8146115: Improve docker container detection and resource configuration usage
Reviewed-by: dholmes, kbarrett, phh
--- a/src/hotspot/os/aix/os_aix.cpp Wed Nov 15 08:25:28 2017 -0500
+++ b/src/hotspot/os/aix/os_aix.cpp Wed Nov 15 10:34:17 2017 -0500
@@ -3545,6 +3545,14 @@
};
int os::active_processor_count() {
+ // User has overridden the number of active processors
+ if (ActiveProcessorCount > 0) {
+ log_trace(os)("active_processor_count: "
+ "active processor count set by user : %d",
+ ActiveProcessorCount);
+ return ActiveProcessorCount;
+ }
+
int online_cpus = ::sysconf(_SC_NPROCESSORS_ONLN);
assert(online_cpus > 0 && online_cpus <= processor_count(), "sanity check");
return online_cpus;
--- a/src/hotspot/os/bsd/os_bsd.cpp Wed Nov 15 08:25:28 2017 -0500
+++ b/src/hotspot/os/bsd/os_bsd.cpp Wed Nov 15 10:34:17 2017 -0500
@@ -3477,6 +3477,14 @@
}
int os::active_processor_count() {
+ // User has overridden the number of active processors
+ if (ActiveProcessorCount > 0) {
+ log_trace(os)("active_processor_count: "
+ "active processor count set by user : %d",
+ ActiveProcessorCount);
+ return ActiveProcessorCount;
+ }
+
return _processor_count;
}
--- a/src/hotspot/os/linux/globals_linux.hpp Wed Nov 15 08:25:28 2017 -0500
+++ b/src/hotspot/os/linux/globals_linux.hpp Wed Nov 15 10:34:17 2017 -0500
@@ -59,6 +59,9 @@
product(bool, UseSHM, false, \
"Use SYSV shared memory for large pages") \
\
+ product(bool, UseContainerSupport, true, \
+ "Enable detection and runtime container configuration support") \
+ \
diagnostic(bool, UseCpuAllocPath, false, \
"Use CPU_ALLOC code path in os::active_processor_count ")
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/src/hotspot/os/linux/osContainer_linux.cpp Wed Nov 15 10:34:17 2017 -0500
@@ -0,0 +1,594 @@
+/*
+ * Copyright (c) 2017, Oracle and/or its affiliates. All rights reserved.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This code is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ *
+ */
+
+#include <string.h>
+#include <math.h>
+#include <errno.h>
+#include "utilities/globalDefinitions.hpp"
+#include "memory/allocation.hpp"
+#include "runtime/os.hpp"
+#include "logging/log.hpp"
+#include "osContainer_linux.hpp"
+
+/*
+ * Warning: Some linux distros use 0x7FFFFFFFFFFFF000
+ * and others use 0x7FFFFFFFFFFFFFFF for unlimited.
+ */
+#define UNLIMITED_MEM CONST64(0x7FFFFFFFFFFFF000)
+
+#define PER_CPU_SHARES 1024
+
+bool OSContainer::_is_initialized = false;
+bool OSContainer::_is_containerized = false;
+
+class CgroupSubsystem: CHeapObj<mtInternal> {
+ friend class OSContainer;
+
+ private:
+ /* mountinfo contents */
+ char *_root;
+ char *_mount_point;
+
+ /* Constructed subsystem directory */
+ char *_path;
+
+ public:
+ CgroupSubsystem(char *root, char *mountpoint) {
+ _root = os::strdup(root);
+ _mount_point = os::strdup(mountpoint);
+ _path = NULL;
+ }
+
+ /*
+ * Set directory to subsystem specific files based
+ * on the contents of the mountinfo and cgroup files.
+ */
+ void set_subsystem_path(char *cgroup_path) {
+ char buf[MAXPATHLEN+1];
+ if (_root != NULL && cgroup_path != NULL) {
+ if (strcmp(_root, "/") == 0) {
+ int buflen;
+ strncpy(buf, _mount_point, MAXPATHLEN);
+ buf[MAXPATHLEN-1] = '\0';
+ if (strcmp(cgroup_path,"/") != 0) {
+ buflen = strlen(buf);
+ if ((buflen + strlen(cgroup_path)) > (MAXPATHLEN-1)) {
+ return;
+ }
+ strncat(buf, cgroup_path, MAXPATHLEN-buflen);
+ buf[MAXPATHLEN-1] = '\0';
+ }
+ _path = os::strdup(buf);
+ } else {
+ if (strcmp(_root, cgroup_path) == 0) {
+ strncpy(buf, _mount_point, MAXPATHLEN);
+ buf[MAXPATHLEN-1] = '\0';
+ _path = os::strdup(buf);
+ } else {
+ char *p = strstr(_root, cgroup_path);
+ if (p != NULL && p == _root) {
+ if (strlen(cgroup_path) > strlen(_root)) {
+ int buflen;
+ strncpy(buf, _mount_point, MAXPATHLEN);
+ buf[MAXPATHLEN-1] = '\0';
+ buflen = strlen(buf);
+ if ((buflen + strlen(cgroup_path)) > (MAXPATHLEN-1)) {
+ return;
+ }
