€•      Œsphinx.addnodes”Œdocument”“”)”}”(Œ	rawsource”Œ ”Œchildren”]”(Œtranslations”ŒLanguagesNode”“”)”}”(hhh]”(h Œpending_xref”“”)”}”(hhh]”Œdocutils.nodes”ŒText”“”ŒChinese (Simplified)”…””}”Œparent”hsbaŒ
attributes”}”(Œids”]”Œclasses”]”Œnames”]”Œdupnames”]”Œbackrefs”]”Œ	refdomain”Œstd”Œreftype”Œdoc”Œ	reftarget”Œ1/translations/zh_CN/admin-guide/cgroup-v1/cpusets”Œmodname”NŒ	classname”NŒrefexplicit”ˆuŒtagname”hhhubh)”}”(hhh]”hŒChinese (Traditional)”…””}”hh2sbah}”(h]”h ]”h"]”h$]”h&]”Œ	refdomain”h)Œreftype”h+Œ	reftarget”Œ1/translations/zh_TW/admin-guide/cgroup-v1/cpusets”Œmodname”NŒ	classname”NŒrefexplicit”ˆuh1hhhubh)”}”(hhh]”hŒItalian”…””}”hhFsbah}”(h]”h ]”h"]”h$]”h&]”Œ	refdomain”h)Œreftype”h+Œ	reftarget”Œ1/translations/it_IT/admin-guide/cgroup-v1/cpusets”Œmodname”NŒ	classname”NŒrefexplicit”ˆuh1hhhubh)”}”(hhh]”hŒJapanese”…””}”hhZsbah}”(h]”h ]”h"]”h$]”h&]”Œ	refdomain”h)Œreftype”h+Œ	reftarget”Œ1/translations/ja_JP/admin-guide/cgroup-v1/cpusets”Œmodname”NŒ	classname”NŒrefexplicit”ˆuh1hhhubh)”}”(hhh]”hŒKorean”…””}”hhnsbah}”(h]”h ]”h"]”h$]”h&]”Œ	refdomain”h)Œreftype”h+Œ	reftarget”Œ1/translations/ko_KR/admin-guide/cgroup-v1/cpusets”Œmodname”NŒ	classname”NŒrefexplicit”ˆuh1hhhubh)”}”(hhh]”hŒSpanish”…””}”hh‚sbah}”(h]”h ]”h"]”h$]”h&]”Œ	refdomain”h)Œreftype”h+Œ	reftarget”Œ1/translations/sp_SP/admin-guide/cgroup-v1/cpusets”Œmodname”NŒ	classname”NŒrefexplicit”ˆuh1hhhubeh}”(h]”h ]”h"]”h$]”h&]”Œcurrent_language”ŒEnglish”uh1h
hhŒ	_document”hŒsource”NŒline”NubhŒtarget”“”)”}”(hŒ.. _cpusets:”h]”h}”(h]”h ]”h"]”h$]”h&]”Œrefid”Œcpusets”uh1h¡h KhhhžhhŸŒK/var/lib/git/docbuild/linux/Documentation/admin-guide/cgroup-v1/cpusets.rst”ubhŒsection”“”)”}”(hhh]”(hŒtitle”“”)”}”(hŒCPUSETS”h]”hŒCPUSETS”…””}”(hh·hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhh²hžhhŸh¯h KubhŒ	paragraph”“”)”}”(hŒCopyright (C) 2004 BULL SA.”h]”hŒCopyright (C) 2004 BULL SA.”…””}”(hhÇhžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Khh²hžhubhÆ)”}”(hŒWritten by Simon.Derr@bull.net”h]”(hŒWritten by ”…””}”(hhÕhžhhŸNh NubhŒ	reference”“”)”}”(hŒSimon.Derr@bull.net”h]”hŒSimon.Derr@bull.net”…””}”(hhßhžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”Œrefuri”Œmailto:Simon.Derr@bull.net”uh1hÝhhÕubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K	hh²hžhubhŒbullet_list”“”)”}”(hhh]”(hŒ	list_item”“”)”}”(hŒ7Portions Copyright (c) 2004-2006 Silicon Graphics, Inc.”h]”hÆ)”}”(hhþh]”hŒ7Portions Copyright (c) 2004-2006 Silicon Graphics, Inc.”…””}”(hj   hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Khhüubah}”(h]”h ]”h"]”h$]”h&]”uh1húhh÷hžhhŸh¯h Nubhû)”}”(hŒ%Modified by Paul Jackson <pj@sgi.com>”h]”hÆ)”}”(hj  h]”(hŒModified by Paul Jackson <”…””}”(hj  hžhhŸNh NubhÞ)”}”(hŒ
pj@sgi.com”h]”hŒ
pj@sgi.com”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”Œrefuri”Œmailto:pj@sgi.com”uh1hÝhj  ubhŒ>”…””}”(hj  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Khj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhh÷hžhhŸh¯h Nubhû)”}”(hŒ,Modified by Christoph Lameter <cl@linux.com>”h]”hÆ)”}”(hj@  h]”(hŒModified by Christoph Lameter <”…””}”(hjB  hžhhŸNh NubhÞ)”}”(hŒcl@linux.com”h]”hŒcl@linux.com”…””}”(hjI  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”Œrefuri”Œmailto:cl@linux.com”uh1hÝhjB  ubhŒ>”…””}”(hjB  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Khj>  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhh÷hžhhŸh¯h Nubhû)”}”(hŒ+Modified by Paul Menage <menage@google.com>”h]”hÆ)”}”(hjk  h]”(hŒModified by Paul Menage <”…””}”(hjm  hžhhŸNh NubhÞ)”}”(hŒmenage@google.com”h]”hŒmenage@google.com”…””}”(hjt  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”Œrefuri”Œmailto:menage@google.com”uh1hÝhjm  ubhŒ>”…””}”(hjm  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Khji  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhh÷hžhhŸh¯h Nubhû)”}”(hŒ;Modified by Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
”h]”hÆ)”}”(hŒ:Modified by Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>”h]”(hŒModified by Hidetoshi Seto <”…””}”(hj˜  hžhhŸNh NubhÞ)”}”(hŒseto.hidetoshi@jp.fujitsu.com”h]”hŒseto.hidetoshi@jp.fujitsu.com”…””}”(hj   hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”Œrefuri”Œ$mailto:seto.hidetoshi@jp.fujitsu.com”uh1hÝhj˜  ubhŒ>”…””}”(hj˜  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Khj”  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhh÷hžhhŸh¯h Nubeh}”(h]”h ]”h"]”h$]”h&]”Œbullet”Œ-”uh1hõhŸh¯h Khh²hžhubhŒcomment”“”)”}”(hXË  CONTENTS:

1. Cpusets
  1.1 What are cpusets ?
  1.2 Why are cpusets needed ?
  1.3 How are cpusets implemented ?
  1.4 What are exclusive cpusets ?
  1.5 What is memory_pressure ?
  1.6 What is memory spread ?
  1.7 What is sched_load_balance ?
  1.8 What is sched_relax_domain_level ?
  1.9 How do I use cpusets ?
2. Usage Examples and Syntax
  2.1 Basic Usage
  2.2 Adding/removing cpus
  2.3 Setting flags
  2.4 Attaching processes
3. Questions
4. Contact”h]”hXË  CONTENTS:

1. Cpusets
  1.1 What are cpusets ?
  1.2 Why are cpusets needed ?
  1.3 How are cpusets implemented ?
  1.4 What are exclusive cpusets ?
  1.5 What is memory_pressure ?
  1.6 What is memory spread ?
  1.7 What is sched_load_balance ?
  1.8 What is sched_relax_domain_level ?
  1.9 How do I use cpusets ?
2. Usage Examples and Syntax
  2.1 Basic Usage
  2.2 Adding/removing cpus
  2.3 Setting flags
  2.4 Attaching processes
3. Questions
4. Contact”…””}”hjÊ  sbah}”(h]”h ]”h"]”h$]”h&]”Œ	xml:space”Œpreserve”uh1jÈ  hh²hžhhŸh¯h K$ubh±)”}”(hhh]”(h¶)”}”(hŒ
1. Cpusets”h]”hŒ
1. Cpusets”…””}”(hjÝ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhjÚ  hžhhŸh¯h K&ubh±)”}”(hhh]”(h¶)”}”(hŒ1.1 What are cpusets ?”h]”hŒ1.1 What are cpusets ?”…””}”(hjî  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhjë  hžhhŸh¯h K)ubhÆ)”}”(hŒ¬Cpusets provide a mechanism for assigning a set of CPUs and Memory
Nodes to a set of tasks.   In this document "Memory Node" refers to
an on-line node that contains memory.”h]”hŒ°Cpusets provide a mechanism for assigning a set of CPUs and Memory
Nodes to a set of tasks.   In this document â€œMemory Nodeâ€ refers to
an on-line node that contains memory.”…””}”(hjü  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K+hjë  hžhubhÆ)”}”(hX(  Cpusets constrain the CPU and Memory placement of tasks to only
the resources within a task's current cpuset.  They form a nested
hierarchy visible in a virtual file system.  These are the essential
hooks, beyond what is already present, required to manage dynamic
job placement on large systems.”h]”hX*  Cpusets constrain the CPU and Memory placement of tasks to only
the resources within a taskâ€™s current cpuset.  They form a nested
hierarchy visible in a virtual file system.  These are the essential
hooks, beyond what is already present, required to manage dynamic
job placement on large systems.”…””}”(hj
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K/hjë  hžhubhÆ)”}”(hŒfCpusets use the generic cgroup subsystem described in
Documentation/admin-guide/cgroup-v1/cgroups.rst.”h]”hŒfCpusets use the generic cgroup subsystem described in
Documentation/admin-guide/cgroup-v1/cgroups.rst.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K5hjë  hžhubhÆ)”}”(hX  Requests by a task, using the sched_setaffinity(2) system call to
include CPUs in its CPU affinity mask, and using the mbind(2) and
set_mempolicy(2) system calls to include Memory Nodes in its memory
policy, are both filtered through that task's cpuset, filtering out any
CPUs or Memory Nodes not in that cpuset.  The scheduler will not
schedule a task on a CPU that is not allowed in its cpus_allowed
vector, and the kernel page allocator will not allocate a page on a
node that is not allowed in the requesting task's mems_allowed vector.”h]”hX   Requests by a task, using the sched_setaffinity(2) system call to
include CPUs in its CPU affinity mask, and using the mbind(2) and
set_mempolicy(2) system calls to include Memory Nodes in its memory
policy, are both filtered through that taskâ€™s cpuset, filtering out any
CPUs or Memory Nodes not in that cpuset.  The scheduler will not
schedule a task on a CPU that is not allowed in its cpus_allowed
vector, and the kernel page allocator will not allocate a page on a
node that is not allowed in the requesting taskâ€™s mems_allowed vector.”…””}”(hj&  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K8hjë  hžhubhÆ)”}”(hX0  User level code may create and destroy cpusets by name in the cgroup
virtual file system, manage the attributes and permissions of these
cpusets and which CPUs and Memory Nodes are assigned to each cpuset,
specify and query to which cpuset a task is assigned, and list the
task pids assigned to a cpuset.”h]”hX0  User level code may create and destroy cpusets by name in the cgroup
virtual file system, manage the attributes and permissions of these
cpusets and which CPUs and Memory Nodes are assigned to each cpuset,
specify and query to which cpuset a task is assigned, and list the
task pids assigned to a cpuset.”…””}”(hj4  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KAhjë  hžhubeh}”(h]”Œwhat-are-cpusets”ah ]”h"]”Œ1.1 what are cpusets ?”ah$]”h&]”uh1h°hjÚ  hžhhŸh¯h K)ubh±)”}”(hhh]”(h¶)”}”(hŒ1.2 Why are cpusets needed ?”h]”hŒ1.2 Why are cpusets needed ?”…””}”(hjM  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhjJ  hžhhŸh¯h KIubhÆ)”}”(hX  The management of large computer systems, with many processors (CPUs),
complex memory cache hierarchies and multiple Memory Nodes having
non-uniform access times (NUMA) presents additional challenges for
the efficient scheduling and memory placement of processes.”h]”hX  The management of large computer systems, with many processors (CPUs),
complex memory cache hierarchies and multiple Memory Nodes having
non-uniform access times (NUMA) presents additional challenges for
the efficient scheduling and memory placement of processes.”…””}”(hj[  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KKhjJ  hžhubhÆ)”}”(hŒËFrequently more modest sized systems can be operated with adequate
efficiency just by letting the operating system automatically share
the available CPU and Memory resources amongst the requesting tasks.”h]”hŒËFrequently more modest sized systems can be operated with adequate
efficiency just by letting the operating system automatically share
the available CPU and Memory resources amongst the requesting tasks.”…””}”(hji  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KPhjJ  hžhubhÆ)”}”(hX  But larger systems, which benefit more from careful processor and
memory placement to reduce memory access times and contention,
and which typically represent a larger investment for the customer,
can benefit from explicitly placing jobs on properly sized subsets of
the system.”h]”hX  But larger systems, which benefit more from careful processor and
memory placement to reduce memory access times and contention,
and which typically represent a larger investment for the customer,
can benefit from explicitly placing jobs on properly sized subsets of
the system.”…””}”(hjw  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KThjJ  hžhubhÆ)”}”(hŒ#This can be especially valuable on:”h]”hŒ#This can be especially valuable on:”…””}”(hj…  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KZhjJ  hžhubhŒblock_quote”“”)”}”(hŒý* Web Servers running multiple instances of the same web application,
* Servers running different applications (for instance, a web server
  and a database), or
* NUMA systems running large HPC applications with demanding
  performance characteristics.
