# HG changeset patch
# User poonam
# Date 1530903013 0
# Fri Jul 06 18:50:13 2018 +0000
# Node ID 2f2d2af6fa5c44e67e0a9987f56392315a1e4b64
# Parent 95b72537801cc9946c27ad27f07e3f0790a21b08
8146115, PR3508, RH1463098: Improve docker container detection and resource configuration usage
Reviewed-by: bobv, dbuck
diff --git openjdk.orig/hotspot/src/os/aix/vm/os_aix.cpp openjdk/hotspot/src/os/aix/vm/os_aix.cpp
--- openjdk.orig/hotspot/src/os/aix/vm/os_aix.cpp
+++ openjdk/hotspot/src/os/aix/vm/os_aix.cpp
@@ -4008,6 +4008,16 @@
};
int os::active_processor_count() {
+ // User has overridden the number of active processors
+ if (ActiveProcessorCount > 0) {
+ if (PrintActiveCpus) {
+ tty->print_cr("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;
diff --git openjdk.orig/hotspot/src/os/bsd/vm/os_bsd.cpp openjdk/hotspot/src/os/bsd/vm/os_bsd.cpp
--- openjdk.orig/hotspot/src/os/bsd/vm/os_bsd.cpp
+++ openjdk/hotspot/src/os/bsd/vm/os_bsd.cpp
@@ -3770,6 +3770,16 @@
};
int os::active_processor_count() {
+ // User has overridden the number of active processors
+ if (ActiveProcessorCount > 0) {
+ if (PrintActiveCpus) {
+ tty->print_cr("active_processor_count: "
+ "active processor count set by user : %d",
+ ActiveProcessorCount);
+ }
+ return ActiveProcessorCount;
+ }
+
return _processor_count;
}
diff --git openjdk.orig/hotspot/src/os/linux/vm/globals_linux.hpp openjdk/hotspot/src/os/linux/vm/globals_linux.hpp
--- openjdk.orig/hotspot/src/os/linux/vm/globals_linux.hpp
+++ openjdk/hotspot/src/os/linux/vm/globals_linux.hpp
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2005, 2016, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2005, 2018, 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
@@ -49,8 +49,13 @@
product(bool, UseSHM, false, \
"Use SYSV shared memory for large pages") \
\
- diagnostic(bool, PrintActiveCpus, false, \
- "Print the number of CPUs detected in os::active_processor_count")
+ product(bool, UseContainerSupport, true, \
+ "Enable detection and runtime container configuration support") \
+ \
+ product(bool, PreferContainerQuotaForCPUCount, true, \
+ "Calculate the container CPU availability based on the value" \
+ " of quotas (if set), when true. Otherwise, use the CPU" \
+ " shares value, provided it is less than quota.")
//
// Defines Linux-specific default values. The flags are available on all
diff --git openjdk.orig/hotspot/src/os/linux/vm/osContainer_linux.cpp openjdk/hotspot/src/os/linux/vm/osContainer_linux.cpp
new file mode 100644
--- /dev/null
+++ openjdk/hotspot/src/os/linux/vm/osContainer_linux.cpp
@@ -0,0 +1,680 @@
+/*
+ * Copyright (c) 2017, 2018, 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 "osContainer_linux.hpp"
+
+#define PER_CPU_SHARES 1024
+
+bool OSContainer::_is_initialized = false;
+bool OSContainer::_is_containerized = false;
+julong _unlimited_memory;
+
+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) {
+ if (PrintContainerInfo) {
+ tty->print_cr("subsystem_file_contents: CgroupSubsytem* is NULL");
+ }
+ return OSCONTAINER_ERROR;
+ }
+ if (c->subsystem_path() == NULL) {
+ if (PrintContainerInfo) {
+ tty->print_cr("subsystem_file_contents: subsystem path is NULL");
+ }
+ return OSCONTAINER_ERROR;
+ }
+
+ strncpy(file, c->subsystem_path(), MAXPATHLEN);
+ file[MAXPATHLEN-1] = '\0';
+ int filelen = strlen(file);
+ if ((filelen + strlen(filename)) > (MAXPATHLEN-1)) {
+ if (PrintContainerInfo) {
+ tty->print_cr("File path too long %s, %s", file, filename);
+ }
+ return OSCONTAINER_ERROR;
+ }
+ strncat(file, filename, MAXPATHLEN-filelen);
+ if (PrintContainerInfo) {
+ tty->print_cr("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 {
+ if (PrintContainerInfo) {
+ tty->print_cr("Type %s not found in file %s", scan_fmt, file);
+ }
+ }
+ } else {
+ if (PrintContainerInfo) {