+ strncat(buf, cgroup_path + strlen(_root), MAXPATHLEN-buflen);
+ buf[MAXPATHLEN-1] = '\0';
+ _path = os::strdup(buf);
+ }
+ }
+ }
+ }
+ }
+ }
+
+ char *subsystem_path() { return _path; }
+};
+
+CgroupSubsystem* memory = NULL;
+CgroupSubsystem* cpuset = NULL;
+CgroupSubsystem* cpu = NULL;
+CgroupSubsystem* cpuacct = NULL;
+
+typedef char * cptr;
+
+PRAGMA_DIAG_PUSH
+PRAGMA_FORMAT_NONLITERAL_IGNORED
+template <typename T> int subsystem_file_contents(CgroupSubsystem* c,
+ const char *filename,
+ const char *scan_fmt,
+ T returnval) {
+ FILE *fp = NULL;
+ char *p;
+ char file[MAXPATHLEN+1];
+ char buf[MAXPATHLEN+1];
+
+ if (c != NULL && c->subsystem_path() != NULL) {
+ strncpy(file, c->subsystem_path(), MAXPATHLEN);
+ file[MAXPATHLEN-1] = '\0';
+ int filelen = strlen(file);
+ if ((filelen + strlen(filename)) > (MAXPATHLEN-1)) {
+ log_debug(os, container)("File path too long %s, %s", file, filename);
+ return OSCONTAINER_ERROR;
+ }
+ strncat(file, filename, MAXPATHLEN-filelen);
+ log_trace(os, container)("Path to %s is %s", filename, file);
+ fp = fopen(file, "r");
+ if (fp != NULL) {
+ p = fgets(buf, MAXPATHLEN, fp);
+ if (p != NULL) {
+ int matched = sscanf(p, scan_fmt, returnval);
+ if (matched == 1) {
+ fclose(fp);
+ return 0;
+ } else {
+ log_debug(os, container)("Type %s not found in file %s",
+ scan_fmt , file);
+ }
+ } else {
+ log_debug(os, container)("Empty file %s", file);
+ }
+ } else {
+ log_debug(os, container)("Open of file %s failed, %s", file,
+ os::strerror(errno));
+ }
+ }
+ if (fp != NULL)
+ fclose(fp);
+ return OSCONTAINER_ERROR;
+}
+PRAGMA_DIAG_POP
+
+#define GET_CONTAINER_INFO(return_type, subsystem, filename, \
+ logstring, scan_fmt, variable) \
+ return_type variable; \
+{ \
+ int err; \
+ err = subsystem_file_contents(subsystem, \
+ filename, \
+ scan_fmt, \
+ &variable); \
+ if (err != 0) \
+ return (return_type) OSCONTAINER_ERROR; \
+ \
+ log_trace(os, container)(logstring, variable); \
+}
+
+#define GET_CONTAINER_INFO_CPTR(return_type, subsystem, filename, \
+ logstring, scan_fmt, variable, bufsize) \
+ char variable[bufsize]; \
+{ \
+ int err; \
+ err = subsystem_file_contents(subsystem, \
+ filename, \
+ scan_fmt, \
+ variable); \
+ if (err != 0) \
+ return (return_type) NULL; \
+ \
+ log_trace(os, container)(logstring, variable); \
+}
+
+/* init
+ *
+ * Initialize the container support and determine if
+ * we are running under cgroup control.
+ */
+void OSContainer::init() {
+ int mountid;
+ int parentid;
+ int major;
+ int minor;
+ FILE *mntinfo = NULL;
+ FILE *cgroup = NULL;
+ char buf[MAXPATHLEN+1];
+ char tmproot[MAXPATHLEN+1];
+ char tmpmount[MAXPATHLEN+1];
+ char tmpbase[MAXPATHLEN+1];
+ char *p;
+ jlong mem_limit;
+
+ assert(!_is_initialized, "Initializing OSContainer more than once");
+
+ _is_initialized = true;
+ _is_containerized = false;
+
+ log_trace(os, container)("OSContainer::init: Initializing Container Support");
+ if (!UseContainerSupport) {
+ log_trace(os, container)("Container Support not enabled");
+ return;
+ }
+
+ /*
+ * Find the cgroup mount point for memory and cpuset
+ * by reading /proc/self/mountinfo
+ *
+ * Example for docker:
+ * 219 214 0:29 /docker/7208cebd00fa5f2e342b1094f7bed87fa25661471a4637118e65f1c995be8a34 /sys/fs/cgroup/memory ro,nosuid,nodev,noexec,relatime - cgroup cgroup rw,memory
+ *
+ * Example for host:
+ * 34 28 0:29 / /sys/fs/cgroup/memory rw,nosuid,nodev,noexec,relatime shared:16 - cgroup cgroup rw,memory
+ */
+ mntinfo = fopen("/proc/self/mountinfo", "r");
+ if (mntinfo == NULL) {
+ log_debug(os, container)("Can't open /proc/self/mountinfo, %s",
+ os::strerror(errno));
+ return;
+ }
+
+ while ( (p = fgets(buf, MAXPATHLEN, mntinfo)) != NULL) {
+ // Look for the filesystem type and see if it's cgroup
+ char fstype[MAXPATHLEN+1];
+ fstype[0] = '\0';
+ char *s = strstr(p, " - ");
+ if (s != NULL &&
+ sscanf(s, " - %s", fstype) == 1 &&
+ strcmp(fstype, "cgroup") == 0) {
+
+ if (strstr(p, "memory") != NULL) {