”h]”hö)”}”(hhh]”(hû)”}”(hŒCWeb Servers running multiple instances of the same web application,”h]”hÆ)”}”(hjž  h]”hŒCWeb Servers running multiple instances of the same web application,”…””}”(hj   hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K\hjœ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj™  ubhû)”}”(hŒVServers running different applications (for instance, a web server
and a database), or”h]”hÆ)”}”(hŒVServers running different applications (for instance, a web server
and a database), or”h]”hŒVServers running different applications (for instance, a web server
and a database), or”…””}”(hj·  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K]hj³  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj™  ubhû)”}”(hŒXNUMA systems running large HPC applications with demanding
performance characteristics.
”h]”hÆ)”}”(hŒWNUMA systems running large HPC applications with demanding
performance characteristics.”h]”hŒWNUMA systems running large HPC applications with demanding
performance characteristics.”…””}”(hjÏ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K_hjË  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj™  ubeh}”(h]”h ]”h"]”h$]”h&]”jÆ  Œ*”uh1hõhŸh¯h K\hj•  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j“  hŸh¯h K\hjJ  hžhubhÆ)”}”(hŒùThese subsets, or "soft partitions" must be able to be dynamically
adjusted, as the job mix changes, without impacting other concurrently
executing jobs. The location of the running jobs pages may also be moved
when the memory locations are changed.”h]”hŒýThese subsets, or â€œsoft partitionsâ€ must be able to be dynamically
adjusted, as the job mix changes, without impacting other concurrently
executing jobs. The location of the running jobs pages may also be moved
when the memory locations are changed.”…””}”(hjð  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KbhjJ  hžhubhÆ)”}”(hX  The kernel cpuset patch provides the minimum essential kernel
mechanisms required to efficiently implement such subsets.  It
leverages existing CPU and Memory Placement facilities in the Linux
kernel to avoid any additional impact on the critical scheduler or
memory allocator code.”h]”hX  The kernel cpuset patch provides the minimum essential kernel
mechanisms required to efficiently implement such subsets.  It
leverages existing CPU and Memory Placement facilities in the Linux
kernel to avoid any additional impact on the critical scheduler or
memory allocator code.”…””}”(hjþ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KghjJ  hžhubeh}”(h]”Œwhy-are-cpusets-needed”ah ]”h"]”Œ1.2 why are cpusets needed ?”ah$]”h&]”uh1h°hjÚ  hžhhŸh¯h KIubh±)”}”(hhh]”(h¶)”}”(hŒ!1.3 How are cpusets implemented ?”h]”hŒ!1.3 How are cpusets implemented ?”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhj  hžhhŸh¯h KoubhÆ)”}”(hŒ|Cpusets provide a Linux kernel mechanism to constrain which CPUs and
Memory Nodes are used by a process or set of processes.”h]”hŒ|Cpusets provide a Linux kernel mechanism to constrain which CPUs and
Memory Nodes are used by a process or set of processes.”…””}”(hj%  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Kqhj  hžhubhÆ)”}”(hŒ½The Linux kernel already has a pair of mechanisms to specify on which
CPUs a task may be scheduled (sched_setaffinity) and on which Memory
Nodes it may obtain memory (mbind, set_mempolicy).”h]”hŒ½The Linux kernel already has a pair of mechanisms to specify on which
CPUs a task may be scheduled (sched_setaffinity) and on which Memory
Nodes it may obtain memory (mbind, set_mempolicy).”…””}”(hj3  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Kthj  hžhubhÆ)”}”(hŒ0Cpusets extends these two mechanisms as follows:”h]”hŒ0Cpusets extends these two mechanisms as follows:”…””}”(hjA  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Kxhj  hžhubj”  )”}”(hXŸ  - Cpusets are sets of allowed CPUs and Memory Nodes, known to the
  kernel.
- Each task in the system is attached to a cpuset, via a pointer
  in the task structure to a reference counted cgroup structure.
- Calls to sched_setaffinity are filtered to just those CPUs
  allowed in that task's cpuset.
- Calls to mbind and set_mempolicy are filtered to just
  those Memory Nodes allowed in that task's cpuset.
- The root cpuset contains all the systems CPUs and Memory
  Nodes.
- For any cpuset, one can define child cpusets containing a subset
  of the parents CPU and Memory Node resources.
- The hierarchy of cpusets can be mounted at /dev/cpuset, for
  browsing and manipulation from user space.
- A cpuset may be marked exclusive, which ensures that no other
  cpuset (except direct ancestors and descendants) may contain
  any overlapping CPUs or Memory Nodes.
- You can list all the tasks (by pid) attached to any cpuset.
”h]”hö)”}”(hhh]”(hû)”}”(hŒGCpusets are sets of allowed CPUs and Memory Nodes, known to the
kernel.”h]”hÆ)”}”(hŒGCpusets are sets of allowed CPUs and Memory Nodes, known to the
kernel.”h]”hŒGCpusets are sets of allowed CPUs and Memory Nodes, known to the
kernel.”…””}”(hjZ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KzhjV  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjS  ubhû)”}”(hŒ}Each task in the system is attached to a cpuset, via a pointer
in the task structure to a reference counted cgroup structure.”h]”hÆ)”}”(hŒ}Each task in the system is attached to a cpuset, via a pointer
in the task structure to a reference counted cgroup structure.”h]”hŒ}Each task in the system is attached to a cpuset, via a pointer
in the task structure to a reference counted cgroup structure.”…””}”(hjr  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K|hjn  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjS  ubhû)”}”(hŒYCalls to sched_setaffinity are filtered to just those CPUs
allowed in that task's cpuset.”h]”hÆ)”}”(hŒYCalls to sched_setaffinity are filtered to just those CPUs
allowed in that task's cpuset.”h]”hŒ[Calls to sched_setaffinity are filtered to just those CPUs
allowed in that taskâ€™s cpuset.”…””}”(hjŠ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K~hj†  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjS  ubhû)”}”(hŒgCalls to mbind and set_mempolicy are filtered to just
those Memory Nodes allowed in that task's cpuset.”h]”hÆ)”}”(hŒgCalls to mbind and set_mempolicy are filtered to just
those Memory Nodes allowed in that task's cpuset.”h]”hŒiCalls to mbind and set_mempolicy are filtered to just
those Memory Nodes allowed in that taskâ€™s cpuset.”…””}”(hj¢  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K€hjž  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjS  ubhû)”}”(hŒ?The root cpuset contains all the systems CPUs and Memory
Nodes.”h]”hÆ)”}”(hŒ?The root cpuset contains all the systems CPUs and Memory
Nodes.”h]”hŒ?The root cpuset contains all the systems CPUs and Memory
Nodes.”…””}”(hjº  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K‚hj¶  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjS  ubhû)”}”(hŒnFor any cpuset, one can define child cpusets containing a subset
of the parents CPU and Memory Node resources.”h]”hÆ)”}”(hŒnFor any cpuset, one can define child cpusets containing a subset
of the parents CPU and Memory Node resources.”h]”hŒnFor any cpuset, one can define child cpusets containing a subset
of the parents CPU and Memory Node resources.”…””}”(hjÒ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K„hjÎ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjS  ubhû)”}”(hŒfThe hierarchy of cpusets can be mounted at /dev/cpuset, for
browsing and manipulation from user space.”h]”hÆ)”}”(hŒfThe hierarchy of cpusets can be mounted at /dev/cpuset, for
browsing and manipulation from user space.”h]”hŒfThe hierarchy of cpusets can be mounted at /dev/cpuset, for
browsing and manipulation from user space.”…””}”(hjê  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K†hjæ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjS  ubhû)”}”(hŒ A cpuset may be marked exclusive, which ensures that no other
cpuset (except direct ancestors and descendants) may contain
any overlapping CPUs or Memory Nodes.”h]”hÆ)”}”(hŒ A cpuset may be marked exclusive, which ensures that no other
cpuset (except direct ancestors and descendants) may contain
any overlapping CPUs or Memory Nodes.”h]”hŒ A cpuset may be marked exclusive, which ensures that no other
cpuset (except direct ancestors and descendants) may contain
any overlapping CPUs or Memory Nodes.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Kˆhjþ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjS  ubhû)”}”(hŒ<You can list all the tasks (by pid) attached to any cpuset.
”h]”hÆ)”}”(hŒ;You can list all the tasks (by pid) attached to any cpuset.”h]”hŒ;You can list all the tasks (by pid) attached to any cpuset.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K‹hj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjS  ubeh}”(h]”h ]”h"]”h$]”h&]”jÆ  jÇ  uh1hõhŸh¯h KzhjO  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j“  hŸh¯h Kzhj  hžhubhÆ)”}”(hŒ{The implementation of cpusets requires a few, simple hooks
into the rest of the kernel, none in performance critical paths:”h]”hŒ{The implementation of cpusets requires a few, simple hooks
into the rest of the kernel, none in performance critical paths:”…””}”(hj:  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Khj  hžhubj”  )”}”(hXG  - in init/main.c, to initialize the root cpuset at system boot.
- in fork and exit, to attach and detach a task from its cpuset.
- in sched_setaffinity, to mask the requested CPUs by what's
  allowed in that task's cpuset.
- in sched.c migrate_live_tasks(), to keep migrating tasks within
  the CPUs allowed by their cpuset, if possible.
- in the mbind and set_mempolicy system calls, to mask the requested
  Memory Nodes by what's allowed in that task's cpuset.
- in page_alloc.c, to restrict memory to allowed nodes.
- in vmscan.c, to restrict page recovery to the current cpuset.
”h]”hö)”}”(hhh]”(hû)”}”(hŒ=in init/main.c, to initialize the root cpuset at system boot.”h]”hÆ)”}”(hjQ  h]”hŒ=in init/main.c, to initialize the root cpuset at system boot.”…””}”(hjS  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KhjO  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjL  ubhû)”}”(hŒ>in fork and exit, to attach and detach a task from its cpuset.”h]”hÆ)”}”(hjh  h]”hŒ>in fork and exit, to attach and detach a task from its cpuset.”…””}”(hjj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K‘hjf  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjL  ubhû)”}”(hŒYin sched_setaffinity, to mask the requested CPUs by what's
allowed in that task's cpuset.”h]”hÆ)”}”(hŒYin sched_setaffinity, to mask the requested CPUs by what's
allowed in that task's cpuset.”h]”hŒ]in sched_setaffinity, to mask the requested CPUs by whatâ€™s
allowed in that taskâ€™s cpuset.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K’hj}  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjL  ubhû)”}”(hŒnin sched.c migrate_live_tasks(), to keep migrating tasks within
the CPUs allowed by their cpuset, if possible.”h]”hÆ)”}”(hŒnin sched.c migrate_live_tasks(), to keep migrating tasks within
the CPUs allowed by their cpuset, if possible.”h]”hŒnin sched.c migrate_live_tasks(), to keep migrating tasks within
the CPUs allowed by their cpuset, if possible.”…””}”(hj™  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K”hj•  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjL  ubhû)”}”(hŒxin the mbind and set_mempolicy system calls, to mask the requested
Memory Nodes by what's allowed in that task's cpuset.”h]”hÆ)”}”(hŒxin the mbind and set_mempolicy system calls, to mask the requested
Memory Nodes by what's allowed in that task's cpuset.”h]”hŒ|in the mbind and set_mempolicy system calls, to mask the requested
Memory Nodes by whatâ€™s allowed in that taskâ€™s cpuset.”…””}”(hj±  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K–hj­  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjL  ubhû)”}”(hŒ5in page_alloc.c, to restrict memory to allowed nodes.”h]”hÆ)”}”(hjÇ  h]”hŒ5in page_alloc.c, to restrict memory to allowed nodes.”…””}”(hjÉ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K˜hjÅ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjL  ubhû)”}”(hŒ>in vmscan.c, to restrict page recovery to the current cpuset.