+ tty->print_cr("Empty file %s", file);
+ }
+ }
+ } else {
+ if (PrintContainerInfo) {
+ tty->print_cr("Open of file %s failed, %s", file, 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; \
+ \
+ if (PrintContainerInfo) \
+ tty->print_cr(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; \
+ \
+ if (PrintContainerInfo) \
+ tty->print_cr(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;
+
+ _unlimited_memory = (LONG_MAX / os::vm_page_size()) * os::vm_page_size();
+
+ if (PrintContainerInfo) {
+ tty->print_cr("OSContainer::init: Initializing Container Support");
+ }
+ if (!UseContainerSupport) {
+ if (PrintContainerInfo) {
+ tty->print_cr("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) {
+ if (PrintContainerInfo) {
+ tty->print_cr("Can't open /proc/self/mountinfo, %s",
+ 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
+ if (PrintContainerInfo) {
+ tty->print_cr("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 {
+ if (PrintContainerInfo) {
+ tty->print_cr("Incompatible str containing cgroup and cpuset: %s", p);
+ }
+ }
+ } else if (strstr(p, "cpu,cpuacct") != NULL || strstr(p, "cpuacct,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);
+ cpuacct = new CgroupSubsystem(tmproot, tmpmount);
+ }
+ else {
+ if (PrintContainerInfo) {
+ tty->print_cr("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 {
+ if (PrintContainerInfo) {
+ tty->print_cr("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 {
+ if (PrintContainerInfo) {
+ tty->print_cr("Incompatible str containing cgroup and cpu: %s", p);
+ }
+ }
+ }
+ }
+ }
+
+ fclose(mntinfo);
+
+ if (memory == NULL) {
+ if (PrintContainerInfo) {
+ tty->print_cr("Required cgroup memory subsystem not found");
+ }
+ return;
+ }
+ if (cpuset == NULL) {
+ if (PrintContainerInfo) {
+ tty->print_cr("Required cgroup cpuset subsystem not found");
+ }
+ return;
+ }
+ if (cpu == NULL) {
+ if (PrintContainerInfo) {
+ tty->print_cr("Required cgroup cpu subsystem not found");
+ }
+ return;
+ }
+ if (cpuacct == NULL) {
+ if (PrintContainerInfo) {
+ tty->print_cr("Required cgroup cpuacct subsystem not found");
+ }
+ return;
+ }
+
+ /*
+ * 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) {
+ if (PrintContainerInfo) {
+ tty->print_cr("Can't open /proc/self/cgroup, %s",
+ 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 || strstr(controller, "cpuacct,cpu") != 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);
+ }
+ }
+ }
+
+ fclose(cgroup);
+
+ // 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;
+
+}
+
+const char * OSContainer::container_type() {
+ if (is_containerized()) {
+ return "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(julong, memory, "/memory.limit_in_bytes",
+ "Memory Limit is: " JULONG_FORMAT, JULONG_FORMAT, memlimit);
+
+ if (memlimit >= _unlimited_memory) {
+ if (PrintContainerInfo) {
+ tty->print_cr("Memory Limit is: Unlimited");
+ }
+ return (jlong)-1;
+ }
+ else {
+ return (jlong)memlimit;
+ }
+}
+
+jlong OSContainer::memory_and_swap_limit_in_bytes() {
+ GET_CONTAINER_INFO(julong, memory, "/memory.memsw.limit_in_bytes",
+ "Memory and Swap Limit is: " JULONG_FORMAT, JULONG_FORMAT, memswlimit);
+ if (memswlimit >= _unlimited_memory) {
+ if (PrintContainerInfo) {
+ tty->print_cr("Memory and Swap Limit is: Unlimited");
+ }
+ return (jlong)-1;
+ } else {
+ return (jlong)memswlimit;
+ }
+}
+
+jlong OSContainer::memory_soft_limit_in_bytes() {
+ GET_CONTAINER_INFO(julong, memory, "/memory.soft_limit_in_bytes",
+ "Memory Soft Limit is: " JULONG_FORMAT, JULONG_FORMAT, memsoftlimit);
+ if (memsoftlimit >= _unlimited_memory) {
+ if (PrintContainerInfo) {
+ tty->print_cr("Memory Soft Limit is: Unlimited");
+ }
+ return (jlong)-1;
+ } else {
+ return (jlong)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 inputs.