+ int matched = sscanf(p, "%d %d %d:%d %s %s",
+ &mountid,
+ &parentid,
+ &major,
+ &minor,
+ tmproot,
+ tmpmount);
+ if (matched == 6) {
+ memory = new CgroupSubsystem(tmproot, tmpmount);
+ }
+ else
+ log_debug(os, container)("Incompatible str containing cgroup and memory: %s", p);
+ } else if (strstr(p, "cpuset") != NULL) {
+ int matched = sscanf(p, "%d %d %d:%d %s %s",
+ &mountid,
+ &parentid,
+ &major,
+ &minor,
+ tmproot,
+ tmpmount);
+ if (matched == 6) {
+ cpuset = new CgroupSubsystem(tmproot, tmpmount);
+ }
+ else {
+ log_debug(os, container)("Incompatible str containing cgroup and cpuset: %s", p);
+ }
+ } else if (strstr(p, "cpu,cpuacct") != NULL) {
+ int matched = sscanf(p, "%d %d %d:%d %s %s",
+ &mountid,
+ &parentid,
+ &major,
+ &minor,
+ tmproot,
+ tmpmount);
+ if (matched == 6) {
+ cpu = new CgroupSubsystem(tmproot, tmpmount);
+ cpuacct = new CgroupSubsystem(tmproot, tmpmount);
+ }
+ else {
+ log_debug(os, container)("Incompatible str containing cgroup and cpu,cpuacct: %s", p);
+ }
+ } else if (strstr(p, "cpuacct") != NULL) {
+ int matched = sscanf(p, "%d %d %d:%d %s %s",
+ &mountid,
+ &parentid,
+ &major,
+ &minor,
+ tmproot,
+ tmpmount);
+ if (matched == 6) {
+ cpuacct = new CgroupSubsystem(tmproot, tmpmount);
+ }
+ else {
+ log_debug(os, container)("Incompatible str containing cgroup and cpuacct: %s", p);
+ }
+ } else if (strstr(p, "cpu") != NULL) {
+ int matched = sscanf(p, "%d %d %d:%d %s %s",
+ &mountid,
+ &parentid,
+ &major,
+ &minor,
+ tmproot,
+ tmpmount);
+ if (matched == 6) {
+ cpu = new CgroupSubsystem(tmproot, tmpmount);
+ }
+ else {
+ log_debug(os, container)("Incompatible str containing cgroup and cpu: %s", p);
+ }
+ }
+ }
+ }
+
+ if (mntinfo != NULL) fclose(mntinfo);
+
+ /*
+ * Read /proc/self/cgroup and map host mount point to
+ * local one via /proc/self/mountinfo content above
+ *
+ * Docker example:
+ * 5:memory:/docker/6558aed8fc662b194323ceab5b964f69cf36b3e8af877a14b80256e93aecb044
+ *
+ * Host example:
+ * 5:memory:/user.slice
+ *
+ * Construct a path to the process specific memory and cpuset
+ * cgroup directory.
+ *
+ * For a container running under Docker from memory example above
+ * the paths would be:
+ *
+ * /sys/fs/cgroup/memory
+ *
+ * For a Host from memory example above the path would be:
+ *
+ * /sys/fs/cgroup/memory/user.slice
+ *
+ */
+ cgroup = fopen("/proc/self/cgroup", "r");
+ if (cgroup == NULL) {
+ log_debug(os, container)("Can't open /proc/self/cgroup, %s",
+ os::strerror(errno));
+ return;
+ }
+
+ while ( (p = fgets(buf, MAXPATHLEN, cgroup)) != NULL) {
+ int cgno;
+ int matched;
+ char *controller;
+ char *base;
+
+ /* Skip cgroup number */
+ strsep(&p, ":");
+ /* Get controller and base */
+ controller = strsep(&p, ":");
+ base = strsep(&p, "\n");
+
+ if (controller != NULL) {
+ if (strstr(controller, "memory") != NULL) {
+ memory->set_subsystem_path(base);
+ } else if (strstr(controller, "cpuset") != NULL) {
+ cpuset->set_subsystem_path(base);
+ } else if (strstr(controller, "cpu,cpuacct") != NULL) {
+ cpu->set_subsystem_path(base);
+ cpuacct->set_subsystem_path(base);
+ } else if (strstr(controller, "cpuacct") != NULL) {
+ cpuacct->set_subsystem_path(base);
+ } else if (strstr(controller, "cpu") != NULL) {
+ cpu->set_subsystem_path(base);
+ }
+ }
+ }
+
+ if (cgroup != NULL) fclose(cgroup);
+
+ if (memory == NULL || cpuset == NULL || cpu == NULL) {
+ log_debug(os, container)("Required cgroup subsystems not found");
+ return;
+ }
+
+ // We need to update the amount of physical memory now that
+ // command line arguments have been processed.
+ if ((mem_limit = memory_limit_in_bytes()) > 0) {
+ os::Linux::set_physical_memory(mem_limit);
+ }
+
+ _is_containerized = true;
+
+}
+
+char * OSContainer::container_type() {
+ if (is_containerized()) {
+ return (char *)"cgroupv1";
+ } else {
+ return NULL;
+ }
+}
+
+
+/* memory_limit_in_bytes
+ *
+ * Return the limit of available memory for this process.