”h]”hÆ)”}”(hŒ=in vmscan.c, to restrict page recovery to the current cpuset.”h]”hŒ=in vmscan.c, to restrict page recovery to the current cpuset.”…””}”(hjà  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K™hjÜ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjL  ubeh}”(h]”h ]”h"]”h$]”h&]”jÆ  jÇ  uh1hõhŸh¯h KhjH  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j“  hŸh¯h Khj  hžhubhÆ)”}”(hŒþYou should mount the "cgroup" filesystem type in order to enable
browsing and modifying the cpusets presently known to the kernel.  No
new system calls are added for cpusets - all support for querying and
modifying cpusets is via this cpuset file system.”h]”hX  You should mount the â€œcgroupâ€ filesystem type in order to enable
browsing and modifying the cpusets presently known to the kernel.  No
new system calls are added for cpusets - all support for querying and
modifying cpusets is via this cpuset file system.”…””}”(hj   hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K›hj  hžhubhÆ)”}”(hŒøThe /proc/<pid>/status file for each task has four added lines,
displaying the task's cpus_allowed (on which CPUs it may be scheduled)
and mems_allowed (on which Memory Nodes it may obtain memory),
in the two formats seen in the following example::”h]”hŒùThe /proc/<pid>/status file for each task has four added lines,
displaying the taskâ€™s cpus_allowed (on which CPUs it may be scheduled)
and mems_allowed (on which Memory Nodes it may obtain memory),
in the two formats seen in the following example:”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K hj  hžhubhŒliteral_block”“”)”}”(hŒCpus_allowed:   ffffffff,ffffffff,ffffffff,ffffffff
Cpus_allowed_list:      0-127
Mems_allowed:   ffffffff,ffffffff
Mems_allowed_list:      0-63”h]”hŒCpus_allowed:   ffffffff,ffffffff,ffffffff,ffffffff
Cpus_allowed_list:      0-127
Mems_allowed:   ffffffff,ffffffff
Mems_allowed_list:      0-63”…””}”hj  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h K¥hj  hžhubhÆ)”}”(hŒ Each cpuset is represented by a directory in the cgroup file system
containing (on top of the standard cgroup files) the following
files describing that cpuset:”h]”hŒ Each cpuset is represented by a directory in the cgroup file system
containing (on top of the standard cgroup files) the following
files describing that cpuset:”…””}”(hj,  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Kªhj  hžhubj”  )”}”(hXÓ  - cpuset.cpus: list of CPUs in that cpuset
- cpuset.mems: list of Memory Nodes in that cpuset
- cpuset.memory_migrate flag: if set, move pages to cpusets nodes
- cpuset.cpu_exclusive flag: is cpu placement exclusive?
- cpuset.mem_exclusive flag: is memory placement exclusive?
- cpuset.mem_hardwall flag:  is memory allocation hardwalled
- cpuset.memory_pressure: measure of how much paging pressure in cpuset
- cpuset.memory_spread_page flag: if set, spread page cache evenly on allowed nodes
- cpuset.memory_spread_slab flag: OBSOLETE. Doesn't have any function.
- cpuset.sched_load_balance flag: if set, load balance within CPUs on that cpuset
- cpuset.sched_relax_domain_level: the searching range when migrating tasks
”h]”hö)”}”(hhh]”(hû)”}”(hŒ(cpuset.cpus: list of CPUs in that cpuset”h]”hÆ)”}”(hjC  h]”hŒ(cpuset.cpus: list of CPUs in that cpuset”…””}”(hjE  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K®hjA  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj>  ubhû)”}”(hŒ0cpuset.mems: list of Memory Nodes in that cpuset”h]”hÆ)”}”(hjZ  h]”hŒ0cpuset.mems: list of Memory Nodes in that cpuset”…””}”(hj\  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K¯hjX  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj>  ubhû)”}”(hŒ?cpuset.memory_migrate flag: if set, move pages to cpusets nodes”h]”hÆ)”}”(hjq  h]”hŒ?cpuset.memory_migrate flag: if set, move pages to cpusets nodes”…””}”(hjs  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K°hjo  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj>  ubhû)”}”(hŒ6cpuset.cpu_exclusive flag: is cpu placement exclusive?”h]”hÆ)”}”(hjˆ  h]”hŒ6cpuset.cpu_exclusive flag: is cpu placement exclusive?”…””}”(hjŠ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K±hj†  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj>  ubhû)”}”(hŒ9cpuset.mem_exclusive flag: is memory placement exclusive?”h]”hÆ)”}”(hjŸ  h]”hŒ9cpuset.mem_exclusive flag: is memory placement exclusive?”…””}”(hj¡  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K²hj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj>  ubhû)”}”(hŒ:cpuset.mem_hardwall flag:  is memory allocation hardwalled”h]”hÆ)”}”(hj¶  h]”hŒ:cpuset.mem_hardwall flag:  is memory allocation hardwalled”…””}”(hj¸  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K³hj´  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj>  ubhû)”}”(hŒEcpuset.memory_pressure: measure of how much paging pressure in cpuset”h]”hÆ)”}”(hjÍ  h]”hŒEcpuset.memory_pressure: measure of how much paging pressure in cpuset”…””}”(hjÏ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K´hjË  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj>  ubhû)”}”(hŒQcpuset.memory_spread_page flag: if set, spread page cache evenly on allowed nodes”h]”hÆ)”}”(hjä  h]”hŒQcpuset.memory_spread_page flag: if set, spread page cache evenly on allowed nodes”…””}”(hjæ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Kµhjâ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj>  ubhû)”}”(hŒDcpuset.memory_spread_slab flag: OBSOLETE. Doesn't have any function.”h]”hÆ)”}”(hjû  h]”hŒFcpuset.memory_spread_slab flag: OBSOLETE. Doesnâ€™t have any function.”…””}”(hjý  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K¶hjù  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj>  ubhû)”}”(hŒOcpuset.sched_load_balance flag: if set, load balance within CPUs on that cpuset”h]”hÆ)”}”(hj  h]”hŒOcpuset.sched_load_balance flag: if set, load balance within CPUs on that cpuset”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K·hj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj>  ubhû)”}”(hŒJcpuset.sched_relax_domain_level: the searching range when migrating tasks
”h]”hÆ)”}”(hŒIcpuset.sched_relax_domain_level: the searching range when migrating tasks”h]”hŒIcpuset.sched_relax_domain_level: the searching range when migrating tasks”…””}”(hj+  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K¸hj'  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj>  ubeh}”(h]”h ]”h"]”h$]”h&]”jÆ  jÇ  uh1hõhŸh¯h K®hj:  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j“  hŸh¯h K®hj  hžhubhÆ)”}”(hŒ9In addition, only the root cpuset has the following file:”h]”hŒ9In addition, only the root cpuset has the following file:”…””}”(hjK  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Kºhj  hžhubj”  )”}”(hŒ@- cpuset.memory_pressure_enabled flag: compute memory_pressure?
”h]”hö)”}”(hhh]”hû)”}”(hŒ>cpuset.memory_pressure_enabled flag: compute memory_pressure?
”h]”hÆ)”}”(hŒ=cpuset.memory_pressure_enabled flag: compute memory_pressure?”h]”hŒ=cpuset.memory_pressure_enabled flag: compute memory_pressure?”…””}”(hjd  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K¼hj`  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj]  ubah}”(h]”h ]”h"]”h$]”h&]”jÆ  jÇ  uh1hõhŸh¯h K¼hjY  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j“  hŸh¯h K¼hj  hžhubhÆ)”}”(hX  New cpusets are created using the mkdir system call or shell
command.  The properties of a cpuset, such as its flags, allowed
CPUs and Memory Nodes, and attached tasks, are modified by writing
to the appropriate file in that cpusets directory, as listed above.”h]”hX  New cpusets are created using the mkdir system call or shell
command.  The properties of a cpuset, such as its flags, allowed
CPUs and Memory Nodes, and attached tasks, are modified by writing
to the appropriate file in that cpusets directory, as listed above.”…””}”(hj„  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K¾hj  hžhubhÆ)”}”(hŒThe named hierarchical structure of nested cpusets allows partitioning
a large system into nested, dynamically changeable, "soft-partitions".”h]”hŒ‘The named hierarchical structure of nested cpusets allows partitioning
a large system into nested, dynamically changeable, â€œsoft-partitionsâ€.”…””}”(hj’  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KÃhj  hžhubhÆ)”}”(hX‰  The attachment of each task, automatically inherited at fork by any
children of that task, to a cpuset allows organizing the work load
on a system into related sets of tasks such that each set is constrained
to using the CPUs and Memory Nodes of a particular cpuset.  A task
may be re-attached to any other cpuset, if allowed by the permissions
on the necessary cpuset file system directories.”h]”hX‰  The attachment of each task, automatically inherited at fork by any
children of that task, to a cpuset allows organizing the work load
on a system into related sets of tasks such that each set is constrained
to using the CPUs and Memory Nodes of a particular cpuset.  A task
may be re-attached to any other cpuset, if allowed by the permissions
on the necessary cpuset file system directories.”…””}”(hj   hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KÆhj  hžhubhÆ)”}”(hŒÉSuch management of a system "in the large" integrates smoothly with
the detailed placement done on individual tasks and memory regions
using the sched_setaffinity, mbind and set_mempolicy system calls.”h]”hŒÍSuch management of a system â€œin the largeâ€ integrates smoothly with
the detailed placement done on individual tasks and memory regions
using the sched_setaffinity, mbind and set_mempolicy system calls.”…””}”(hj®  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KÍhj  hžhubhÆ)”}”(hŒ)The following rules apply to each cpuset:”h]”hŒ)The following rules apply to each cpuset:”…””}”(hj¼  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KÑhj  hžhubj”  )”}”(hŒ¹- Its CPUs and Memory Nodes must be a subset of its parents.
- It can't be marked exclusive unless its parent is.
- If its cpu or memory is exclusive, they may not overlap any sibling.
”h]”hö)”}”(hhh]”(hû)”}”(hŒ:Its CPUs and Memory Nodes must be a subset of its parents.”h]”hÆ)”}”(hjÓ  h]”hŒ:Its CPUs and Memory Nodes must be a subset of its parents.”…””}”(hjÕ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KÓhjÑ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjÎ  ubhû)”}”(hŒ2It can't be marked exclusive unless its parent is.”h]”hÆ)”}”(hjê  h]”hŒ4It canâ€™t be marked exclusive unless its parent is.”…””}”(hjì  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KÔhjè  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjÎ  ubhû)”}”(hŒEIf its cpu or memory is exclusive, they may not overlap any sibling.
”h]”hÆ)”}”(hŒDIf its cpu or memory is exclusive, they may not overlap any sibling.”h]”hŒDIf its cpu or memory is exclusive, they may not overlap any sibling.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KÕhjÿ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjÎ  ubeh}”(h]”h ]”h"]”h$]”h&]”jÆ  jÇ  uh1hõhŸh¯h KÓhjÊ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j“  hŸh¯h KÓhj  hžhubhÆ)”}”(hXš  These rules, and the natural hierarchy of cpusets, enable efficient
enforcement of the exclusive guarantee, without having to scan all
cpusets every time any of them change to ensure nothing overlaps a
exclusive cpuset.  Also, the use of a Linux virtual file system (vfs)
to represent the cpuset hierarchy provides for a familiar permission
and name space for cpusets, with a minimum of additional kernel code.”h]”hXš  These rules, and the natural hierarchy of cpusets, enable efficient
enforcement of the exclusive guarantee, without having to scan all
cpusets every time any of them change to ensure nothing overlaps a
exclusive cpuset.  Also, the use of a Linux virtual file system (vfs)
to represent the cpuset hierarchy provides for a familiar permission
and name space for cpusets, with a minimum of additional kernel code.”…””}”(hj#  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K×hj  hžhubhÆ)”}”(hX5  The cpus and mems files in the root (top_cpuset) cpuset are
read-only.  The cpus file automatically tracks the value of
cpu_online_mask using a CPU hotplug notifier, and the mems file
automatically tracks the value of node_states[N_MEMORY]--i.e.,
nodes with memory--using the cpuset_track_online_nodes() hook.”h]”hX5  The cpus and mems files in the root (top_cpuset) cpuset are
read-only.  The cpus file automatically tracks the value of
cpu_online_mask using a CPU hotplug notifier, and the mems file
automatically tracks the value of node_states[N_MEMORY]--i.e.,
nodes with memory--using the cpuset_track_online_nodes() hook.”…””}”(hj1  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KÞhj  hžhubhÆ)”}”(hX„  The cpuset.effective_cpus and cpuset.effective_mems files are
normally read-only copies of cpuset.cpus and cpuset.mems files
respectively.  If the cpuset cgroup filesystem is mounted with the
special "cpuset_v2_mode" option, the behavior of these files will become
similar to the corresponding files in cpuset v2.  In other words, hotplug
events will not change cpuset.cpus and cpuset.mems.  Those events will
only affect cpuset.effective_cpus and cpuset.effective_mems which show
the actual cpus and memory nodes that are currently used by this cpuset.
See Documentation/admin-guide/cgroup-v2.rst for more information about
cpuset v2 behavior.”h]”hXˆ  The cpuset.effective_cpus and cpuset.effective_mems files are
normally read-only copies of cpuset.cpus and cpuset.mems files
respectively.  If the cpuset cgroup filesystem is mounted with the
special â€œcpuset_v2_modeâ€ option, the behavior of these files will become
similar to the corresponding files in cpuset v2.  In other words, hotplug
events will not change cpuset.cpus and cpuset.mems.  Those events will
only affect cpuset.effective_cpus and cpuset.effective_mems which show
the actual cpus and memory nodes that are currently used by this cpuset.