+ *
+ * cpu affinity
+ * cgroup cpu quota & cpu period
+ * cgroup 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.
+ *
+ * If neither shares or quotas have been specified, return the
+ * number of active processors in the system.
+ *
+ * If both shares and quotas have been specified, the results are
+ * based on the flag PreferContainerQuotaForCPUCount. If true,
+ * return the quota value. If false return the smallest value
+ * between shares or quotas.
+ *
+ * If shares and/or quotas have been specified, the resulting number
+ * returned will never exceed the number of active processors.
+ *
+ * return:
+ * number of CPUs
+ */
+int OSContainer::active_processor_count() {
+ int quota_count = 0, share_count = 0;
+ int cpu_count, limit_count;
+ int result;
+
+ cpu_count = limit_count = os::Linux::active_processor_count();
+ int quota = cpu_quota();
+ int period = cpu_period();
+ int share = cpu_shares();
+
+ if (quota > -1 && period > 0) {
+ quota_count = ceilf((float)quota / (float)period);
+ if (PrintContainerInfo) {
+ tty->print_cr("CPU Quota count based on quota/period: %d", quota_count);
+ }
+ }
+ if (share > -1) {
+ share_count = ceilf((float)share / (float)PER_CPU_SHARES);
+ if (PrintContainerInfo) {
+ tty->print_cr("CPU Share count based on shares: %d", share_count);
+ }
+ }
+
+ // If both shares and quotas are setup results depend
+ // on flag PreferContainerQuotaForCPUCount.
+ // If true, limit CPU count to quota
+ // If false, use minimum of shares and quotas
+ if (quota_count !=0 && share_count != 0) {
+ if (PreferContainerQuotaForCPUCount) {
+ limit_count = quota_count;
+ } else {
+ limit_count = MIN2(quota_count, share_count);
+ }
+ } else if (quota_count != 0) {
+ limit_count = quota_count;
+ } else if (share_count != 0) {
+ limit_count = share_count;
+ }
+
+ result = MIN2(cpu_count, limit_count);
+ if (PrintContainerInfo) {
+ tty->print_cr("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;
+}
+
diff --git openjdk.orig/hotspot/src/os/linux/vm/osContainer_linux.hpp openjdk/hotspot/src/os/linux/vm/osContainer_linux.hpp
new file mode 100644
--- /dev/null
+++ openjdk/hotspot/src/os/linux/vm/osContainer_linux.hpp
@@ -0,0 +1,68 @@
+/*
+ * Copyright (c) 2018, 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 const 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
diff --git openjdk.orig/hotspot/src/os/linux/vm/os_linux.cpp openjdk/hotspot/src/os/linux/vm/os_linux.cpp
--- openjdk.orig/hotspot/src/os/linux/vm/os_linux.cpp
+++ openjdk/hotspot/src/os/linux/vm/os_linux.cpp
@@ -37,6 +37,7 @@
#include "mutex_linux.inline.hpp"
#include "oops/oop.inline.hpp"
#include "os_share_linux.hpp"
+#include "osContainer_linux.hpp"
#include "prims/jniFastGetField.hpp"
#include "prims/jvm.h"
#include "prims/jvm_misc.hpp"
@@ -179,13 +180,62 @@
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()) < 1) {
+ if (PrintContainerInfo) {
+ tty->print_cr("container memory limit %s: " JLONG_FORMAT ", using host value",
+ mem_limit == OSCONTAINER_ERROR ? "failed" : "unlimited", mem_limit);
+ }
+ }
+
+ if (mem_limit > 0 && (mem_usage = OSContainer::memory_usage_in_bytes()) < 1) {