+ *
+ * return:
+ * memory limit in bytes or
+ * -1 for unlimited
+ * OSCONTAINER_ERROR for not supported
+ */
+jlong OSContainer::memory_limit_in_bytes() {
+ GET_CONTAINER_INFO(jlong, memory, "/memory.limit_in_bytes",
+ "Memory Limit is: " JLONG_FORMAT, JLONG_FORMAT, memlimit);
+
+ if (memlimit >= UNLIMITED_MEM) {
+ log_trace(os, container)("Memory Limit is: Unlimited");
+ return (jlong)-1;
+ }
+ else {
+ return memlimit;
+ }
+}
+
+jlong OSContainer::memory_and_swap_limit_in_bytes() {
+ GET_CONTAINER_INFO(jlong, memory, "/memory.memsw.limit_in_bytes",
+ "Memory and Swap Limit is: " JLONG_FORMAT, JLONG_FORMAT, memswlimit);
+ if (memswlimit >= UNLIMITED_MEM) {
+ log_trace(os, container)("Memory and Swap Limit is: Unlimited");
+ return (jlong)-1;
+ } else {
+ return memswlimit;
+ }
+}
+
+jlong OSContainer::memory_soft_limit_in_bytes() {
+ GET_CONTAINER_INFO(jlong, memory, "/memory.soft_limit_in_bytes",
+ "Memory Soft Limit is: " JLONG_FORMAT, JLONG_FORMAT, memsoftlimit);
+ if (memsoftlimit >= UNLIMITED_MEM) {
+ log_trace(os, container)("Memory Soft Limit is: Unlimited");
+ return (jlong)-1;
+ } else {
+ return memsoftlimit;
+ }
+}
+
+/* memory_usage_in_bytes
+ *
+ * Return the amount of used memory for this process.
+ *
+ * return:
+ * memory usage in bytes or
+ * -1 for unlimited
+ * OSCONTAINER_ERROR for not supported
+ */
+jlong OSContainer::memory_usage_in_bytes() {
+ GET_CONTAINER_INFO(jlong, memory, "/memory.usage_in_bytes",
+ "Memory Usage is: " JLONG_FORMAT, JLONG_FORMAT, memusage);
+ return memusage;
+}
+
+/* memory_max_usage_in_bytes
+ *
+ * Return the maximum amount of used memory for this process.
+ *
+ * return:
+ * max memory usage in bytes or
+ * OSCONTAINER_ERROR for not supported
+ */
+jlong OSContainer::memory_max_usage_in_bytes() {
+ GET_CONTAINER_INFO(jlong, memory, "/memory.max_usage_in_bytes",
+ "Maximum Memory Usage is: " JLONG_FORMAT, JLONG_FORMAT, memmaxusage);
+ return memmaxusage;
+}
+
+/* active_processor_count
+ *
+ * Calculate an appropriate number of active processors for the
+ * VM to use based on these three cgroup options.
+ *
+ * cpu affinity
+ * cpu quota & cpu period
+ * cpu shares
+ *
+ * Algorithm:
+ *
+ * Determine the number of available CPUs from sched_getaffinity
+ *
+ * If user specified a quota (quota != -1), calculate the number of
+ * required CPUs by dividing quota by period.
+ *
+ * If shares are in effect (shares != -1), calculate the number
+ * of cpus required for the shares by dividing the share value
+ * by PER_CPU_SHARES.
+ *
+ * All results of division are rounded up to the next whole number.
+ *
+ * Return the smaller number from the three different settings.
+ *
+ * return:
+ * number of cpus
+ * OSCONTAINER_ERROR if failure occured during extract of cpuset info
+ */
+int OSContainer::active_processor_count() {
+ int cpu_count, share_count, quota_count;
+ int share, quota, period;
+ int result;
+
+ cpu_count = os::Linux::active_processor_count();
+
+ share = cpu_shares();
+ if (share > -1) {
+ share_count = ceilf((float)share / (float)PER_CPU_SHARES);
+ log_trace(os, container)("cpu_share count: %d", share_count);
+ } else {
+ share_count = cpu_count;
+ }
+
+ quota = cpu_quota();
+ period = cpu_period();
+ if (quota > -1 && period > 0) {
+ quota_count = ceilf((float)quota / (float)period);
+ log_trace(os, container)("quota_count: %d", quota_count);
+ } else {
+ quota_count = cpu_count;
+ }
+
+ result = MIN2(cpu_count, MIN2(share_count, quota_count));
+ log_trace(os, container)("OSContainer::active_processor_count: %d", result);
+ return result;
+}
+
+char * OSContainer::cpu_cpuset_cpus() {
+ GET_CONTAINER_INFO_CPTR(cptr, cpuset, "/cpuset.cpus",
+ "cpuset.cpus is: %s", "%1023s", cpus, 1024);
+ return os::strdup(cpus);
+}
+
+char * OSContainer::cpu_cpuset_memory_nodes() {
+ GET_CONTAINER_INFO_CPTR(cptr, cpuset, "/cpuset.mems",
+ "cpuset.mems is: %s", "%1023s", mems, 1024);
+ return os::strdup(mems);
+}
+
+/* cpu_quota
+ *
+ * Return the number of milliseconds per period
+ * process is guaranteed to run.
+ *
+ * return:
+ * quota time in milliseconds
+ * -1 for no quota
+ * OSCONTAINER_ERROR for not supported
+ */
+int OSContainer::cpu_quota() {
+ GET_CONTAINER_INFO(int, cpu, "/cpu.cfs_quota_us",
+ "CPU Quota is: %d", "%d", quota);
+ return quota;
+}
+
+int OSContainer::cpu_period() {
+ GET_CONTAINER_INFO(int, cpu, "/cpu.cfs_period_us",
+ "CPU Period is: %d", "%d", period);
+ return period;
+}
+
+/* cpu_shares
+ *
+ * Return the amount of cpu shares available to the process
+ *
+ * return:
+ * Share number (typically a number relative to 1024)
+ * (2048 typically expresses 2 CPUs worth of processing)
+ * -1 for no share setup
+ * OSCONTAINER_ERROR for not supported
+ */
+int OSContainer::cpu_shares() {
+ GET_CONTAINER_INFO(int, cpu, "/cpu.shares",
+ "CPU Shares is: %d", "%d", shares);
+ // Convert 1024 to no shares setup
+ if (shares == 1024) return -1;
+
+ return shares;
+}
+
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/src/hotspot/os/linux/osContainer_linux.hpp Wed Nov 15 10:34:17 2017 -0500
@@ -0,0 +1,68 @@
+/*
+ * Copyright (c) 2017, Oracle and/or its affiliates. All rights reserved.
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This code is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ *
+ */
+
+#ifndef OS_LINUX_VM_OSCONTAINER_LINUX_HPP
+#define OS_LINUX_VM_OSCONTAINER_LINUX_HPP
+
+#include "utilities/globalDefinitions.hpp"
+#include "utilities/macros.hpp"
+#include "memory/allocation.hpp"
+
+#define OSCONTAINER_ERROR (-2)
+
+class OSContainer: AllStatic {
+
+ private:
+ static bool _is_initialized;
+ static bool _is_containerized;
+
+ public:
+ static void init();
+ static inline bool is_containerized();
+ static char * container_type();
+
+ static jlong memory_limit_in_bytes();
+ static jlong memory_and_swap_limit_in_bytes();
+ static jlong memory_soft_limit_in_bytes();
+ static jlong memory_usage_in_bytes();
+ static jlong memory_max_usage_in_bytes();
+
+ static int active_processor_count();
+
+ static char * cpu_cpuset_cpus();
+ static char * cpu_cpuset_memory_nodes();
+
+ static int cpu_quota();
+ static int cpu_period();
+
+ static int cpu_shares();
+
+};
+
+inline bool OSContainer::is_containerized() {
+ assert(_is_initialized, "OSContainer not initialized");
+ return _is_containerized;
+}
+
+#endif // OS_LINUX_VM_OSCONTAINER_LINUX_HPP
--- a/src/hotspot/os/linux/os_linux.cpp Wed Nov 15 08:25:28 2017 -0500
+++ b/src/hotspot/os/linux/os_linux.cpp Wed Nov 15 10:34:17 2017 -0500
@@ -38,6 +38,7 @@
#include "oops/oop.inline.hpp"
#include "os_linux.inline.hpp"
#include "os_share_linux.hpp"
+#include "osContainer_linux.hpp"
#include "prims/jniFastGetField.hpp"
#include "prims/jvm_misc.hpp"
#include "runtime/arguments.hpp"
@@ -171,13 +172,52 @@
julong os::Linux::available_memory() {
// values in struct sysinfo are "unsigned long"
struct sysinfo si;
+ julong avail_mem;
+
+ if (OSContainer::is_containerized()) {
+ jlong mem_limit, mem_usage;
+ if ((mem_limit = OSContainer::memory_limit_in_bytes()) > 0) {
+ if ((mem_usage = OSContainer::memory_usage_in_bytes()) > 0) {
+ if (mem_limit > mem_usage) {
+ avail_mem = (julong)mem_limit - (julong)mem_usage;
+ } else {
+ avail_mem = 0;
+ }
+ log_trace(os)("available container memory: " JULONG_FORMAT, avail_mem);
+ return avail_mem;
+ } else {
+ log_debug(os,container)("container memory usage call failed: " JLONG_FORMAT, mem_usage);
+ }
+ } else {
+ log_debug(os,container)("container memory unlimited or failed: " JLONG_FORMAT, mem_limit);
+ }
+ }
+
sysinfo(&si);
-
- return (julong)si.freeram * si.mem_unit;
+ avail_mem = (julong)si.freeram * si.mem_unit;
+ log_trace(os)("available memory: " JULONG_FORMAT, avail_mem);
+ return avail_mem;
}
julong os::physical_memory() {
- return Linux::physical_memory();
+ if (OSContainer::is_containerized()) {
+ jlong mem_limit;
+ if ((mem_limit = OSContainer::memory_limit_in_bytes()) > 0) {
+ log_trace(os)("total container memory: " JLONG_FORMAT, mem_limit);
+ return (julong)mem_limit;
+ } else {
+ if (mem_limit == OSCONTAINER_ERROR) {
+ log_debug(os,container)("container memory limit call failed");
+ }
+ if (mem_limit == -1) {
+ log_debug(os,container)("container memory unlimited, using host value");
+ }
+ }
+ }
+
+ jlong phys_mem = Linux::physical_memory();
+ log_trace(os)("total system memory: " JLONG_FORMAT, phys_mem);
+ return phys_mem;
}
// Return true if user is running as root.