See Documentation/admin-guide/cgroup-v2.rst for more information about
cpuset v2 behavior.”…””}”(hj?  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Kähj  hžhubeh}”(h]”Œhow-are-cpusets-implemented”ah ]”h"]”Œ!1.3 how are cpusets implemented ?”ah$]”h&]”uh1h°hjÚ  hžhhŸh¯h Koubh±)”}”(hhh]”(h¶)”}”(hŒ 1.4 What are exclusive cpusets ?”h]”hŒ 1.4 What are exclusive cpusets ?”…””}”(hjX  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhjU  hžhhŸh¯h KñubhÆ)”}”(hŒ‘If a cpuset is cpu or mem exclusive, no other cpuset, other than
a direct ancestor or descendant, may share any of the same CPUs or
Memory Nodes.”h]”hŒ‘If a cpuset is cpu or mem exclusive, no other cpuset, other than
a direct ancestor or descendant, may share any of the same CPUs or
Memory Nodes.”…””}”(hjf  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h KóhjU  hžhubhÆ)”}”(hX  A cpuset that is cpuset.mem_exclusive *or* cpuset.mem_hardwall is "hardwalled",
i.e. it restricts kernel allocations for page, buffer and other data
commonly shared by the kernel across multiple users.  All cpusets,
whether hardwalled or not, restrict allocations of memory for user
space.  This enables configuring a system so that several independent
jobs can share common kernel data, such as file system pages, while
isolating each job's user allocation in its own cpuset.  To do this,
construct a large mem_exclusive cpuset to hold all the jobs, and
construct child, non-mem_exclusive cpusets for each individual job.
Only a small amount of typical kernel memory, such as requests from
interrupt handlers, is allowed to be taken outside even a
mem_exclusive cpuset.”h]”(hŒ&A cpuset that is cpuset.mem_exclusive ”…””}”(hjt  hžhhŸNh NubhŒemphasis”“”)”}”(hŒ*or*”h]”hŒor”…””}”(hj~  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1j|  hjt  ubhXÞ   cpuset.mem_hardwall is â€œhardwalledâ€,
i.e. it restricts kernel allocations for page, buffer and other data
commonly shared by the kernel across multiple users.  All cpusets,
whether hardwalled or not, restrict allocations of memory for user
space.  This enables configuring a system so that several independent
jobs can share common kernel data, such as file system pages, while
isolating each jobâ€™s user allocation in its own cpuset.  To do this,
construct a large mem_exclusive cpuset to hold all the jobs, and
construct child, non-mem_exclusive cpusets for each individual job.
Only a small amount of typical kernel memory, such as requests from
interrupt handlers, is allowed to be taken outside even a
mem_exclusive cpuset.”…””}”(hjt  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h K÷hjU  hžhubeh}”(h]”Œwhat-are-exclusive-cpusets”ah ]”h"]”Œ 1.4 what are exclusive cpusets ?”ah$]”h&]”uh1h°hjÚ  hžhhŸh¯h Kñubh±)”}”(hhh]”(h¶)”}”(hŒ1.5 What is memory_pressure ?”h]”hŒ1.5 What is memory_pressure ?”…””}”(hj¡  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhjž  hžhhŸh¯h MubhÆ)”}”(hŒÔThe memory_pressure of a cpuset provides a simple per-cpuset metric
of the rate that the tasks in a cpuset are attempting to free up in
use memory on the nodes of the cpuset to satisfy additional memory
requests.”h]”hŒÔThe memory_pressure of a cpuset provides a simple per-cpuset metric
of the rate that the tasks in a cpuset are attempting to free up in
use memory on the nodes of the cpuset to satisfy additional memory
requests.”…””}”(hj¯  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhjž  hžhubhÆ)”}”(hŒ‘This enables batch managers monitoring jobs running in dedicated
cpusets to efficiently detect what level of memory pressure that job
is causing.”h]”hŒ‘This enables batch managers monitoring jobs running in dedicated
cpusets to efficiently detect what level of memory pressure that job
is causing.”…””}”(hj½  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhjž  hžhubhÆ)”}”(hX¢  This is useful both on tightly managed systems running a wide mix of
submitted jobs, which may choose to terminate or re-prioritize jobs that
are trying to use more memory than allowed on the nodes assigned to them,
and with tightly coupled, long running, massively parallel scientific
computing jobs that will dramatically fail to meet required performance
goals if they start to use more memory than allowed to them.”h]”hX¢  This is useful both on tightly managed systems running a wide mix of
submitted jobs, which may choose to terminate or re-prioritize jobs that
are trying to use more memory than allowed on the nodes assigned to them,
and with tightly coupled, long running, massively parallel scientific
computing jobs that will dramatically fail to meet required performance
goals if they start to use more memory than allowed to them.”…””}”(hjË  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhjž  hžhubhÆ)”}”(hŒÕThis mechanism provides a very economical way for the batch manager
to monitor a cpuset for signs of memory pressure.  It's up to the
batch manager or other user code to decide what to do about it and
take action.”h]”hŒ×This mechanism provides a very economical way for the batch manager
to monitor a cpuset for signs of memory pressure.  Itâ€™s up to the
batch manager or other user code to decide what to do about it and
take action.”…””}”(hjÙ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhjž  hžhubhŒdefinition_list”“”)”}”(hhh]”hŒdefinition_list_item”“”)”}”(hXA  ==>
Unless this feature is enabled by writing "1" to the special file
/dev/cpuset/memory_pressure_enabled, the hook in the rebalance
code of __alloc_pages() for this metric reduces to simply noticing
that the cpuset_memory_pressure_enabled flag is zero.  So only
systems that enable this feature will compute the metric.
”h]”(hŒterm”“”)”}”(hŒ==>”h]”hŒ==>”…””}”(hjô  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jò  hŸh¯h M!hjî  ubhŒ
definition”“”)”}”(hhh]”hÆ)”}”(hX<  Unless this feature is enabled by writing "1" to the special file
/dev/cpuset/memory_pressure_enabled, the hook in the rebalance
code of __alloc_pages() for this metric reduces to simply noticing
that the cpuset_memory_pressure_enabled flag is zero.  So only
systems that enable this feature will compute the metric.”h]”hX@  Unless this feature is enabled by writing â€œ1â€ to the special file
/dev/cpuset/memory_pressure_enabled, the hook in the rebalance
code of __alloc_pages() for this metric reduces to simply noticing
that the cpuset_memory_pressure_enabled flag is zero.  So only
systems that enable this feature will compute the metric.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j  hjî  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jì  hŸh¯h M!hjé  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jç  hjž  hžhhŸh¯h NubhÆ)”}”(hŒ"Why a per-cpuset, running average:”h]”hŒ"Why a per-cpuset, running average:”…””}”(hj'  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M#hjž  hžhubj”  )”}”(hXÙ  Because this meter is per-cpuset, rather than per-task or mm,
the system load imposed by a batch scheduler monitoring this
metric is sharply reduced on large systems, because a scan of
the tasklist can be avoided on each set of queries.

Because this meter is a running average, instead of an accumulating
counter, a batch scheduler can detect memory pressure with a
single read, instead of having to read and accumulate results
for a period of time.

Because this meter is per-cpuset rather than per-task or mm,
the batch scheduler can obtain the key information, memory
pressure in a cpuset, with a single read, rather than having to
query and accumulate results over all the (dynamically changing)
set of tasks in the cpuset.
”h]”(hÆ)”}”(hŒìBecause this meter is per-cpuset, rather than per-task or mm,
the system load imposed by a batch scheduler monitoring this
metric is sharply reduced on large systems, because a scan of
the tasklist can be avoided on each set of queries.”h]”hŒìBecause this meter is per-cpuset, rather than per-task or mm,
the system load imposed by a batch scheduler monitoring this
metric is sharply reduced on large systems, because a scan of
the tasklist can be avoided on each set of queries.”…””}”(hj9  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M%hj5  ubhÆ)”}”(hŒÔBecause this meter is a running average, instead of an accumulating
counter, a batch scheduler can detect memory pressure with a
single read, instead of having to read and accumulate results
for a period of time.”h]”hŒÔBecause this meter is a running average, instead of an accumulating
counter, a batch scheduler can detect memory pressure with a
single read, instead of having to read and accumulate results
for a period of time.”…””}”(hjG  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M*hj5  ubhÆ)”}”(hX  Because this meter is per-cpuset rather than per-task or mm,
the batch scheduler can obtain the key information, memory
pressure in a cpuset, with a single read, rather than having to
query and accumulate results over all the (dynamically changing)
set of tasks in the cpuset.”h]”hX  Because this meter is per-cpuset rather than per-task or mm,
the batch scheduler can obtain the key information, memory
pressure in a cpuset, with a single read, rather than having to
query and accumulate results over all the (dynamically changing)
set of tasks in the cpuset.”…””}”(hjU  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M/hj5  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1j“  hŸh¯h M%hjž  hžhubhÆ)”}”(hŒÊA per-cpuset simple digital filter (requires a spinlock and 3 words
of data per-cpuset) is kept, and updated by any task attached to that
cpuset, if it enters the synchronous (direct) page reclaim code.”h]”hŒÊA per-cpuset simple digital filter (requires a spinlock and 3 words
of data per-cpuset) is kept, and updated by any task attached to that
cpuset, if it enters the synchronous (direct) page reclaim code.”…””}”(hji  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M5hjž  hžhubhÆ)”}”(hŒÕA per-cpuset file provides an integer number representing the recent
(half-life of 10 seconds) rate of direct page reclaims caused by
the tasks in the cpuset, in units of reclaims attempted per second,
times 1000.”h]”hŒÕA per-cpuset file provides an integer number representing the recent
(half-life of 10 seconds) rate of direct page reclaims caused by
the tasks in the cpuset, in units of reclaims attempted per second,
times 1000.”…””}”(hjw  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M9hjž  hžhubeh}”(h]”Œwhat-is-memory-pressure”ah ]”h"]”Œ1.5 what is memory_pressure ?”ah$]”h&]”uh1h°hjÚ  hžhhŸh¯h Mubh±)”}”(hhh]”(h¶)”}”(hŒ1.6 What is memory spread ?”h]”hŒ1.6 What is memory spread ?”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhj  hžhhŸh¯h M@ubhÆ)”}”(hŒêThere are two boolean flag files per cpuset that control where the
kernel allocates pages for the file system buffers and related in
kernel data structures.  They are called 'cpuset.memory_spread_page' and
'cpuset.memory_spread_slab'.”h]”hŒòThere are two boolean flag files per cpuset that control where the
kernel allocates pages for the file system buffers and related in
kernel data structures.  They are called â€˜cpuset.memory_spread_pageâ€™ and
â€˜cpuset.memory_spread_slabâ€™.”…””}”(hjž  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MAhj  hžhubhÆ)”}”(hX  If the per-cpuset boolean flag file 'cpuset.memory_spread_page' is set, then
the kernel will spread the file system buffers (page cache) evenly
over all the nodes that the faulting task is allowed to use, instead
of preferring to put those pages on the node where the task is running.”h]”hX   If the per-cpuset boolean flag file â€˜cpuset.memory_spread_pageâ€™ is set, then
the kernel will spread the file system buffers (page cache) evenly
over all the nodes that the faulting task is allowed to use, instead
of preferring to put those pages on the node where the task is running.”…””}”(hj¬  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MFhj  hžhubhÆ)”}”(hX=  If the per-cpuset boolean flag file 'cpuset.memory_spread_slab' is set,
then the kernel will spread some file system related slab caches,
such as for inodes and dentries evenly over all the nodes that the
faulting task is allowed to use, instead of preferring to put those
pages on the node where the task is running.”h]”hXA  If the per-cpuset boolean flag file â€˜cpuset.memory_spread_slabâ€™ is set,
then the kernel will spread some file system related slab caches,
such as for inodes and dentries evenly over all the nodes that the
faulting task is allowed to use, instead of preferring to put those
pages on the node where the task is running.”…””}”(hjº  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MKhj  hžhubhÆ)”}”(hŒcThe setting of these flags does not affect anonymous data segment or
stack segment pages of a task.”h]”hŒcThe setting of these flags does not affect anonymous data segment or
stack segment pages of a task.”…””}”(hjÈ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MQhj  hžhubhÆ)”}”(hX
  By default, both kinds of memory spreading are off, and memory
pages are allocated on the node local to where the task is running,
except perhaps as modified by the task's NUMA mempolicy or cpuset
configuration, so long as sufficient free memory pages are available.”h]”hX  By default, both kinds of memory spreading are off, and memory
pages are allocated on the node local to where the task is running,
except perhaps as modified by the taskâ€™s NUMA mempolicy or cpuset
configuration, so long as sufficient free memory pages are available.”…””}”(hjÖ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MThj  hžhubhÆ)”}”(hŒVWhen new cpusets are created, they inherit the memory spread settings
of their parent.”h]”hŒVWhen new cpusets are created, they inherit the memory spread settings
of their parent.”…””}”(hjä  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MYhj  hžhubhÆ)”}”(hX»  Setting memory spreading causes allocations for the affected page
or slab caches to ignore the task's NUMA mempolicy and be spread
instead.    Tasks using mbind() or set_mempolicy() calls to set NUMA