+ if (PrintContainerInfo) {
+ tty->print_cr("container memory usage failed: " JLONG_FORMAT ", using host value", mem_usage);
+ }
+ }
+
+ if (mem_limit > 0 && mem_usage > 0 ) {
+ avail_mem = mem_limit > mem_usage ? (julong)mem_limit - (julong)mem_usage : 0;
+ if (PrintContainerInfo) {
+ tty->print_cr("available container memory: " JULONG_FORMAT, avail_mem);
+ }
+ return avail_mem;
+ }
+ }
+
sysinfo(&si);
-
- return (julong)si.freeram * si.mem_unit;
+ avail_mem = (julong)si.freeram * si.mem_unit;
+ if (Verbose) {
+ tty->print_cr("available memory: " JULONG_FORMAT, avail_mem);
+ }
+ return avail_mem;
}
julong os::physical_memory() {
- return Linux::physical_memory();
+ jlong phys_mem = 0;
+ if (OSContainer::is_containerized()) {
+ jlong mem_limit;
+ if ((mem_limit = OSContainer::memory_limit_in_bytes()) > 0) {
+ if (PrintContainerInfo) {
+ tty->print_cr("total container memory: " JLONG_FORMAT, mem_limit);
+ }
+ return mem_limit;
+ }
+
+ if (PrintContainerInfo) {
+ tty->print_cr("container memory limit %s: " JLONG_FORMAT ", using host value",
+ mem_limit == OSCONTAINER_ERROR ? "failed" : "unlimited", mem_limit);
+ }
+ }
+
+ phys_mem = Linux::physical_memory();
+ if (Verbose) {
+ tty->print_cr("total system memory: " JLONG_FORMAT, phys_mem);
+ }
+ return phys_mem;
}
////////////////////////////////////////////////////////////////////////////////
@@ -2129,6 +2179,8 @@
os::Posix::print_load_average(st);
os::Linux::print_full_memory_info(st);
+
+ os::Linux::print_container_info(st);
}
// Try to identify popular distros.
@@ -2194,6 +2246,57 @@
st->cr();
}
+void os::Linux::print_container_info(outputStream* st) {
+if (!OSContainer::is_containerized()) {
+ return;
+ }
+
+ st->print("container (cgroup) information:\n");
+
+ const char *p_ct = OSContainer::container_type();
+ st->print("container_type: %s\n", p_ct != NULL ? p_ct : "failed");
+
+ char *p = OSContainer::cpu_cpuset_cpus();
+ st->print("cpu_cpuset_cpus: %s\n", p != NULL ? p : "failed");
+ free(p);
+
+ p = OSContainer::cpu_cpuset_memory_nodes();
+ st->print("cpu_memory_nodes: %s\n", p != NULL ? p : "failed");
+ free(p);
+
+ int i = OSContainer::active_processor_count();
+ if (i > 0) {
+ st->print("active_processor_count: %d\n", i);
+ } else {
+ st->print("active_processor_count: failed\n");
+ }
+
+ 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:");
@@ -4966,6 +5069,10 @@
}
}
+void os::pd_init_container_support() {
+ OSContainer::init();
+}
+
// this is called _after_ the global arguments have been parsed
jint os::init_2(void)
{
@@ -5146,7 +5253,7 @@
// sched_getaffinity gives an accurate answer as it accounts for cpusets.
// 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
int cpus_size = sizeof(cpu_set_t);
int cpu_count = 0;
@@ -5164,10 +5271,48 @@
"which may exceed available processors", strerror(errno), cpu_count);
}
- 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) {
+ if (PrintActiveCpus) {
+ tty->print_cr("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();
+ if (PrintActiveCpus) {
+ tty->print_cr("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) {
// Not yet implemented.
