1
0
mirror of https://github.com/systemd/systemd.git synced 2024-12-23 21:35:11 +03:00

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.
This commit is contained in:
Tejun Heo 2016-05-27 09:10:18 -07:00 committed by Lennart Poettering
parent ffe54bf4bf
commit da4d897e75
8 changed files with 207 additions and 34 deletions

View File

@ -114,6 +114,13 @@
prefixed ones. On unified hierarchy, IO resource control also applies to buffered writes.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><option>Memory</option></term>
<listitem>
<para><varname>MemoryMax</varname> replaces <varname>MemoryLimit</varname>. <varname>MemoryLow</varname>
and <varname>MemoryHigh</varname> are effective only on unified hierarchy.</para>
</listitem>
</varlistentry>
</variablelist>
</para>
@ -212,6 +219,67 @@
</listitem>
</varlistentry>
<varlistentry>
<term><varname>MemoryLow=<replaceable>bytes</replaceable></varname></term>
<listitem>
<para>Specify the best-effort memory usage protection of the executed processes in this unit. If the memory
usages of this unit and all its ancestors are below their low boundaries, this unit's memory won't be
reclaimed as long as memory can be reclaimed from unprotected units.</para>
<para>Takes a memory size in bytes. If the value is suffixed with K, M, G or T, the specified memory size is
parsed as Kilobytes, Megabytes, Gigabytes, or Terabytes (with the base 1024), respectively. This controls the
<literal>memory.low</literal> control group attribute. For details about this control group attribute, see
<ulink url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink>.</para>
<para>Implies <literal>MemoryAccounting=true</literal>.</para>
<para>This setting is supported only if the unified control group hierarchy is used.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>MemoryHigh=<replaceable>bytes</replaceable></varname></term>
<listitem>
<para>Specify the high limit on memory usage of the executed processes in this unit. Memory usage may go
above the limit if unavoidable, but the processes are heavily slowed down and memory is taken away
aggressively in such cases. This is the main mechanism to control memory usage of a unit.</para>
<para>Takes a memory size in bytes. If the value is suffixed with K, M, G or T, the specified memory size is
parsed as Kilobytes, Megabytes, Gigabytes, or Terabytes (with the base 1024), respectively. If assigned the
special value <literal>max</literal>, no memory limit is applied. This controls the
<literal>memory.high</literal> control group attribute. For details about this control group attribute, see
<ulink url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink>.</para>
<para>Implies <literal>MemoryAccounting=true</literal>.</para>
<para>This setting is supported only if the unified control group hierarchy is used.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>MemoryMax=<replaceable>bytes</replaceable></varname></term>
<listitem>
<para>Specify the absolute limit on memory usage of the executed processes in this unit. If memory usage
cannot be contained under the limit, out-of-memory killer is invoked inside the unit. It is recommended to
use <varname>MemoryHigh=</varname> as the main control mechanism and use <varname>MemoryMax=</varname> as the
last line of defense.</para>
<para>Takes a memory size in bytes. If the value is suffixed with K, M, G or T, the specified memory size is
parsed as Kilobytes, Megabytes, Gigabytes, or Terabytes (with the base 1024), respectively. If assigned the
special value <literal>max</literal>, no memory limit is applied. This controls the
<literal>memory.max</literal> control group attribute. For details about this control group attribute, see
<ulink url="https://www.kernel.org/doc/Documentation/cgroup-v2.txt">cgroup-v2.txt</ulink>.</para>
<para>Implies <literal>MemoryAccounting=true</literal>.</para>
<para>This setting is supported only if the unified control group hierarchy is used. Use
<varname>MemoryLimit=</varname> on systems using the legacy control group hierarchy.</para>
</listitem>
</varlistentry>
<varlistentry>
<term><varname>MemoryLimit=<replaceable>bytes</replaceable></varname></term>
@ -230,6 +298,9 @@
url="https://www.kernel.org/doc/Documentation/cgroup-v1/memory.txt">memory.txt</ulink>.</para>
<para>Implies <literal>MemoryAccounting=true</literal>.</para>
<para>This setting is supported only if the legacy control group hierarchy is used. Use
<varname>MemoryMax=</varname> on systems using the unified control group hierarchy.</para>
</listitem>
</varlistentry>