mempolicies will not notice any change in these calls as a result of
their containing task's memory spread settings.  If memory spreading
is turned off, then the currently specified NUMA mempolicy once again
applies to memory page allocations.”h]”hX¿  Setting memory spreading causes allocations for the affected page
or slab caches to ignore the taskâ€™s NUMA mempolicy and be spread
instead.    Tasks using mbind() or set_mempolicy() calls to set NUMA
mempolicies will not notice any change in these calls as a result of
their containing taskâ€™s memory spread settings.  If memory spreading
is turned off, then the currently specified NUMA mempolicy once again
applies to memory page allocations.”…””}”(hjò  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M\hj  hžhubhÆ)”}”(hŒðBoth 'cpuset.memory_spread_page' and 'cpuset.memory_spread_slab' are boolean flag
files.  By default they contain "0", meaning that the feature is off
for that cpuset.  If a "1" is written to that file, then that turns
the named feature on.”h]”hX   Both â€˜cpuset.memory_spread_pageâ€™ and â€˜cpuset.memory_spread_slabâ€™ are boolean flag
files.  By default they contain â€œ0â€, meaning that the feature is off
for that cpuset.  If a â€œ1â€ is written to that file, then that turns
the named feature on.”…””}”(hj 	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mdhj  hžhubhÆ)”}”(hŒThe implementation is simple.”h]”hŒThe implementation is simple.”…””}”(hj	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mihj  hžhubhÆ)”}”(hX…  Setting the flag 'cpuset.memory_spread_page' turns on a per-process flag
PFA_SPREAD_PAGE for each task that is in that cpuset or subsequently
joins that cpuset.  The page allocation calls for the page cache
is modified to perform an inline check for this PFA_SPREAD_PAGE task
flag, and if set, a call to a new routine cpuset_mem_spread_node()
returns the node to prefer for the allocation.”h]”hX‰  Setting the flag â€˜cpuset.memory_spread_pageâ€™ turns on a per-process flag
PFA_SPREAD_PAGE for each task that is in that cpuset or subsequently
joins that cpuset.  The page allocation calls for the page cache
is modified to perform an inline check for this PFA_SPREAD_PAGE task
flag, and if set, a call to a new routine cpuset_mem_spread_node()
returns the node to prefer for the allocation.”…””}”(hj	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mkhj  hžhubhÆ)”}”(hŒ¾Similarly, setting 'cpuset.memory_spread_slab' turns on the flag
PFA_SPREAD_SLAB, and appropriately marked slab caches will allocate
pages from the node returned by cpuset_mem_spread_node().”h]”hŒÂSimilarly, setting â€˜cpuset.memory_spread_slabâ€™ turns on the flag
PFA_SPREAD_SLAB, and appropriately marked slab caches will allocate
pages from the node returned by cpuset_mem_spread_node().”…””}”(hj*	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mrhj  hžhubhÆ)”}”(hŒÌThe cpuset_mem_spread_node() routine is also simple.  It uses the
value of a per-task rotor cpuset_mem_spread_rotor to select the next
node in the current task's mems_allowed to prefer for the allocation.”h]”hŒÎThe cpuset_mem_spread_node() routine is also simple.  It uses the
value of a per-task rotor cpuset_mem_spread_rotor to select the next
node in the current taskâ€™s mems_allowed to prefer for the allocation.”…””}”(hj8	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mvhj  hžhubhÆ)”}”(hŒ\This memory placement policy is also known (in other contexts) as
round-robin or interleave.”h]”hŒ\This memory placement policy is also known (in other contexts) as
round-robin or interleave.”…””}”(hjF	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mzhj  hžhubhÆ)”}”(hX±  This policy can provide substantial improvements for jobs that need
to place thread local data on the corresponding node, but that need
to access large file system data sets that need to be spread across
the several nodes in the jobs cpuset in order to fit.  Without this
policy, especially for jobs that might have one thread reading in the
data set, the memory allocation across the nodes in the jobs cpuset
can become very uneven.”h]”hX±  This policy can provide substantial improvements for jobs that need
to place thread local data on the corresponding node, but that need
to access large file system data sets that need to be spread across
the several nodes in the jobs cpuset in order to fit.  Without this
policy, especially for jobs that might have one thread reading in the
data set, the memory allocation across the nodes in the jobs cpuset
can become very uneven.”…””}”(hjT	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M}hj  hžhubeh}”(h]”Œwhat-is-memory-spread”ah ]”h"]”Œ1.6 what is memory spread ?”ah$]”h&]”uh1h°hjÚ  hžhhŸh¯h M@ubh±)”}”(hhh]”(h¶)”}”(hŒ 1.7 What is sched_load_balance ?”h]”hŒ 1.7 What is sched_load_balance ?”…””}”(hjm	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhjj	  hžhhŸh¯h M†ubhÆ)”}”(hX3  The kernel scheduler (kernel/sched/core.c) automatically load balances
tasks.  If one CPU is underutilized, kernel code running on that
CPU will look for tasks on other more overloaded CPUs and move those
tasks to itself, within the constraints of such placement mechanisms
as cpusets and sched_setaffinity.”h]”hX3  The kernel scheduler (kernel/sched/core.c) automatically load balances
tasks.  If one CPU is underutilized, kernel code running on that
CPU will look for tasks on other more overloaded CPUs and move those
tasks to itself, within the constraints of such placement mechanisms
as cpusets and sched_setaffinity.”…””}”(hj{	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mˆhjj	  hžhubhÆ)”}”(hXõ  The algorithmic cost of load balancing and its impact on key shared
kernel data structures such as the task list increases more than
linearly with the number of CPUs being balanced.  So the scheduler
has support to partition the systems CPUs into a number of sched
domains such that it only load balances within each sched domain.
Each sched domain covers some subset of the CPUs in the system;
no two sched domains overlap; some CPUs might not be in any sched
domain and hence won't be load balanced.”h]”hX÷  The algorithmic cost of load balancing and its impact on key shared
kernel data structures such as the task list increases more than
linearly with the number of CPUs being balanced.  So the scheduler
has support to partition the systems CPUs into a number of sched
domains such that it only load balances within each sched domain.
Each sched domain covers some subset of the CPUs in the system;
no two sched domains overlap; some CPUs might not be in any sched
domain and hence wonâ€™t be load balanced.”…””}”(hj‰	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MŽhjj	  hžhubhÆ)”}”(hŒ½Put simply, it costs less to balance between two smaller sched domains
than one big one, but doing so means that overloads in one of the
two domains won't be load balanced to the other one.”h]”hŒ¿Put simply, it costs less to balance between two smaller sched domains
than one big one, but doing so means that overloads in one of the
two domains wonâ€™t be load balanced to the other one.”…””}”(hj—	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M—hjj	  hžhubhÆ)”}”(hX  By default, there is one sched domain covering all CPUs, including those
marked isolated using the kernel boot time "isolcpus=" argument. However,
the isolated CPUs will not participate in load balancing, and will not
have tasks running on them unless explicitly assigned.”h]”hX  By default, there is one sched domain covering all CPUs, including those
marked isolated using the kernel boot time â€œisolcpus=â€ argument. However,
the isolated CPUs will not participate in load balancing, and will not
have tasks running on them unless explicitly assigned.”…””}”(hj¥	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M›hjj	  hžhubhÆ)”}”(hŒ`This default load balancing across all CPUs is not well suited for
the following two situations:”h]”hŒ`This default load balancing across all CPUs is not well suited for
the following two situations:”…””}”(hj³	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M hjj	  hžhubj”  )”}”(hXh  1) On large systems, load balancing across many CPUs is expensive.
   If the system is managed using cpusets to place independent jobs
   on separate sets of CPUs, full load balancing is unnecessary.
2) Systems supporting realtime on some CPUs need to minimize
   system overhead on those CPUs, including avoiding task load
   balancing if that is not needed.
”h]”hŒenumerated_list”“”)”}”(hhh]”(hû)”}”(hŒ¾On large systems, load balancing across many CPUs is expensive.
If the system is managed using cpusets to place independent jobs
on separate sets of CPUs, full load balancing is unnecessary.”h]”hÆ)”}”(hŒ¾On large systems, load balancing across many CPUs is expensive.
If the system is managed using cpusets to place independent jobs
on separate sets of CPUs, full load balancing is unnecessary.”h]”hŒ¾On large systems, load balancing across many CPUs is expensive.
If the system is managed using cpusets to place independent jobs
on separate sets of CPUs, full load balancing is unnecessary.”…””}”(hjÎ	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M£hjÊ	  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjÇ	  ubhû)”}”(hŒ—Systems supporting realtime on some CPUs need to minimize
system overhead on those CPUs, including avoiding task load
balancing if that is not needed.
”h]”hÆ)”}”(hŒ–Systems supporting realtime on some CPUs need to minimize
system overhead on those CPUs, including avoiding task load
balancing if that is not needed.”h]”hŒ–Systems supporting realtime on some CPUs need to minimize
system overhead on those CPUs, including avoiding task load
balancing if that is not needed.”…””}”(hjæ	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M¦hjâ	  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjÇ	  ubeh}”(h]”h ]”h"]”h$]”h&]”Œenumtype”Œarabic”Œprefix”hŒsuffix”Œ)”uh1jÅ	  hjÁ	  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j“  hŸh¯h M£hjj	  hžhubhÆ)”}”(hXI  When the per-cpuset flag "cpuset.sched_load_balance" is enabled (the default
setting), it requests that all the CPUs in that cpusets allowed 'cpuset.cpus'
be contained in a single sched domain, ensuring that load balancing
can move a task (not otherwised pinned, as by sched_setaffinity)
from any CPU in that cpuset to any other.”h]”hXQ  When the per-cpuset flag â€œcpuset.sched_load_balanceâ€ is enabled (the default
setting), it requests that all the CPUs in that cpusets allowed â€˜cpuset.cpusâ€™
be contained in a single sched domain, ensuring that load balancing
can move a task (not otherwised pinned, as by sched_setaffinity)
from any CPU in that cpuset to any other.”…””}”(hj
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mªhjj	  hžhubhÆ)”}”(hŒõWhen the per-cpuset flag "cpuset.sched_load_balance" is disabled, then the
scheduler will avoid load balancing across the CPUs in that cpuset,
--except-- in so far as is necessary because some overlapping cpuset
has "sched_load_balance" enabled.”h]”hŒýWhen the per-cpuset flag â€œcpuset.sched_load_balanceâ€ is disabled, then the
scheduler will avoid load balancing across the CPUs in that cpuset,
--except-- in so far as is necessary because some overlapping cpuset
has â€œsched_load_balanceâ€ enabled.”…””}”(hj
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M°hjj	  hžhubhÆ)”}”(hX  So, for example, if the top cpuset has the flag "cpuset.sched_load_balance"
enabled, then the scheduler will have one sched domain covering all
CPUs, and the setting of the "cpuset.sched_load_balance" flag in any other
cpusets won't matter, as we're already fully load balancing.”h]”hX#  So, for example, if the top cpuset has the flag â€œcpuset.sched_load_balanceâ€
enabled, then the scheduler will have one sched domain covering all
CPUs, and the setting of the â€œcpuset.sched_load_balanceâ€ flag in any other
cpusets wonâ€™t matter, as weâ€™re already fully load balancing.”…””}”(hj'
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mµhjj	  hžhubhÆ)”}”(hŒ®Therefore in the above two situations, the top cpuset flag
"cpuset.sched_load_balance" should be disabled, and only some of the smaller,
child cpusets have this flag enabled.”h]”hŒ²Therefore in the above two situations, the top cpuset flag
â€œcpuset.sched_load_balanceâ€ should be disabled, and only some of the smaller,
child cpusets have this flag enabled.”…””}”(hj5
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mºhjj	  hžhubhÆ)”}”(hX¾  When doing this, you don't usually want to leave any unpinned tasks in
the top cpuset that might use non-trivial amounts of CPU, as such tasks
may be artificially constrained to some subset of CPUs, depending on
the particulars of this flag setting in descendant cpusets.  Even if
such a task could use spare CPU cycles in some other CPUs, the kernel
scheduler might not consider the possibility of load balancing that
task to that underused CPU.”h]”hXÀ  When doing this, you donâ€™t usually want to leave any unpinned tasks in
the top cpuset that might use non-trivial amounts of CPU, as such tasks
may be artificially constrained to some subset of CPUs, depending on
the particulars of this flag setting in descendant cpusets.  Even if
such a task could use spare CPU cycles in some other CPUs, the kernel
scheduler might not consider the possibility of load balancing that
task to that underused CPU.”…””•¡ç      }”(hjC
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M¾hjj	  hžhubhÆ)”}”(hŒŸOf course, tasks pinned to a particular CPU can be left in a cpuset
that disables "cpuset.sched_load_balance" as those tasks aren't going anywhere
else anyway.”h]”hŒ¥Of course, tasks pinned to a particular CPU can be left in a cpuset
that disables â€œcpuset.sched_load_balanceâ€ as those tasks arenâ€™t going anywhere
else anyway.”…””}”(hjQ
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MÆhjj	  hžhubhÆ)”}”(hŒÃThere is an impedance mismatch here, between cpusets and sched domains.