return;
diff --git openjdk.orig/hotspot/src/os/linux/vm/os_linux.hpp openjdk/hotspot/src/os/linux/vm/os_linux.hpp
--- openjdk.orig/hotspot/src/os/linux/vm/os_linux.hpp
+++ openjdk/hotspot/src/os/linux/vm/os_linux.hpp
@@ -35,6 +35,7 @@
class Linux {
friend class os;
+ friend class OSContainer;
friend class TestReserveMemorySpecial;
// For signal-chaining
@@ -79,6 +80,9 @@
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);
@@ -116,6 +120,7 @@
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);
diff --git openjdk.orig/hotspot/src/os/solaris/vm/os_solaris.cpp openjdk/hotspot/src/os/solaris/vm/os_solaris.cpp
--- openjdk.orig/hotspot/src/os/solaris/vm/os_solaris.cpp
+++ openjdk/hotspot/src/os/solaris/vm/os_solaris.cpp
@@ -357,6 +357,16 @@
}
int os::active_processor_count() {
+ // User has overridden the number of active processors
+ if (ActiveProcessorCount > 0) {
+ if (Verbose) {
+ tty->print_cr("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;
diff --git openjdk.orig/hotspot/src/os/windows/vm/os_windows.cpp openjdk/hotspot/src/os/windows/vm/os_windows.cpp
--- openjdk.orig/hotspot/src/os/windows/vm/os_windows.cpp
+++ openjdk/hotspot/src/os/windows/vm/os_windows.cpp
@@ -716,6 +716,16 @@
#endif
int os::active_processor_count() {
+ // User has overridden the number of active processors
+ if (ActiveProcessorCount > 0) {
+ if (PrintActiveCpus) {
+ tty->print_cr("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();
diff --git openjdk.orig/hotspot/src/share/vm/runtime/arguments.cpp openjdk/hotspot/src/share/vm/runtime/arguments.cpp
--- openjdk.orig/hotspot/src/share/vm/runtime/arguments.cpp
+++ openjdk/hotspot/src/share/vm/runtime/arguments.cpp
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 1997, 2017, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2018, 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
@@ -1801,20 +1801,34 @@
}
}
+ // Convert Fraction to Precentage values
+ if (FLAG_IS_DEFAULT(MaxRAMPercentage) &&
+ !FLAG_IS_DEFAULT(MaxRAMFraction))
+ MaxRAMPercentage = 100.0 / MaxRAMFraction;
+
+ if (FLAG_IS_DEFAULT(MinRAMPercentage) &&
+ !FLAG_IS_DEFAULT(MinRAMFraction))
+ MinRAMPercentage = 100.0 / MinRAMFraction;
+
+ if (FLAG_IS_DEFAULT(InitialRAMPercentage) &&
+ !FLAG_IS_DEFAULT(InitialRAMFraction))
+ InitialRAMPercentage = 100.0 / InitialRAMFraction;
+
// If the maximum heap size has not been set with -Xmx,
// then set it as fraction of the size of physical memory,
// respecting the maximum and minimum sizes of the heap.
if (FLAG_IS_DEFAULT(MaxHeapSize)) {
- julong reasonable_max = phys_mem / MaxRAMFraction;
-
- if (phys_mem <= MaxHeapSize * MinRAMFraction) {
+ julong reasonable_max = (julong)((phys_mem * MaxRAMPercentage) / 100);
+ const julong reasonable_min = (julong)((phys_mem * MinRAMPercentage) / 100);
+ if (reasonable_min < MaxHeapSize) {
// Small physical memory, so use a minimum fraction of it for the heap
- reasonable_max = phys_mem / MinRAMFraction;
+ reasonable_max = reasonable_min;
} else {
// Not-small physical memory, so require a heap at least
// as large as MaxHeapSize
reasonable_max = MAX2(reasonable_max, (julong)MaxHeapSize);
}
+
if (!FLAG_IS_DEFAULT(ErgoHeapSizeLimit) && ErgoHeapSizeLimit != 0) {
// Limit the heap size to ErgoHeapSizeLimit
reasonable_max = MIN2(reasonable_max, (julong)ErgoHeapSizeLimit);
@@ -1856,7 +1870,7 @@
reasonable_minimum = limit_by_allocatable_memory(reasonable_minimum);
if (InitialHeapSize == 0) {
- julong reasonable_initial = phys_mem / InitialRAMFraction;
+ julong reasonable_initial = (julong)((phys_mem * InitialRAMPercentage) / 100);
reasonable_initial = MAX3(reasonable_initial, reasonable_minimum, (julong)min_heap_size());
reasonable_initial = MIN2(reasonable_initial, (julong)MaxHeapSize);
@@ -1881,6 +1895,94 @@
}
}
+// This option inspects the machine and attempts to set various
+// parameters to be optimal for long-running, memory allocation
+// intensive jobs. It is intended for machines with large
+// amounts of cpu and memory.
+jint Arguments::set_aggressive_heap_flags() {
+ // initHeapSize is needed since _initial_heap_size is 4 bytes on a 32 bit
+ // VM, but we may not be able to represent the total physical memory
+ // available (like having 8gb of memory on a box but using a 32bit VM).
+ // Thus, we need to make sure we're using a julong for intermediate
+ // calculations.
+ julong initHeapSize;
+ julong total_memory = os::physical_memory();
+
+ if (total_memory < (julong) 256 * M) {
+ jio_fprintf(defaultStream::error_stream(),
+ "You need at least 256mb of memory to use -XX:+AggressiveHeap\n");
+ vm_exit(1);
+ }
+
+ // The heap size is half of available memory, or (at most)
+ // all of possible memory less 160mb (leaving room for the OS
+ // when using ISM). This is the maximum; because adaptive sizing
+ // is turned on below, the actual space used may be smaller.
+
+ initHeapSize = MIN2(total_memory / (julong) 2,
+ total_memory - (julong) 160 * M);
+
+ initHeapSize = limit_by_allocatable_memory(initHeapSize);
+
+ if (FLAG_IS_DEFAULT(MaxHeapSize)) {
+ FLAG_SET_CMDLINE(uintx, MaxHeapSize, initHeapSize);
+ FLAG_SET_CMDLINE(uintx, InitialHeapSize, initHeapSize);
+ // Currently the minimum size and the initial heap sizes are the same.
+ set_min_heap_size(initHeapSize);
+ }
+ if (FLAG_IS_DEFAULT(NewSize)) {
+ // Make the young generation 3/8ths of the total heap.
+ FLAG_SET_CMDLINE(uintx, NewSize,
+ ((julong) MaxHeapSize / (julong) 8) * (julong) 3);
+ FLAG_SET_CMDLINE(uintx, MaxNewSize, NewSize);
+ }
+
+#ifndef _ALLBSD_SOURCE // UseLargePages is not yet supported on BSD.
+ FLAG_SET_DEFAULT(UseLargePages, true);
+#endif
+
+ // Increase some data structure sizes for efficiency
+ FLAG_SET_CMDLINE(uintx, BaseFootPrintEstimate, MaxHeapSize);
+ FLAG_SET_CMDLINE(bool, ResizeTLAB, false);
+ FLAG_SET_CMDLINE(uintx, TLABSize, 256 * K);
+
+ // See the OldPLABSize comment below, but replace 'after promotion'
+ // with 'after copying'. YoungPLABSize is the size of the survivor
+ // space per-gc-thread buffers. The default is 4kw.
+ FLAG_SET_CMDLINE(uintx, YoungPLABSize, 256 * K); // Note: this is in words
+
+ // OldPLABSize is the size of the buffers in the old gen that
+ // UseParallelGC uses to promote live data that doesn't fit in the
+ // survivor spaces. At any given time, there's one for each gc thread.
+ // The default size is 1kw. These buffers are rarely used, since the
+ // survivor spaces are usually big enough. For specjbb, however, there
+ // are occasions when there's lots of live data in the young gen
+ // and we end up promoting some of it. We don't have a definite
+ // explanation for why bumping OldPLABSize helps, but the theory
+ // is that a bigger PLAB results in retaining something like the
+ // original allocation order after promotion, which improves mutator
+ // locality. A minor effect may be that larger PLABs reduce the
+ // number of PLAB allocation events during gc. The value of 8kw
+ // was arrived at by experimenting with specjbb.
+ FLAG_SET_CMDLINE(uintx, OldPLABSize, 8 * K); // Note: this is in words
+
+ // Enable parallel GC and adaptive generation sizing
+ FLAG_SET_CMDLINE(bool, UseParallelGC, true);
+
+ // Encourage steady state memory management
+ FLAG_SET_CMDLINE(uintx, ThresholdTolerance, 100);
+
+ // This appears to improve mutator locality
+ FLAG_SET_CMDLINE(bool, ScavengeBeforeFullGC, false);
+
+ // Get around early Solaris scheduling bug
+ // (affinity vs other jobs on system)
+ // but disallow DR and offlining (5008695).
+ FLAG_SET_CMDLINE(bool, BindGCTaskThreadsToCPUs, true);
+
+ return JNI_OK;
+}
+
// This must be called after ergonomics because we want bytecode rewriting
// if the server compiler is used, or if UseSharedSpaces is disabled.
void Arguments::set_bytecode_flags() {
@@ -2644,6 +2746,14 @@
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(&scp, scp_assembly_required);
if (result != JNI_OK) {
@@ -3117,94 +3227,6 @@
_exit_hook = CAST_TO_FN_PTR(exit_hook_t, option->extraInfo);
} else if (match_option(option, "abort", &tail)) {
_abort_hook = CAST_TO_FN_PTR(abort_hook_t, option->extraInfo);
- // -XX:+AggressiveHeap
- } else if (match_option(option, "-XX:+AggressiveHeap", &tail)) {
-
- // This option inspects the machine and attempts to set various
- // parameters to be optimal for long-running, memory allocation
- // intensive jobs. It is intended for machines with large
- // amounts of cpu and memory.
-
- // initHeapSize is needed since _initial_heap_size is 4 bytes on a 32 bit
- // VM, but we may not be able to represent the total physical memory
- // available (like having 8gb of memory on a box but using a 32bit VM).
- // Thus, we need to make sure we're using a julong for intermediate
- // calculations.
- julong initHeapSize;
- julong total_memory = os::physical_memory();
-
- if (total_memory < (julong)256*M) {
- jio_fprintf(defaultStream::error_stream(),
- "You need at least 256mb of memory to use -XX:+AggressiveHeap\n");
- vm_exit(1);
- }
-
- // The heap size is half of available memory, or (at most)
- // all of possible memory less 160mb (leaving room for the OS
- // when using ISM). This is the maximum; because adaptive sizing
- // is turned on below, the actual space used may be smaller.
-
- initHeapSize = MIN2(total_memory / (julong)2,
- total_memory - (julong)160*M);
-
- initHeapSize = limit_by_allocatable_memory(initHeapSize);
-
- if (FLAG_IS_DEFAULT(MaxHeapSize)) {
- FLAG_SET_CMDLINE(uintx, MaxHeapSize, initHeapSize);
- FLAG_SET_CMDLINE(uintx, InitialHeapSize, initHeapSize);
- // Currently the minimum size and the initial heap sizes are the same.
- set_min_heap_size(initHeapSize);
- }
- if (FLAG_IS_DEFAULT(NewSize)) {
- // Make the young generation 3/8ths of the total heap.
- FLAG_SET_CMDLINE(uintx, NewSize,
- ((julong)MaxHeapSize / (julong)8) * (julong)3);
- FLAG_SET_CMDLINE(uintx, MaxNewSize, NewSize);
- }
-
-#ifndef _ALLBSD_SOURCE // UseLargePages is not yet supported on BSD.
- FLAG_SET_DEFAULT(UseLargePages, true);
-#endif
-
- // Increase some data structure sizes for efficiency
- FLAG_SET_CMDLINE(uintx, BaseFootPrintEstimate, MaxHeapSize);
- FLAG_SET_CMDLINE(bool, ResizeTLAB, false);
- FLAG_SET_CMDLINE(uintx, TLABSize, 256*K);
-
- // See the OldPLABSize comment below, but replace 'after promotion'
- // with 'after copying'. YoungPLABSize is the size of the survivor
- // space per-gc-thread buffers. The default is 4kw.
- FLAG_SET_CMDLINE(uintx, YoungPLABSize, 256*K); // Note: this is in words
-
- // OldPLABSize is the size of the buffers in the old gen that
- // UseParallelGC uses to promote live data that doesn't fit in the
- // survivor spaces. At any given time, there's one for each gc thread.
- // The default size is 1kw. These buffers are rarely used, since the
- // survivor spaces are usually big enough. For specjbb, however, there
- // are occasions when there's lots of live data in the young gen
- // and we end up promoting some of it. We don't have a definite
- // explanation for why bumping OldPLABSize helps, but the theory
- // is that a bigger PLAB results in retaining something like the
- // original allocation order after promotion, which improves mutator
- // locality. A minor effect may be that larger PLABs reduce the
- // number of PLAB allocation events during gc. The value of 8kw
- // was arrived at by experimenting with specjbb.
- FLAG_SET_CMDLINE(uintx, OldPLABSize, 8*K); // Note: this is in words
-
- // Enable parallel GC and adaptive generation sizing
- FLAG_SET_CMDLINE(bool, UseParallelGC, true);
-
- // Encourage steady state memory management
- FLAG_SET_CMDLINE(uintx, ThresholdTolerance, 100);
-
- // This appears to improve mutator locality
- FLAG_SET_CMDLINE(bool, ScavengeBeforeFullGC, false);
-
- // Get around early Solaris scheduling bug
- // (affinity vs other jobs on system)
- // but disallow DR and offlining (5008695).
- FLAG_SET_CMDLINE(bool, BindGCTaskThreadsToCPUs, true);
-
} else if (match_option(option, "-XX:+NeverTenure", &tail)) {
// The last option must always win.
FLAG_SET_CMDLINE(bool, AlwaysTenure, false);
@@ -3605,6 +3627,15 @@
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()) {
diff --git openjdk.orig/hotspot/src/share/vm/runtime/arguments.hpp openjdk/hotspot/src/share/vm/runtime/arguments.hpp
--- openjdk.orig/hotspot/src/share/vm/runtime/arguments.hpp
+++ openjdk/hotspot/src/share/vm/runtime/arguments.hpp
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 1997, 2014, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2018, 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
@@ -365,6 +365,8 @@
// Aggressive optimization flags.
static void set_aggressive_opts_flags();
+ static jint set_aggressive_heap_flags();
+
// Argument parsing
static void do_pd_flag_adjustments();
static bool parse_argument(const char* arg, Flag::Flags origin);
diff --git openjdk.orig/hotspot/src/share/vm/runtime/globals.hpp openjdk/hotspot/src/share/vm/runtime/globals.hpp
--- openjdk.orig/hotspot/src/share/vm/runtime/globals.hpp
+++ openjdk/hotspot/src/share/vm/runtime/globals.hpp
@@ -2076,13 +2076,23 @@
product_pd(uint64_t, MaxRAM, \
"Real memory size (in bytes) used to set maximum heap size") \
\
+ product(bool, AggressiveHeap, false, \
+ "Optimize heap options for long-running memory intensive apps") \
+ \
product(uintx, ErgoHeapSizeLimit, 0, \
"Maximum ergonomically set heap size (in bytes); zero means use " \
- "MaxRAM / MaxRAMFraction") \
+ "MaxRAM * MaxRAMPercentage / 100") \
\
experimental(bool, UseCGroupMemoryLimitForHeap, false, \
"Use CGroup memory limit as physical memory limit for heap " \
- "sizing") \
+ "sizing" \
+ "Deprecated, replaced by container support") \
+ \
+ diagnostic(bool, PrintContainerInfo, false, \
+ "Print container related information") \
+ \
+ diagnostic(bool, PrintActiveCpus, false, \
+ "Print the number of CPUs detected in os::active_processor_count") \
\
product(uintx, MaxRAMFraction, 4, \
"Maximum fraction (1/n) of real memory used for maximum heap " \
@@ -2099,6 +2109,19 @@
product(uintx, InitialRAMFraction, 64, \
"Fraction (1/n) of real memory used for initial heap size") \
\
+ product(double, MaxRAMPercentage, 25.0, \
+ "Maximum percentage of real memory used for maximum heap size") \
+ \
+ product(double, MinRAMPercentage, 50.0, \
+ "Minimum percentage of real memory used for maximum heap" \
+ "size on systems with small physical memory size") \
+ \
+ product(double, InitialRAMPercentage, 1.5625, \
+ "Percentage of real memory used for initial heap size") \
+ \
+ product(intx, 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") \
diff --git openjdk.orig/hotspot/src/share/vm/runtime/os.hpp openjdk/hotspot/src/share/vm/runtime/os.hpp
--- openjdk.orig/hotspot/src/share/vm/runtime/os.hpp
+++ openjdk/hotspot/src/share/vm/runtime/os.hpp
@@ -152,8 +152,16 @@
static size_t page_size_for_region(size_t region_size, size_t min_pages, bool must_be_aligned);
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
diff --git openjdk.orig/hotspot/src/share/vm/runtime/thread.cpp openjdk/hotspot/src/share/vm/runtime/thread.cpp
--- openjdk.orig/hotspot/src/share/vm/runtime/thread.cpp
+++ openjdk/hotspot/src/share/vm/runtime/thread.cpp
@@ -3332,6 +3332,7 @@
Arguments::init_version_specific_system_properties();
// Parse arguments
+ // Note: this internally calls os::init_container_support()
jint parse_result = Arguments::parse(args);
if (parse_result != JNI_OK) return parse_result;