Copyright (c) 2003, Sun Microsystems, Inc. All Rights Reserved
The contents of this file are subject to the terms of the Common Development and Distribution License (the "License"). You may not use this file except in compliance with the License.
You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE or http://www.opensolaris.org/os/licensing. See the License for the specific language governing permissions and limitations under the License.
When distributing Covered Code, include this CDDL HEADER in each file and include the License file at usr/src/OPENSOLARIS.LICENSE. If applicable, add the following below this CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your own identifying information: Portions Copyright [yyyy] [name of copyright owner]
/usr/bin/ppgsz [-F] -o option[,option] cmd | -p pid...
The ppgsz utility sets the preferred page size for stack, heap, and/or other anonymous segments for the target process(es), that is, the launched cmd or the process(es) in the pid list. ppgsz stops the target process(es) while changing the page size. See memcntl(2).
The following options are supported:
-F
Force. Sets the preferred page size options(s) for target process(es) even if controlled by other process(es). Caution should be exercised when using the -F flag. See proc(1).
-o option[,option]
The options are:
heap=size
This option specifies the preferred page size for the heap of the target process(es). heap is defined to be the bss (uninitialized data) and the brk area that immediately follows the bss (see brk(2)). The preferred heap page size is set for the existing heap and for any additional heap memory allocated in the future. See NOTES.
stack=size
This option specifies the preferred page size for the stack of the target process(es). The preferred stack page size is set for the existing stack and newly allocated parts of the stack as it expands.
anon=size
This option specifies the preferred page size for all existing MAP_PRIVATE anonymous segments of the target process(es), other than heap and stack, which are large enough to fit at least one aligned page of the specified size. For the segments that are large enough, the preferred page size is set starting at the first size-aligned address in the segment. The anon preferred pagesize is not applied to MAP_PRIVATE anonymous segments created in the future. See MAP_ANON in mmap(2). Anonymous memory refers to MAP_PRIVATE pages that are not directly associated with a file in some filesystem. The ppgsz command uses memcntl(2) to set the preferred page size for anonymous segments. See MC_HAT_ADVISE in memcntl(2).
-p pid
Sets the preferred page size option(s) for the target process(es) in the process-id (pid) list following the -p option. The pid list can also consist of names in the /proc directory. Only the process owner or the super-user is permitted to set page size. cmd is interpreted if -p is not specified. ppgsz launches cmd and applies page size option(s) to the new process. The heap and stack preferred page sizes are inherited. Child process(es) created (see fork(2)) from the launched process or the target process(es) in the pid list after ppgsz completes will inherit the preferred heap and stack page sizes. The preferred page sizes of all segments are set back to the default system page size on exec(2) (see getpagesize(3C)). The preferred page size for all other anonymous segments is not inherited by children of the launched or target process(es).
Example 1 Setting the preferred heap and stack page size
The following example sets the preferred heap page size to 4M and the preferred stack page size to 512K for all ora\(emowned processes running commands that begin with ora:
example% ppgsz -o heap=4M,stack=512K -p `pgrep -u ora '^ora'`
Example 2 Setting the preferred anonymous page size
The following example sets the preferred page size of existing qualifying anonymous segments to 512k for process ID 953:
example% ppgsz -o anon=512k -p 953
If cmd is specified and successfully invoked (see exec(2)), the exit status of ppgsz will be the exit status of cmd. Otherwise, ppgsz will exit with one of the following values:
0
Successfully set preferred page size(s) for processes in the pid list.
125
An error occurred in ppgsz. Errors include: invalid argument, invalid page size(s) specified, and failure to set preferred page size(s) for one or more processes in the pid list or cmd.
126
cmd was found but could not be invoked.
127
cmd could not be found.
/proc/*
Process files.
A template link-editor mapfile for aligning bss (see NOTES).
See attributes(5) for descriptions of the following attributes:
ATTRIBUTE TYPE | ATTRIBUTE VALUE |
Interface Stability | Evolving |
ld(1), mpss.so.1(1), pagesize(1), pgrep(1), pmap(1), proc(1), brk(2), exec(2), fork(2), memcntl(2), mmap(2), sbrk(2), getpagesize(3C), proc(4), attributes(5)
Linker and Libraries Guide
Due to resource constraints, the setting of the preferred page size does not necessarily guarantee that the target process(es) will get the preferred page size. Use pmap(1) to view the actual heap and stack page sizes of the target process(es) (see pmap -s option).
Large pages are required to be mapped at addresses that are multiples of the size of the large page. Given that the heap is typically not large page aligned, the starting portions of the heap (below the first large page aligned address) are mapped with the system memory page size. See getpagesize(3C).
To provide a heap that will be mapped with a large page size, an application can be built using a link-editor (ld(1)) mapfile containing the bss segment declaration directive. Refer to the section Mapfile Option in the Linker and Libraries Guide for more details of this directive and the template mapfile provided in /usr/lib/ld/map.bssalign. Users are cautioned that an alignment specification may be machine-specific and may lose its benefit on different hardware platforms. A more flexible means of requesting the most optimal underlying page size may evolve in future releases.
mpss.so.1(1), a preloadable shared object, can also be used to set the preferred stack and/or heap page sizes.