Cpusets are hierarchical and nest.  Sched domains are flat; they don't
overlap and each CPU is in at most one sched domain.”h]”hŒÅThere is an impedance mismatch here, between cpusets and sched domains.
Cpusets are hierarchical and nest.  Sched domains are flat; they donâ€™t
overlap and each CPU is in at most one sched domain.”…””}”(hj_
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MÊhjj	  hžhubhÆ)”}”(hXç  It is necessary for sched domains to be flat because load balancing
across partially overlapping sets of CPUs would risk unstable dynamics
that would be beyond our understanding.  So if each of two partially
overlapping cpusets enables the flag 'cpuset.sched_load_balance', then we
form a single sched domain that is a superset of both.  We won't move
a task to a CPU outside its cpuset, but the scheduler load balancing
code might waste some compute cycles considering that possibility.”h]”hXí  It is necessary for sched domains to be flat because load balancing
across partially overlapping sets of CPUs would risk unstable dynamics
that would be beyond our understanding.  So if each of two partially
overlapping cpusets enables the flag â€˜cpuset.sched_load_balanceâ€™, then we
form a single sched domain that is a superset of both.  We wonâ€™t move
a task to a CPU outside its cpuset, but the scheduler load balancing
code might waste some compute cycles considering that possibility.”…””}”(hjm
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MÎhjj	  hžhubhÆ)”}”(hXp  This mismatch is why there is not a simple one-to-one relation
between which cpusets have the flag "cpuset.sched_load_balance" enabled,
and the sched domain configuration.  If a cpuset enables the flag, it
will get balancing across all its CPUs, but if it disables the flag,
it will only be assured of no load balancing if no other overlapping
cpuset enables the flag.”h]”hXt  This mismatch is why there is not a simple one-to-one relation
between which cpusets have the flag â€œcpuset.sched_load_balanceâ€ enabled,
and the sched domain configuration.  If a cpuset enables the flag, it
will get balancing across all its CPUs, but if it disables the flag,
it will only be assured of no load balancing if no other overlapping
cpuset enables the flag.”…””}”(hj{
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MÖhjj	  hžhubhÆ)”}”(hX8  If two cpusets have partially overlapping 'cpuset.cpus' allowed, and only
one of them has this flag enabled, then the other may find its
tasks only partially load balanced, just on the overlapping CPUs.
This is just the general case of the top_cpuset example given a few
paragraphs above.  In the general case, as in the top cpuset case,
don't leave tasks that might use non-trivial amounts of CPU in
such partially load balanced cpusets, as they may be artificially
constrained to some subset of the CPUs allowed to them, for lack of
load balancing to the other CPUs.”h]”hX>  If two cpusets have partially overlapping â€˜cpuset.cpusâ€™ allowed, and only
one of them has this flag enabled, then the other may find its
tasks only partially load balanced, just on the overlapping CPUs.
This is just the general case of the top_cpuset example given a few
paragraphs above.  In the general case, as in the top cpuset case,
donâ€™t leave tasks that might use non-trivial amounts of CPU in
such partially load balanced cpusets, as they may be artificially
constrained to some subset of the CPUs allowed to them, for lack of
load balancing to the other CPUs.”…””}”(hj‰
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MÝhjj	  hžhubhÆ)”}”(hŒÆCPUs in "cpuset.isolcpus" were excluded from load balancing by the
isolcpus= kernel boot option, and will never be load balanced regardless
of the value of "cpuset.sched_load_balance" in any cpuset.”h]”hŒÎCPUs in â€œcpuset.isolcpusâ€ were excluded from load balancing by the
isolcpus= kernel boot option, and will never be load balanced regardless
of the value of â€œcpuset.sched_load_balanceâ€ in any cpuset.”…””}”(hj—
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mçhjj	  hžhubeh}”(h]”Œwhat-is-sched-load-balance”ah ]”h"]”Œ 1.7 what is sched_load_balance ?”ah$]”h&]”uh1h°hjÚ  hžhhŸh¯h M†ubh±)”}”(hhh]”(h¶)”}”(hŒ01.7.1 sched_load_balance implementation details.”h]”hŒ01.7.1 sched_load_balance implementation details.”…””}”(hj°
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhj­
  hžhhŸh¯h MìubhÆ)”}”(hX3  The per-cpuset flag 'cpuset.sched_load_balance' defaults to enabled (contrary
to most cpuset flags.)  When enabled for a cpuset, the kernel will
ensure that it can load balance across all the CPUs in that cpuset
(makes sure that all the CPUs in the cpus_allowed of that cpuset are
in the same sched domain.)”h]”hX7  The per-cpuset flag â€˜cpuset.sched_load_balanceâ€™ defaults to enabled (contrary
to most cpuset flags.)  When enabled for a cpuset, the kernel will
ensure that it can load balance across all the CPUs in that cpuset
(makes sure that all the CPUs in the cpus_allowed of that cpuset are
in the same sched domain.)”…””}”(hj¾
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mîhj­
  hžhubhÆ)”}”(hŒ„If two overlapping cpusets both have 'cpuset.sched_load_balance' enabled,
then they will be (must be) both in the same sched domain.”h]”hŒˆIf two overlapping cpusets both have â€˜cpuset.sched_load_balanceâ€™ enabled,
then they will be (must be) both in the same sched domain.”…””}”(hjÌ
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Môhj­
  hžhubhÆ)”}”(hŒÎIf, as is the default, the top cpuset has 'cpuset.sched_load_balance' enabled,
then by the above that means there is a single sched domain covering
the whole system, regardless of any other cpuset settings.”h]”hŒÒIf, as is the default, the top cpuset has â€˜cpuset.sched_load_balanceâ€™ enabled,
then by the above that means there is a single sched domain covering
the whole system, regardless of any other cpuset settings.”…””}”(hjÚ
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M÷hj­
  hžhubhÆ)”}”(hX  The kernel commits to user space that it will avoid load balancing
where it can.  It will pick as fine a granularity partition of sched
domains as it can while still providing load balancing for any set
of CPUs allowed to a cpuset having 'cpuset.sched_load_balance' enabled.”h]”hX  The kernel commits to user space that it will avoid load balancing
where it can.  It will pick as fine a granularity partition of sched
domains as it can while still providing load balancing for any set
of CPUs allowed to a cpuset having â€˜cpuset.sched_load_balanceâ€™ enabled.”…””}”(hjè
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mûhj­
  hžhubhÆ)”}”(hX7  The internal kernel cpuset to scheduler interface passes from the
cpuset code to the scheduler code a partition of the load balanced
CPUs in the system. This partition is a set of subsets (represented
as an array of struct cpumask) of CPUs, pairwise disjoint, that cover
all the CPUs that must be load balanced.”h]”hX7  The internal kernel cpuset to scheduler interface passes from the
cpuset code to the scheduler code a partition of the load balanced
CPUs in the system. This partition is a set of subsets (represented
as an array of struct cpumask) of CPUs, pairwise disjoint, that cover
all the CPUs that must be load balanced.”…””}”(hjö
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M hj­
  hžhubhÆ)”}”(hŒThe cpuset code builds a new such partition and passes it to the
scheduler sched domain setup code, to have the sched domains rebuilt
as necessary, whenever:”h]”hŒThe cpuset code builds a new such partition and passes it to the
scheduler sched domain setup code, to have the sched domains rebuilt
as necessary, whenever:”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj­
  hžhubj”  )”}”(hXg  - the 'cpuset.sched_load_balance' flag of a cpuset with non-empty CPUs changes,
- or CPUs come or go from a cpuset with this flag enabled,
- or 'cpuset.sched_relax_domain_level' value of a cpuset with non-empty CPUs
  and with this flag enabled changes,
- or a cpuset with non-empty CPUs and with this flag enabled is removed,
- or a cpu is offlined/onlined.
”h]”hö)”}”(hhh]”(hû)”}”(hŒMthe 'cpuset.sched_load_balance' flag of a cpuset with non-empty CPUs changes,”h]”hÆ)”}”(hj  h]”hŒQthe â€˜cpuset.sched_load_balanceâ€™ flag of a cpuset with non-empty CPUs changes,”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M
hj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj  ubhû)”}”(hŒ8or CPUs come or go from a cpuset with this flag enabled,”h]”hÆ)”}”(hj2  h]”hŒ8or CPUs come or go from a cpuset with this flag enabled,”…””}”(hj4  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj0  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj  ubhû)”}”(hŒnor 'cpuset.sched_relax_domain_level' value of a cpuset with non-empty CPUs
and with this flag enabled changes,”h]”hÆ)”}”(hŒnor 'cpuset.sched_relax_domain_level' value of a cpuset with non-empty CPUs
and with this flag enabled changes,”h]”hŒror â€˜cpuset.sched_relax_domain_levelâ€™ value of a cpuset with non-empty CPUs
and with this flag enabled changes,”…””}”(hjK  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MhjG  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj  ubhû)”}”(hŒFor a cpuset with non-empty CPUs and with this flag enabled is removed,”h]”hÆ)”}”(hja  h]”hŒFor a cpuset with non-empty CPUs and with this flag enabled is removed,”…””}”(hjc  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj_  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj  ubhû)”}”(hŒor a cpu is offlined/onlined.
”h]”hÆ)”}”(hŒor a cpu is offlined/onlined.”h]”hŒor a cpu is offlined/onlined.”…””}”(hjz  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhjv  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj  ubeh}”(h]”h ]”h"]”h$]”h&]”jÆ  jÇ  uh1hõhŸh¯h M
hj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j“  hŸh¯h M
hj­
  hžhubhÆ)”}”(hŒ“This partition exactly defines what sched domains the scheduler should
setup - one sched domain for each element (struct cpumask) in the
partition.”h]”hŒ“This partition exactly defines what sched domains the scheduler should
setup - one sched domain for each element (struct cpumask) in the
partition.”…””}”(hjš  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj­
  hžhubhÆ)”}”(hXH  The scheduler remembers the currently active sched domain partitions.
When the scheduler routine partition_sched_domains() is invoked from
the cpuset code to update these sched domains, it compares the new
partition requested with the current, and updates its sched domains,
removing the old and adding the new, for each change.”h]”hXH  The scheduler remembers the currently active sched domain partitions.
When the scheduler routine partition_sched_domains() is invoked from
the cpuset code to update these sched domains, it compares the new
partition requested with the current, and updates its sched domains,
removing the old and adding the new, for each change.”…””}”(hj¨  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj­
  hžhubeh}”(h]”Œ)sched-load-balance-implementation-details”ah ]”h"]”Œ01.7.1 sched_load_balance implementation details.”ah$]”h&]”uh1h°hjÚ  hžhhŸh¯h Mìubh±)”}”(hhh]”(h¶)”}”(hŒ&1.8 What is sched_relax_domain_level ?”h]”hŒ&1.8 What is sched_relax_domain_level ?”…””}”(hjÁ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhj¾  hžhhŸh¯h MubhÆ)”}”(hŒ|In sched domain, the scheduler migrates tasks in 2 ways; periodic load
balance on tick, and at time of some schedule events.”h]”hŒ|In sched domain, the scheduler migrates tasks in 2 ways; periodic load
balance on tick, and at time of some schedule events.”…””}”(hjÏ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj¾  hžhubhÆ)”}”(hX5  When a task is woken up, scheduler try to move the task on idle CPU.
For example, if a task A running on CPU X activates another task B
on the same CPU X, and if CPU Y is X's sibling and performing idle,
then scheduler migrate task B to CPU Y so that task B can start on
CPU Y without waiting task A on CPU X.”h]”hX7  When a task is woken up, scheduler try to move the task on idle CPU.
For example, if a task A running on CPU X activates another task B
on the same CPU X, and if CPU Y is Xâ€™s sibling and performing idle,
then scheduler migrate task B to CPU Y so that task B can start on
CPU Y without waiting task A on CPU X.”…””}”(hjÝ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M"hj¾  hžhubhÆ)”}”(hŒAnd if a CPU run out of tasks in its runqueue, the CPU try to pull
extra tasks from other busy CPUs to help them before it is going to
be idle.”h]”hŒAnd if a CPU run out of tasks in its runqueue, the CPU try to pull
extra tasks from other busy CPUs to help them before it is going to
be idle.”…””}”(hjë  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M(hj¾  hžhubhÆ)”}”(hX;  Of course it takes some searching cost to find movable tasks and/or
idle CPUs, the scheduler might not search all CPUs in the domain
every time.  In fact, in some architectures, the searching ranges on
events are limited in the same socket or node where the CPU locates,
while the load balance on tick searches all.”h]”hX;  Of course it takes some searching cost to find movable tasks and/or
idle CPUs, the scheduler might not search all CPUs in the domain
every time.  In fact, in some architectures, the searching ranges on
events are limited in the same socket or node where the CPU locates,
while the load balance on tick searches all.”…””}”(hjù  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M,hj¾  hžhubhÆ)”}”(hXu  For example, assume CPU Z is relatively far from CPU X.  Even if CPU Z
is idle while CPU X and the siblings are busy, scheduler can't migrate
woken task B from X to Z since it is out of its searching range.