@@ -1950,6 +1990,8 @@
os::Posix::print_load_average(st);
os::Linux::print_full_memory_info(st);
+
+ os::Linux::print_container_info(st);
}
// Try to identify popular distros.
@@ -2087,6 +2129,66 @@
st->cr();
}
+void os::Linux::print_container_info(outputStream* st) {
+ if (OSContainer::is_containerized()) {
+ st->print("container (cgroup) information:\n");
+
+ char *p = OSContainer::container_type();
+ if (p == NULL)
+ st->print("container_type() failed\n");
+ else {
+ st->print("container_type: %s\n", p);
+ }
+
+ p = OSContainer::cpu_cpuset_cpus();
+ if (p == NULL)
+ st->print("cpu_cpuset_cpus() failed\n");
+ else {
+ st->print("cpu_cpuset_cpus: %s\n", p);
+ free(p);
+ }
+
+ p = OSContainer::cpu_cpuset_memory_nodes();
+ if (p < 0)
+ st->print("cpu_memory_nodes() failed\n");
+ else {
+ st->print("cpu_memory_nodes: %s\n", p);
+ free(p);
+ }
+
+ int i = OSContainer::active_processor_count();
+ if (i < 0)
+ st->print("active_processor_count() failed\n");
+ else
+ st->print("active_processor_count: %d\n", i);
+
+ i = OSContainer::cpu_quota();
+ st->print("cpu_quota: %d\n", i);
+
+ i = OSContainer::cpu_period();
+ st->print("cpu_period: %d\n", i);
+
+ i = OSContainer::cpu_shares();
+ st->print("cpu_shares: %d\n", i);
+
+ jlong j = OSContainer::memory_limit_in_bytes();
+ st->print("memory_limit_in_bytes: " JLONG_FORMAT "\n", j);
+
+ j = OSContainer::memory_and_swap_limit_in_bytes();
+ st->print("memory_and_swap_limit_in_bytes: " JLONG_FORMAT "\n", j);
+
+ j = OSContainer::memory_soft_limit_in_bytes();
+ st->print("memory_soft_limit_in_bytes: " JLONG_FORMAT "\n", j);
+
+ j = OSContainer::OSContainer::memory_usage_in_bytes();
+ st->print("memory_usage_in_bytes: " JLONG_FORMAT "\n", j);
+
+ j = OSContainer::OSContainer::memory_max_usage_in_bytes();
+ st->print("memory_max_usage_in_bytes: " JLONG_FORMAT "\n", j);
+ st->cr();
+ }
+}
+
void os::print_memory_info(outputStream* st) {
st->print("Memory:");
@@ -4798,6 +4900,10 @@
}
}
+void os::pd_init_container_support() {
+ OSContainer::init();
+}
+
// this is called _after_ the global arguments have been parsed
jint os::init_2(void) {
@@ -4946,12 +5052,12 @@
// dynamic check - see 6515172 for details.
// If anything goes wrong we fallback to returning the number of online
// processors - which can be greater than the number available to the process.
-int os::active_processor_count() {
+int os::Linux::active_processor_count() {
cpu_set_t cpus; // can represent at most 1024 (CPU_SETSIZE) processors
cpu_set_t* cpus_p = &cpus;
int cpus_size = sizeof(cpu_set_t);
- int configured_cpus = processor_count(); // upper bound on available cpus
+ int configured_cpus = os::processor_count(); // upper bound on available cpus
int cpu_count = 0;
// old build platforms may not support dynamic cpu sets
@@ -5014,10 +5120,44 @@
CPU_FREE(cpus_p);
}
- assert(cpu_count > 0 && cpu_count <= processor_count(), "sanity check");
+ assert(cpu_count > 0 && cpu_count <= os::processor_count(), "sanity check");
return cpu_count;
}
+// Determine the active processor count from one of
+// three different sources:
+//
+// 1. User option -XX:ActiveProcessorCount
+// 2. kernel os calls (sched_getaffinity or sysconf(_SC_NPROCESSORS_ONLN)
+// 3. extracted from cgroup cpu subsystem (shares and quotas)
+//
+// Option 1, if specified, will always override.
+// If the cgroup subsystem is active and configured, we
+// will return the min of the cgroup and option 2 results.
+// This is required since tools, such as numactl, that
+// alter cpu affinity do not update cgroup subsystem
+// cpuset configuration files.
+int os::active_processor_count() {
+ // User has overridden the number of active processors
+ if (ActiveProcessorCount > 0) {
+ log_trace(os)("active_processor_count: "
+ "active processor count set by user : %d",
+ ActiveProcessorCount);
+ return ActiveProcessorCount;
+ }
+
+ int active_cpus;
+ if (OSContainer::is_containerized()) {
+ active_cpus = OSContainer::active_processor_count();
+ log_trace(os)("active_processor_count: determined by OSContainer: %d",
+ active_cpus);
+ } else {
+ active_cpus = os::Linux::active_processor_count();
+ }
+
+ return active_cpus;
+}
+
void os::set_native_thread_name(const char *name) {
if (Linux::_pthread_setname_np) {
char buf [16]; // according to glibc manpage, 16 chars incl. '/0'
--- a/src/hotspot/os/linux/os_linux.hpp Wed Nov 15 08:25:28 2017 -0500
+++ b/src/hotspot/os/linux/os_linux.hpp Wed Nov 15 10:34:17 2017 -0500
@@ -32,6 +32,7 @@
class Linux {
friend class os;
+ friend class OSContainer;
friend class TestReserveMemorySpecial;
static bool libjsig_is_loaded; // libjsig that interposes sigaction(),
@@ -75,6 +76,9 @@
static julong available_memory();
static julong physical_memory() { return _physical_memory; }
+ static void set_physical_memory(julong phys_mem) { _physical_memory = phys_mem; }
+ static int active_processor_count();
+
static void initialize_system_info();
static int commit_memory_impl(char* addr, size_t bytes, bool exec);
@@ -106,6 +110,7 @@
static bool release_memory_special_huge_tlbfs(char* base, size_t bytes);
static void print_full_memory_info(outputStream* st);
+ static void print_container_info(outputStream* st);
static void print_distro_info(outputStream* st);
static void print_libversion_info(outputStream* st);
--- a/src/hotspot/os/solaris/os_solaris.cpp Wed Nov 15 08:25:28 2017 -0500
+++ b/src/hotspot/os/solaris/os_solaris.cpp Wed Nov 15 10:34:17 2017 -0500
@@ -290,6 +290,14 @@
}
int os::active_processor_count() {
+ // User has overridden the number of active processors
+ if (ActiveProcessorCount > 0) {
+ log_trace(os)("active_processor_count: "
+ "active processor count set by user : %d",
+ ActiveProcessorCount);
+ return ActiveProcessorCount;
+ }
+
int online_cpus = sysconf(_SC_NPROCESSORS_ONLN);
pid_t pid = getpid();
psetid_t pset = PS_NONE;
--- a/src/hotspot/os/windows/os_windows.cpp Wed Nov 15 08:25:28 2017 -0500
+++ b/src/hotspot/os/windows/os_windows.cpp Wed Nov 15 10:34:17 2017 -0500
@@ -723,6 +723,14 @@
}
int os::active_processor_count() {
+ // User has overridden the number of active processors
+ if (ActiveProcessorCount > 0) {
+ log_trace(os)("active_processor_count: "
+ "active processor count set by user : %d",
+ ActiveProcessorCount);
+ return ActiveProcessorCount;
+ }
+
DWORD_PTR lpProcessAffinityMask = 0;
DWORD_PTR lpSystemAffinityMask = 0;
int proc_count = processor_count();
--- a/src/hotspot/share/logging/logTag.hpp Wed Nov 15 08:25:28 2017 -0500
+++ b/src/hotspot/share/logging/logTag.hpp Wed Nov 15 10:34:17 2017 -0500
@@ -54,6 +54,7 @@
LOG_TAG(compilation) \
LOG_TAG(constraints) \
LOG_TAG(constantpool) \
+ LOG_TAG(container) \
LOG_TAG(coops) \
LOG_TAG(cpu) \
LOG_TAG(cset) \
--- a/src/hotspot/share/runtime/arguments.cpp Wed Nov 15 08:25:28 2017 -0500
+++ b/src/hotspot/share/runtime/arguments.cpp Wed Nov 15 10:34:17 2017 -0500
@@ -385,6 +385,7 @@
{ "InitialRAMFraction", JDK_Version::jdk(10), JDK_Version::undefined(), JDK_Version::undefined() },
{ "UseMembar", JDK_Version::jdk(10), JDK_Version::jdk(11), JDK_Version::jdk(12) },
{ "FastTLABRefill", JDK_Version::jdk(10), JDK_Version::jdk(11), JDK_Version::jdk(12) },
+ { "UseCGroupMemoryLimitForHeap", JDK_Version::jdk(10), JDK_Version::undefined(), JDK_Version::jdk(11) },
{ "IgnoreUnverifiableClassesDuringDump", JDK_Version::jdk(10), JDK_Version::undefined(), JDK_Version::undefined() },
// --- Deprecated alias flags (see also aliased_jvm_flags) - sorted by obsolete_in then expired_in:
@@ -2689,6 +2690,14 @@
return result;
}
+ // We need to ensure processor and memory resources have been properly
+ // configured - which may rely on arguments we just processed - before
+ // doing the final argument processing. Any argument processing that
+ // needs to know about processor and memory resources must occur after
+ // this point.
+
+ os::init_container_support();
+
// Do final processing now that all arguments have been parsed
result = finalize_vm_init_args(patch_mod_javabase);
if (result != JNI_OK) {
@@ -3364,12 +3373,6 @@
_exit_hook = CAST_TO_FN_PTR(exit_hook_t, option->extraInfo);
} else if (match_option(option, "abort")) {
_abort_hook = CAST_TO_FN_PTR(abort_hook_t, option->extraInfo);
- // -XX:+AggressiveHeap
- } else if (match_option(option, "-XX:+AggressiveHeap")) {
- jint result = set_aggressive_heap_flags();
- if (result != JNI_OK) {
- return result;
- }
// Need to keep consistency of MaxTenuringThreshold and AlwaysTenure/NeverTenure;
// and the last option wins.
} else if (match_option(option, "-XX:+NeverTenure")) {
@@ -3651,6 +3654,16 @@
return JNI_ERR;
}
+ // This must be done after all arguments have been processed
+ // and the container support has been initialized since AggressiveHeap
+ // relies on the amount of total memory available.
+ if (AggressiveHeap) {
+ jint result = set_aggressive_heap_flags();
+ if (result != JNI_OK) {
+ return result;
+ }
+ }
+
// This must be done after all arguments have been processed.
// java_compiler() true means set to "NONE" or empty.
if (java_compiler() && !xdebug_mode()) {
--- a/src/hotspot/share/runtime/globals.hpp Wed Nov 15 08:25:28 2017 -0500
+++ b/src/hotspot/share/runtime/globals.hpp Wed Nov 15 10:34:17 2017 -0500
@@ -2036,6 +2036,9 @@
"Real memory size (in bytes) used to set maximum heap size") \
range(0, 0XFFFFFFFFFFFFFFFF) \
\
+ product(bool, AggressiveHeap, false, \
+ "Optimize heap options for long-running memory intensive apps") \
+ \
product(size_t, ErgoHeapSizeLimit, 0, \
"Maximum ergonomically set heap size (in bytes); zero means use " \
"MaxRAM * MaxRAMPercentage / 100") \
@@ -2043,7 +2046,8 @@
\
experimental(bool, UseCGroupMemoryLimitForHeap, false, \
"Use CGroup memory limit as physical memory limit for heap " \
- "sizing") \
+ "sizing" \
+ "Deprecated, replaced by container support") \
\
product(uintx, MaxRAMFraction, 4, \
"Maximum fraction (1/n) of real memory used for maximum heap " \
@@ -2075,6 +2079,9 @@
"Percentage of real memory used for initial heap size") \
range(0.0, 100.0) \
\
+ product(int, ActiveProcessorCount, -1, \
+ "Specify the CPU count the VM should use and report as active") \
+ \
develop(uintx, MaxVirtMemFraction, 2, \
"Maximum fraction (1/n) of virtual memory used for ergonomically "\
"determining maximum heap size") \
--- a/src/hotspot/share/runtime/os.hpp Wed Nov 15 08:25:28 2017 -0500
+++ b/src/hotspot/share/runtime/os.hpp Wed Nov 15 10:34:17 2017 -0500
@@ -142,8 +142,16 @@
static void get_summary_os_info(char* buf, size_t buflen);
static void initialize_initial_active_processor_count();
+
+ LINUX_ONLY(static void pd_init_container_support();)
+
public:
static void init(void); // Called before command line parsing
+
+ static void init_container_support() { // Called during command line parsing.
+ LINUX_ONLY(pd_init_container_support();)
+ }
+
static void init_before_ergo(void); // Called after command line parsing
// before VM ergonomics processing.
static jint init_2(void); // Called after command line parsing
--- a/src/hotspot/share/runtime/thread.cpp Wed Nov 15 08:25:28 2017 -0500
+++ b/src/hotspot/share/runtime/thread.cpp Wed Nov 15 10:34:17 2017 -0500
@@ -3528,6 +3528,7 @@
LogConfiguration::initialize(create_vm_timer.begin_time());
// Parse arguments
+ // Note: this internally calls os::init_container_support()
jint parse_result = Arguments::parse(args);
if (parse_result != JNI_OK) return parse_result;
--- a/test/hotspot/jtreg/runtime/CommandLine/VMDeprecatedOptions.java Wed Nov 15 08:25:28 2017 -0500
+++ b/test/hotspot/jtreg/runtime/CommandLine/VMDeprecatedOptions.java Wed Nov 15 10:34:17 2017 -0500
@@ -88,9 +88,23 @@
output.shouldMatch(match);
}
+ // Deprecated experimental command line options need to be preceded on the
+ // command line by -XX:+UnlockExperimentalVMOption.
+ static void testDeprecatedExperimental(String option, String value) throws Throwable {
+ String XXoption = CommandLineOptionTest.prepareFlag(option, value);
+ ProcessBuilder processBuilder = ProcessTools.createJavaProcessBuilder(
+ CommandLineOptionTest.UNLOCK_EXPERIMENTAL_VM_OPTIONS, XXoption, "-version");
+ OutputAnalyzer output = new OutputAnalyzer(processBuilder.start());
+ // check for option deprecation message:
+ output.shouldHaveExitValue(0);
+ String match = getDeprecationString(option);
+ output.shouldMatch(match);
+ }
+
public static void main(String[] args) throws Throwable {
testDeprecated(DEPRECATED_OPTIONS); // Make sure that each deprecated option is mentioned in the output.
testDeprecatedDiagnostic("UnsyncloadClass", "false");
testDeprecatedDiagnostic("IgnoreUnverifiableClassesDuringDump", "false");
+ testDeprecatedExperimental("UseCGroupMemoryLimitForHeap", "true");
}
}