View File

@ -46,7 +46,10 @@ void cgroup_context_init(CGroupContext *c) {
c->startup_cpu_shares = CGROUP_CPU_SHARES_INVALID;
c->cpu_quota_per_sec_usec = USEC_INFINITY;
c->memory_limit = (uint64_t) -1;
c->memory_high = CGROUP_LIMIT_MAX;
c->memory_max = CGROUP_LIMIT_MAX;
c->memory_limit = CGROUP_LIMIT_MAX;
c->io_weight = CGROUP_WEIGHT_INVALID;
c->startup_io_weight = CGROUP_WEIGHT_INVALID;
@ -147,6 +150,9 @@ void cgroup_context_dump(CGroupContext *c, FILE* f, const char *prefix) {
"%sStartupIOWeight=%" PRIu64 "\n"
"%sBlockIOWeight=%" PRIu64 "\n"
"%sStartupBlockIOWeight=%" PRIu64 "\n"
"%sMemoryLow=%" PRIu64 "\n"
"%sMemoryHigh=%" PRIu64 "\n"
"%sMemoryMax=%" PRIu64 "\n"
"%sMemoryLimit=%" PRIu64 "\n"
"%sTasksMax=%" PRIu64 "\n"
"%sDevicePolicy=%s\n"
@ -163,6 +169,9 @@ void cgroup_context_dump(CGroupContext *c, FILE* f, const char *prefix) {
prefix, c->startup_io_weight,
prefix, c->blockio_weight,
prefix, c->startup_blockio_weight,
prefix, c->memory_low,
prefix, c->memory_high,
prefix, c->memory_max,
prefix, c->memory_limit,
prefix, c->tasks_max,
prefix, cgroup_device_policy_to_string(c->device_policy),
@ -496,6 +505,23 @@ static unsigned cgroup_apply_blkio_device_limit(const char *path, const char *de
return n;
}
static bool cgroup_context_has_unified_memory_config(CGroupContext *c) {
return c->memory_low > 0 || c->memory_high != CGROUP_LIMIT_MAX || c->memory_max != CGROUP_LIMIT_MAX;
}
static void cgroup_apply_unified_memory_limit(const char *path, const char *file, uint64_t v) {
char buf[DECIMAL_STR_MAX(uint64_t) + 1] = "max";
int r;
if (v != CGROUP_LIMIT_MAX)
xsprintf(buf, "%" PRIu64 "\n", v);
r = cg_set_attribute("memory", path, file, buf);
if (r < 0)
log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
"Failed to set %s on %s: %m", file, path);
}
void cgroup_context_apply(CGroupContext *c, CGroupMask mask, const char *path, ManagerState state) {
bool is_root;
int r;
@ -662,26 +688,30 @@ void cgroup_context_apply(CGroupContext *c, CGroupMask mask, const char *path, M
}
if ((mask & CGROUP_MASK_MEMORY) && !is_root) {
if (c->memory_limit != (uint64_t) -1) {
if (cg_unified() > 0) {
uint64_t max = c->memory_max;
if (cgroup_context_has_unified_memory_config(c))
max = c->memory_max;
else
max = c->memory_limit;
cgroup_apply_unified_memory_limit(path, "memory.low", c->memory_low);
cgroup_apply_unified_memory_limit(path, "memory.high", c->memory_high);
cgroup_apply_unified_memory_limit(path, "memory.max", max);
} else {
char buf[DECIMAL_STR_MAX(uint64_t) + 1];
sprintf(buf, "%" PRIu64 "\n", c->memory_limit);
if (cg_unified() <= 0)
r = cg_set_attribute("memory", path, "memory.limit_in_bytes", buf);
if (c->memory_limit != CGROUP_LIMIT_MAX)
xsprintf(buf, "%" PRIu64 "\n", c->memory_limit);
else
r = cg_set_attribute("memory", path, "memory.max", buf);
xsprintf(buf, "%" PRIu64 "\n", c->memory_max);
} else {
if (cg_unified() <= 0)
r = cg_set_attribute("memory", path, "memory.limit_in_bytes", "-1");
else
r = cg_set_attribute("memory", path, "memory.max", "max");
r = cg_set_attribute("memory", path, "memory.limit_in_bytes", buf);
if (r < 0)
log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
"Failed to set memory.limit_in_bytes on %s: %m", path);
}
if (r < 0)
log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
"Failed to set memory.limit_in_bytes/memory.max on %s: %m", path);
}
if ((mask & CGROUP_MASK_DEVICES) && !is_root) {
@ -778,7 +808,8 @@ CGroupMask cgroup_context_get_mask(CGroupContext *c) {
mask |= CGROUP_MASK_IO | CGROUP_MASK_BLKIO;
if (c->memory_accounting ||
c->memory_limit != (uint64_t) -1)
c->memory_limit != CGROUP_LIMIT_MAX ||
cgroup_context_has_unified_memory_config(c))
mask |= CGROUP_MASK_MEMORY;
if (c->device_allow ||

View File

@ -94,6 +94,10 @@ struct CGroupContext {
LIST_HEAD(CGroupIODeviceWeight, io_device_weights);
LIST_HEAD(CGroupIODeviceLimit, io_device_limits);
uint64_t memory_low;
uint64_t memory_high;
uint64_t memory_max;
/* For legacy hierarchies */
uint64_t cpu_shares;
uint64_t startup_cpu_shares;

View File

@ -228,6 +228,9 @@ const sd_bus_vtable bus_cgroup_vtable[] = {
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),
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),
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),
@ -826,6 +829,31 @@ int bus_cgroup_set_property(
return 1;
} else if (STR_IN_SET(name, "MemoryLow", "MemoryHigh", "MemoryMax")) {
uint64_t v;
r = sd_bus_message_read(message, "t", &v);
if (r < 0)
return r;
if (mode != UNIT_CHECK) {
if (streq(name, "MemoryLow"))
c->memory_low = v;
else if (streq(name, "MemoryHigh"))
c->memory_high = v;
else
c->memory_max = v;
unit_invalidate_cgroup(u, CGROUP_MASK_MEMORY);
if (v == CGROUP_LIMIT_MAX)
unit_write_drop_in_private_format(u, mode, name, "%s=max\n", name);
else
unit_write_drop_in_private_format(u, mode, name, "%s=%" PRIu64 "\n", name, v);
}
return 1;
} else if (streq(name, "MemoryLimit")) {
uint64_t limit;

View File

@ -117,6 +117,9 @@ $1.CPUShares, config_parse_cpu_shares, 0,
$1.StartupCPUShares, config_parse_cpu_shares, 0, offsetof($1, cgroup_context.startup_cpu_shares)
$1.CPUQuota, config_parse_cpu_quota, 0, offsetof($1, cgroup_context)
$1.MemoryAccounting, config_parse_bool, 0, offsetof($1, cgroup_context.memory_accounting)
$1.MemoryLow, config_parse_memory_limit, 0, offsetof($1, cgroup_context)
$1.MemoryHigh, config_parse_memory_limit, 0, offsetof($1, cgroup_context)
$1.MemoryMax, config_parse_memory_limit, 0, offsetof($1, cgroup_context)
$1.MemoryLimit, config_parse_memory_limit, 0, offsetof($1, cgroup_context)
$1.DeviceAllow, config_parse_device_allow, 0, offsetof($1, cgroup_context)
$1.DevicePolicy, config_parse_device_policy, 0, offsetof($1, cgroup_context.device_policy)

View File

@ -2793,21 +2793,26 @@ int config_parse_memory_limit(
void *userdata) {
CGroupContext *c = data;
uint64_t bytes;
uint64_t bytes = CGROUP_LIMIT_MAX;
int r;
if (isempty(rvalue) || streq(rvalue, "infinity")) {
c->memory_limit = (uint64_t) -1;
return 0;
if (!isempty(rvalue) && !streq(rvalue, "infinity") && !streq(rvalue, "max")) {
r = parse_size(rvalue, 1024, &bytes);
if (r < 0 || bytes < 1) {
log_syntax(unit, LOG_ERR, filename, line, r, "Memory limit '%s' invalid. Ignoring.", rvalue);
return 0;
}
}
r = parse_size(rvalue, 1024, &bytes);
if (r < 0 || bytes < 1) {
log_syntax(unit, LOG_ERR, filename, line, r, "Memory limit '%s' invalid. Ignoring.", rvalue);
return 0;
}
if (streq(lvalue, "MemoryLow"))
c->memory_low = bytes;
else if (streq(lvalue, "MemoryHigh"))
c->memory_high = bytes;
else if (streq(lvalue, "MemoryMax"))
c->memory_max = bytes;
else
c->memory_limit = bytes;
c->memory_limit = bytes;
return 0;
}

View File

@ -166,11 +166,11 @@ int bus_append_unit_property_assignment(sd_bus_message *m, const char *assignmen
r = sd_bus_message_append(m, "v", "b", r);
} else if (streq(field, "MemoryLimit")) {
} else if (STR_IN_SET(field, "MemoryLow", "MemoryHigh", "MemoryMax", "MemoryLimit")) {
uint64_t bytes;
if (isempty(eq) || streq(eq, "infinity"))
bytes = (uint64_t) -1;
if (isempty(eq) || streq(eq, "max") || streq(eq, "infinity"))
bytes = CGROUP_LIMIT_MAX;
else {
r = parse_size(eq, 1024, &bytes);
if (r < 0) {

View File

@ -3493,6 +3493,9 @@ typedef struct UnitStatusInfo {
/* CGroup */
uint64_t memory_current;
uint64_t memory_low;
uint64_t memory_high;
uint64_t memory_max;
uint64_t memory_limit;
uint64_t cpu_usage_nsec;
uint64_t tasks_current;
@ -3775,10 +3778,30 @@ static void print_status_info(
printf(" Memory: %s", format_bytes(buf, sizeof(buf), i->memory_current));
if (i->memory_limit != (uint64_t) -1)
printf(" (limit: %s)\n", format_bytes(buf, sizeof(buf), i->memory_limit));
else
printf("\n");
if (i->memory_low > 0 || i->memory_high != CGROUP_LIMIT_MAX || i->memory_max != CGROUP_LIMIT_MAX ||
i->memory_limit != CGROUP_LIMIT_MAX) {
const char *prefix = "";
printf(" (");
if (i->memory_low > 0) {
printf("%slow: %s", prefix, format_bytes(buf, sizeof(buf), i->memory_low));
prefix = " ";
}
if (i->memory_high != CGROUP_LIMIT_MAX) {
printf("%shigh: %s", prefix, format_bytes(buf, sizeof(buf), i->memory_high));
prefix = " ";
}
if (i->memory_max != CGROUP_LIMIT_MAX) {
printf("%smax: %s", prefix, format_bytes(buf, sizeof(buf), i->memory_max));
prefix = " ";
}
if (i->memory_limit != CGROUP_LIMIT_MAX) {
printf("%slimit: %s", prefix, format_bytes(buf, sizeof(buf), i->memory_limit));
prefix = " ";
}
printf(")");
}
printf("\n");
}
if (i->cpu_usage_nsec != (uint64_t) -1) {
@ -4007,6 +4030,12 @@ static int status_property(const char *name, sd_bus_message *m, UnitStatusInfo *
i->assert_timestamp = (usec_t) u;
else if (streq(name, "MemoryCurrent"))
i->memory_current = u;
else if (streq(name, "MemoryLow"))
i->memory_low = u;
else if (streq(name, "MemoryHigh"))
i->memory_high = u;
else if (streq(name, "MemoryMax"))
i->memory_max = u;
else if (streq(name, "MemoryLimit"))
i->memory_limit = u;
else if (streq(name, "TasksCurrent"))
@ -4500,6 +4529,8 @@ static int show_one(
_cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL;
UnitStatusInfo info = {
.memory_current = (uint64_t) -1,
.memory_high = CGROUP_LIMIT_MAX,
.memory_max = CGROUP_LIMIT_MAX,
.memory_limit = (uint64_t) -1,
.cpu_usage_nsec = (uint64_t) -1,
.tasks_current = (uint64_t) -1,