As the result, task B on CPU X need to wait task A or wait load balance
on the next tick.  For some applications in special situation, waiting
1 tick may be too long.”h]”hXw  For example, assume CPU Z is relatively far from CPU X.  Even if CPU Z
is idle while CPU X and the siblings are busy, scheduler canâ€™t migrate
woken task B from X to Z since it is out of its searching range.
As the result, task B on CPU X need to wait task A or wait load balance
on the next tick.  For some applications in special situation, waiting
1 tick may be too long.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M2hj¾  hžhubhÆ)”}”(hX  The 'cpuset.sched_relax_domain_level' file allows you to request changing
this searching range as you like.  This file takes int value which
indicates size of searching range in levels approximately as follows,
otherwise initial value -1 that indicates the cpuset has no request.”h]”hX  The â€˜cpuset.sched_relax_domain_levelâ€™ file allows you to request changing
this searching range as you like.  This file takes int value which
indicates size of searching range in levels approximately as follows,
otherwise initial value -1 that indicates the cpuset has no request.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M9hj¾  hžhubhŒtable”“”)”}”(hhh]”hŒtgroup”“”)”}”(hhh]”(hŒcolspec”“”)”}”(hhh]”h}”(h]”h ]”h"]”h$]”h&]”Œcolwidth”Kuh1j-  hj*  ubj.  )”}”(hhh]”h}”(h]”h ]”h"]”h$]”h&]”Œcolwidth”K;uh1j-  hj*  ubhŒtbody”“”)”}”(hhh]”(hŒrow”“”)”}”(hhh]”(hŒentry”“”)”}”(hhh]”hÆ)”}”(hŒ-1”h]”hŒ-1”…””}”(hjR  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M?hjO  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hjJ  ubjN  )”}”(hhh]”hÆ)”}”(hŒ;no request. use system default or follow request of others.”h]”hŒ;no request. use system default or follow request of others.”…””}”(hji  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M?hjf  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hjJ  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jH  hjE  ubjI  )”}”(hhh]”(jN  )”}”(hhh]”hÆ)”}”(hŒ0”h]”hŒ0”…””}”(hj‰  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M@hj†  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hjƒ  ubjN  )”}”(hhh]”hÆ)”}”(hŒ
no search.”h]”hŒ
no search.”…””}”(hj   hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M@hj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hjƒ  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jH  hjE  ubjI  )”}”(hhh]”(jN  )”}”(hhh]”hÆ)”}”(hŒ1”h]”hŒ1”…””}”(hjÀ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MAhj½  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hjº  ubjN  )”}”(hhh]”hÆ)”}”(hŒ)search siblings (hyperthreads in a core).”h]”hŒ)search siblings (hyperthreads in a core).”…””}”(hj×  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MAhjÔ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hjº  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jH  hjE  ubjI  )”}”(hhh]”(jN  )”}”(hhh]”hÆ)”}”(hŒ2”h]”hŒ2”…””}”(hj÷  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MBhjô  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hjñ  ubjN  )”}”(hhh]”hÆ)”}”(hŒsearch cores in a package.”h]”hŒsearch cores in a package.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MBhj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hjñ  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jH  hjE  ubjI  )”}”(hhh]”(jN  )”}”(hhh]”hÆ)”}”(hŒ3”h]”hŒ3”…””}”(hj.  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MChj+  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hj(  ubjN  )”}”(hhh]”hÆ)”}”(hŒ8search cpus in a node [= system wide on non-NUMA system]”h]”hŒ8search cpus in a node [= system wide on non-NUMA system]”…””}”(hjE  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MChjB  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hj(  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jH  hjE  ubjI  )”}”(hhh]”(jN  )”}”(hhh]”hÆ)”}”(hŒ4”h]”hŒ4”…””}”(hje  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MDhjb  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hj_  ubjN  )”}”(hhh]”hÆ)”}”(hŒ0search nodes in a chunk of node [on NUMA system]”h]”hŒ0search nodes in a chunk of node [on NUMA system]”…””}”(hj|  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MDhjy  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hj_  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jH  hjE  ubjI  )”}”(hhh]”(jN  )”}”(hhh]”hÆ)”}”(hŒ5”h]”hŒ5”…””}”(hjœ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MEhj™  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hj–  ubjN  )”}”(hhh]”hÆ)”}”(hŒ#search system wide [on NUMA system]”h]”hŒ#search system wide [on NUMA system]”…””}”(hj³  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MEhj°  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jM  hj–  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jH  hjE  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jC  hj*  ubeh}”(h]”h ]”h"]”h$]”h&]”Œcols”Kuh1j(  hj%  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j#  hj¾  hžhhŸh¯h NubhÆ)”}”(hŒÃNot all levels can be present and values can change depending on the
system architecture and kernel configuration. Check
/sys/kernel/debug/sched/domains/cpu*/domain*/ for system-specific
details.”h]”hŒÃNot all levels can be present and values can change depending on the
system architecture and kernel configuration. Check
/sys/kernel/debug/sched/domains/cpu*/domain*/ for system-specific
details.”…””}”(hjà  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MHhj¾  hžhubhÆ)”}”(hŒ~The system default is architecture dependent.  The system default
can be changed using the relax_domain_level= boot parameter.”h]”hŒ~The system default is architecture dependent.  The system default
can be changed using the relax_domain_level= boot parameter.”…””}”(hjî  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MMhj¾  hžhubhÆ)”}”(hX  This file is per-cpuset and affect the sched domain where the cpuset
belongs to.  Therefore if the flag 'cpuset.sched_load_balance' of a cpuset
is disabled, then 'cpuset.sched_relax_domain_level' have no effect since
there is no sched domain belonging the cpuset.”h]”hX  This file is per-cpuset and affect the sched domain where the cpuset
belongs to.  Therefore if the flag â€˜cpuset.sched_load_balanceâ€™ of a cpuset
is disabled, then â€˜cpuset.sched_relax_domain_levelâ€™ have no effect since
there is no sched domain belonging the cpuset.”…””}”(hjü  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MPhj¾  hžhubhÆ)”}”(hŒ¶If multiple cpusets are overlapping and hence they form a single sched
domain, the largest value among those is used.  Be careful, if one
requests 0 and others are -1 then 0 is used.”h]”hŒ¶If multiple cpusets are overlapping and hence they form a single sched
domain, the largest value among those is used.  Be careful, if one
requests 0 and others are -1 then 0 is used.”…””}”(hj
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MUhj¾  hžhubhÆ)”}”(hŒ­Note that modifying this file will have both good and bad effects,
and whether it is acceptable or not depends on your situation.
Don't modify this file if you are not sure.”h]”hŒ¯Note that modifying this file will have both good and bad effects,
and whether it is acceptable or not depends on your situation.
Donâ€™t modify this file if you are not sure.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MYhj¾  hžhubhÆ)”}”(hŒIf your situation is:”h]”hŒIf your situation is:”…””}”(hj&  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M]hj¾  hžhubj”  )”}”(hX¹  - The migration costs between each cpu can be assumed considerably
  small(for you) due to your special application's behavior or
  special hardware support for CPU cache etc.
- The searching cost doesn't have impact(for you) or you can make
  the searching cost enough small by managing cpuset to compact etc.
- The latency is required even it sacrifices cache hit rate etc.
  then increasing 'sched_relax_domain_level' would benefit you.

”h]”hö)”}”(hhh]”(hû)”}”(hŒ©The migration costs between each cpu can be assumed considerably
small(for you) due to your special application's behavior or
special hardware support for CPU cache etc.”h]”hÆ)”}”(hŒ©The migration costs between each cpu can be assumed considerably
small(for you) due to your special application's behavior or
special hardware support for CPU cache etc.”h]”hŒ«The migration costs between each cpu can be assumed considerably
small(for you) due to your special applicationâ€™s behavior or
special hardware support for CPU cache etc.”…””}”(hj?  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M_hj;  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj8  ubhû)”}”(hŒ‚The searching cost doesn't have impact(for you) or you can make
the searching cost enough small by managing cpuset to compact etc.”h]”hÆ)”}”(hŒ‚The searching cost doesn't have impact(for you) or you can make
the searching cost enough small by managing cpuset to compact etc.”h]”hŒ„The searching cost doesnâ€™t have impact(for you) or you can make
the searching cost enough small by managing cpuset to compact etc.”…””}”(hjW  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MbhjS  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj8  ubhû)”}”(hŒ~The latency is required even it sacrifices cache hit rate etc.
then increasing 'sched_relax_domain_level' would benefit you.

”h]”hÆ)”}”(hŒ|The latency is required even it sacrifices cache hit rate etc.
then increasing 'sched_relax_domain_level' would benefit you.”h]”hŒ€The latency is required even it sacrifices cache hit rate etc.
then increasing â€˜sched_relax_domain_levelâ€™ would benefit you.”…””}”(hjo  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mdhjk  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj8  ubeh}”(h]”h ]”h"]”h$]”h&]”jÆ  jÇ  uh1hõhŸh¯h M_hj4  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j“  hŸh¯h M_hj¾  hžhubeh}”(h]”Œ what-is-sched-relax-domain-level”ah ]”h"]”Œ&1.8 what is sched_relax_domain_level ?”ah$]”h&]”uh1h°hjÚ  hžhhŸh¯h Mubh±)”}”(hhh]”(h¶)”}”(hŒ1.9 How do I use cpusets ?”h]”hŒ1.9 How do I use cpusets ?”…””}”(hjš  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhj—  hžhhŸh¯h MiubhÆ)”}”(hXX  In order to minimize the impact of cpusets on critical kernel
code, such as the scheduler, and due to the fact that the kernel
does not support one task updating the memory placement of another
task directly, the impact on a task of changing its cpuset CPU
or Memory Node placement, or of changing to which cpuset a task
is attached, is subtle.”h]”hXX  In order to minimize the impact of cpusets on critical kernel
code, such as the scheduler, and due to the fact that the kernel
does not support one task updating the memory placement of another
task directly, the impact on a task of changing its cpuset CPU
or Memory Node placement, or of changing to which cpuset a task
is attached, is subtle.”…””}”(hj¨  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mkhj—  hžhubhÆ)”}”(hXé  If a cpuset has its Memory Nodes modified, then for each task attached
to that cpuset, the next time that the kernel attempts to allocate
a page of memory for that task, the kernel will notice the change
in the task's cpuset, and update its per-task memory placement to
remain within the new cpusets memory placement.  If the task was using
mempolicy MPOL_BIND, and the nodes to which it was bound overlap with
its new cpuset, then the task will continue to use whatever subset
of MPOL_BIND nodes are still allowed in the new cpuset.  If the task
was using MPOL_BIND and now none of its MPOL_BIND nodes are allowed
in the new cpuset, then the task will be essentially treated as if it
was MPOL_BIND bound to the new cpuset (even though its NUMA placement,
as queried by get_mempolicy(), doesn't change).  If a task is moved
from one cpuset to another, then the kernel will adjust the task's
memory placement, as above, the next time that the kernel attempts
to allocate a page of memory for that task.”h]”hXï  If a cpuset has its Memory Nodes modified, then for each task attached
to that cpuset, the next time that the kernel attempts to allocate
a page of memory for that task, the kernel will notice the change
in the taskâ€™s cpuset, and update its per-task memory placement to
remain within the new cpusets memory placement.  If the task was using
mempolicy MPOL_BIND, and the nodes to which it was bound overlap with
its new cpuset, then the task will continue to use whatever subset
of MPOL_BIND nodes are still allowed in the new cpuset.  If the task
was using MPOL_BIND and now none of its MPOL_BIND nodes are allowed
in the new cpuset, then the task will be essentially treated as if it
was MPOL_BIND bound to the new cpuset (even though its NUMA placement,
as queried by get_mempolicy(), doesnâ€™t change).  If a task is moved
from one cpuset to another, then the kernel will adjust the taskâ€™s
memory placement, as above, the next time that the kernel attempts
to allocate a page of memory for that task.”…””}”(hj¶  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mrhj—  hžhubhÆ)”}”(hXã  If a cpuset has its 'cpuset.cpus' modified, then each task in that cpuset
will have its allowed CPU placement changed immediately.  Similarly,
if a task's pid is written to another cpuset's 'tasks' file, then its
allowed CPU placement is changed immediately.  If such a task had been
bound to some subset of its cpuset using the sched_setaffinity() call,
the task will be allowed to run on any CPU allowed in its new cpuset,
negating the effect of the prior sched_setaffinity() call.”h]”hXï  If a cpuset has its â€˜cpuset.cpusâ€™ modified, then each task in that cpuset
will have its allowed CPU placement changed immediately.  Similarly,
if a taskâ€™s pid is written to another cpusetâ€™s â€˜tasksâ€™ file, then its
allowed CPU placement is changed immediately.  If such a task had been
bound to some subset of its cpuset using the sched_setaffinity() call,
the task will be allowed to run on any CPU allowed in its new cpuset,
negating the effect of the prior sched_setaffinity() call.”…””}”(hjÄ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M‚hj—  hžhubhÆ)”}”(hŒÀIn summary, the memory placement of a task whose cpuset is changed is
updated by the kernel, on the next allocation of a page for that task,
and the processor placement is updated immediately.”h]”hŒÀIn summary, the memory placement of a task whose cpuset is changed is
updated by the kernel, on the next allocation of a page for that task,
and the processor placement is updated immediately.”…””}”(hjÒ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MŠhj—  hžhubhÆ)”}”(hXÎ  Normally, once a page is allocated (given a physical page
of main memory) then that page stays on whatever node it
was allocated, so long as it remains allocated, even if the
cpusets memory placement policy 'cpuset.mems' subsequently changes.
If the cpuset flag file 'cpuset.memory_migrate' is set true, then when
tasks are attached to that cpuset, any pages that task had
allocated to it on nodes in its previous cpuset are migrated
to the task's new cpuset. The relative placement of the page within
the cpuset is preserved during these migration operations if possible.
For example if the page was on the second valid node of the prior cpuset
then the page will be placed on the second valid node of the new cpuset.”h]”hXØ  Normally, once a page is allocated (given a physical page
of main memory) then that page stays on whatever node it
was allocated, so long as it remains allocated, even if the
cpusets memory placement policy â€˜cpuset.memsâ€™ subsequently changes.
If the cpuset flag file â€˜cpuset.memory_migrateâ€™ is set true, then when
tasks are attached to that cpuset, any pages that task had
allocated to it on nodes in its previous cpuset are migrated
to the taskâ€™s new cpuset. The relative placement of the page within
the cpuset is preserved during these migration operations if possible.
For example if the page was on the second valid node of the prior cpuset
then the page will be placed on the second valid node of the new cpuset.”…””}”(hjà  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MŽhj—  hžhubhÆ)”}”(hXp  Also if 'cpuset.memory_migrate' is set true, then if that cpuset's
'cpuset.mems' file is modified, pages allocated to tasks in that
cpuset, that were on nodes in the previous setting of 'cpuset.mems',
will be moved to nodes in the new setting of 'mems.'
Pages that were not in the task's prior cpuset, or in the cpuset's
prior 'cpuset.mems' setting, will not be moved.”h]”hXŠ  Also if â€˜cpuset.memory_migrateâ€™ is set true, then if that cpusetâ€™s
â€˜cpuset.memsâ€™ file is modified, pages allocated to tasks in that
cpuset, that were on nodes in the previous setting of â€˜cpuset.memsâ€™,
will be moved to nodes in the new setting of â€˜mems.â€™
Pages that were not in the taskâ€™s prior cpuset, or in the cpusetâ€™s
prior â€˜cpuset.memsâ€™ setting, will not be moved.”…””}”(hjî  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mšhj—  hžhubhÆ)”}”(hX%  There is an exception to the above.  If hotplug functionality is used
to remove all the CPUs that are currently assigned to a cpuset,
then all the tasks in that cpuset will be moved to the nearest ancestor
with non-empty cpus.  But the moving of some (or all) tasks might fail if
cpuset is bound with another cgroup subsystem which has some restrictions
on task attaching.  In this failing case, those tasks will stay
in the original cpuset, and the kernel will automatically update
their cpus_allowed to allow all online CPUs.  When memory hotplug
functionality for removing Memory Nodes is available, a similar exception
is expected to apply there as well.  In general, the kernel prefers to
violate cpuset placement, over starving a task that has had all
its allowed CPUs or Memory Nodes taken offline.”h]”hX%  There is an exception to the above.  If hotplug functionality is used
to remove all the CPUs that are currently assigned to a cpuset,
then all the tasks in that cpuset will be moved to the nearest ancestor
with non-empty cpus.  But the moving of some (or all) tasks might fail if
cpuset is bound with another cgroup subsystem which has some restrictions
on task attaching.  In this failing case, those tasks will stay
in the original cpuset, and the kernel will automatically update
their cpus_allowed to allow all online CPUs.  When memory hotplug
functionality for removing Memory Nodes is available, a similar exception
is expected to apply there as well.  In general, the kernel prefers to
violate cpuset placement, over starving a task that has had all
its allowed CPUs or Memory Nodes taken offline.”…””}”(hjü  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M¡hj—  hžhubhÆ)”}”(hX¤  There is a second exception to the above.  GFP_ATOMIC requests are
kernel internal allocations that must be satisfied, immediately.
The kernel may drop some request, in rare cases even panic, if a
GFP_ATOMIC alloc fails.  If the request cannot be satisfied within
the current task's cpuset, then we relax the cpuset, and look for
memory anywhere we can find it.  It's better to violate the cpuset
than stress the kernel.”h]”hX¨  There is a second exception to the above.  GFP_ATOMIC requests are
kernel internal allocations that must be satisfied, immediately.
The kernel may drop some request, in rare cases even panic, if a
GFP_ATOMIC alloc fails.  If the request cannot be satisfied within
the current taskâ€™s cpuset, then we relax the cpuset, and look for
memory anywhere we can find it.  Itâ€™s better to violate the cpuset
than stress the kernel.”…””}”(hj
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M®hj—  hžhubhÆ)”}”(hŒJTo start a new job that is to be contained within a cpuset, the steps are:”h]”hŒJTo start a new job that is to be contained within a cpuset, the steps are:”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M¶hj—  hžhubj”  )”}”(hXÌ  1) mkdir /sys/fs/cgroup/cpuset
2) mount -t cgroup -ocpuset cpuset /sys/fs/cgroup/cpuset
3) Create the new cpuset by doing mkdir's and write's (or echo's) in
   the /sys/fs/cgroup/cpuset virtual file system.
4) Start a task that will be the "founding father" of the new job.
5) Attach that task to the new cpuset by writing its pid to the
   /sys/fs/cgroup/cpuset tasks file for that cpuset.
6) fork, exec or clone the job tasks from this founding father task.
”h]”jÆ	  )”}”(hhh]”(hû)”}”(hŒmkdir /sys/fs/cgroup/cpuset”h]”hÆ)”}”(hj/  h]”hŒmkdir /sys/fs/cgroup/cpuset”…””}”(hj1  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M¸hj-  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj*  ubhû)”}”(hŒ5mount -t cgroup -ocpuset cpuset /sys/fs/cgroup/cpuset”h]”hÆ)”}”(hjF  h]”hŒ5mount -t cgroup -ocpuset cpuset /sys/fs/cgroup/cpuset”…””}”(hjH  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M¹hjD  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj*  ubhû)”}”(hŒpCreate the new cpuset by doing mkdir's and write's (or echo's) in
the /sys/fs/cgroup/cpuset virtual file system.”h]”hÆ)”}”(hŒpCreate the new cpuset by doing mkdir's and write's (or echo's) in
the /sys/fs/cgroup/cpuset virtual file system.”h]”hŒvCreate the new cpuset by doing mkdirâ€™s and writeâ€™s (or echoâ€™s) in
the /sys/fs/cgroup/cpuset virtual file system.”…””}”(hj_  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mºhj[  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj*  ubhû)”}”(hŒ?Start a task that will be the "founding father" of the new job.”h]”hÆ)”}”(hju  h]”hŒCStart a task that will be the â€œfounding fatherâ€ of the new job.”…””}”(hjw  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M¼hjs  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj*  ubhû)”}”(hŒnAttach that task to the new cpuset by writing its pid to the
/sys/fs/cgroup/cpuset tasks file for that cpuset.”h]”hÆ)”}”(hŒnAttach that task to the new cpuset by writing its pid to the
/sys/fs/cgroup/cpuset tasks file for that cpuset.”h]”hŒnAttach that task to the new cpuset by writing its pid to the
/sys/fs/cgroup/cpuset tasks file for that cpuset.”…””}”(hjŽ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M½hjŠ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj*  ubhû)”}”(hŒBfork, exec or clone the job tasks from this founding father task.
”h]”hÆ)”}”(hŒAfork, exec or clone the job tasks from this founding father task.”h]”hŒAfork, exec or clone the job tasks from this founding father task.”…””}”(hj¦  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M¿hj¢  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhj*  ubeh}”(h]”h ]”h"]”h$]”h&]”j 
  j
  j
  hj
  j
  uh1jÅ	  hj&  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j“  hŸh¯h M¸hj—  hžhubhÆ)”}”(hŒµFor example, the following sequence of commands will setup a cpuset
named "Charlie", containing just CPUs 2 and 3, and Memory Node 1,
and then start a subshell 'sh' in that cpuset::”h]”hŒ¼For example, the following sequence of commands will setup a cpuset
named â€œCharlieâ€, containing just CPUs 2 and 3, and Memory Node 1,
and then start a subshell â€˜shâ€™ in that cpuset:”…””}”(hjÆ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MÁhj—  hžhubj  )”}”(hX*  mount -t cgroup -ocpuset cpuset /sys/fs/cgroup/cpuset
cd /sys/fs/cgroup/cpuset
mkdir Charlie
cd Charlie
/bin/echo 2-3 > cpuset.cpus
/bin/echo 1 > cpuset.mems
/bin/echo $$ > tasks
sh
# The subshell 'sh' is now running in cpuset Charlie
# The next line should display '/Charlie'
cat /proc/self/cpuset”h]”hX*  mount -t cgroup -ocpuset cpuset /sys/fs/cgroup/cpuset
cd /sys/fs/cgroup/cpuset
mkdir Charlie
cd Charlie
/bin/echo 2-3 > cpuset.cpus
/bin/echo 1 > cpuset.mems
/bin/echo $$ > tasks
sh
# The subshell 'sh' is now running in cpuset Charlie
# The next line should display '/Charlie'
cat /proc/self/cpuset”…””}”hjÔ  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h MÅhj—  hžhubhÆ)”}”(hŒ*There are ways to query or modify cpusets:”h]”hŒ*There are ways to query or modify cpusets:”…””}”(hjâ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MÑhj—  hžhubj”  )”}”(hX>  - via the cpuset file system directly, using the various cd, mkdir, echo,
  cat, rmdir commands from the shell, or their equivalent from C.
- via the C library libcpuset.
- via the C library libcgroup.
  (https://github.com/libcgroup/libcgroup/)
- via the python application cset.
  (http://code.google.com/p/cpuset/)
”h]”hö)”}”(hhh]”(hû)”}”(hŒ‡via the cpuset file system directly, using the various cd, mkdir, echo,
cat, rmdir commands from the shell, or their equivalent from C.”h]”hÆ)”}”(hŒ‡via the cpuset file system directly, using the various cd, mkdir, echo,
cat, rmdir commands from the shell, or their equivalent from C.”h]”hŒ‡via the cpuset file system directly, using the various cd, mkdir, echo,
cat, rmdir commands from the shell, or their equivalent from C.”…””}”(hjû  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MÓhj÷  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjô  ubhû)”}”(hŒvia the C library libcpuset.”h]”hÆ)”}”(hj  h]”hŒvia the C library libcpuset.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MÕhj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjô  ubhû)”}”(hŒFvia the C library libcgroup.
(https://github.com/libcgroup/libcgroup/)”h]”hÆ)”}”(hŒFvia the C library libcgroup.
(https://github.com/libcgroup/libcgroup/)”h]”(hŒvia the C library libcgroup.
(”…””}”(hj*  hžhhŸNh NubhÞ)”}”(hŒ'https://github.com/libcgroup/libcgroup/”h]”hŒ'https://github.com/libcgroup/libcgroup/”…””}”(hj2  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”Œrefuri”j4  uh1hÝhj*  ubhŒ)”…””}”(hj*  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MÖhj&  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjô  ubhû)”}”(hŒDvia the python application cset.
(http://code.google.com/p/cpuset/)
”h]”hÆ)”}”(hŒCvia the python application cset.
(http://code.google.com/p/cpuset/)”h]”(hŒ"via the python application cset.
(”…””}”(hjU  hžhhŸNh NubhÞ)”}”(hŒ http://code.google.com/p/cpuset/”h]”hŒ http://code.google.com/p/cpuset/”…””}”(hj]  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”Œrefuri”j_  uh1hÝhjU  ubhŒ)”…””}”(hjU  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MØhjQ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1húhjô  ubeh}”(h]”h ]”h"]”h$]”h&]”jÆ  jÇ  uh1hõhŸh¯h MÓhjð  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j“  hