nspawn@.service and user@.service to set this boolean, in order to ensure that container management will just work, and the user systemd instance can run fine.
2014-11-05 17:57:23 +01:00
}
dbus-cgroup: merge several blocks which operate almost same tasks
2017-12-19 10:14:39 +01:00
if (u->transient && u->load_state == UNIT_STUB)
return bus_cgroup_set_transient_property(u, c, name, message, flags, error);
dbus: add infrastructure for changing multiple properties at once on units and hook some cgroup attributes up to it This introduces two bus calls to make runtime changes to selected bus properties, optionally with persistence. This currently hooks this up only for three cgroup atributes, but this brings the infrastructure to add more changable attributes. This allows setting multiple attributes at once, and takes an array rather than a dictionary of properties, in order to implement simple resetting of lists using the same approach as when they are sourced from unit files. This means, that list properties are appended to by this call, unless they are first reset via assigning the empty list.
2013-06-27 21:14:56 +02:00
return 0;
}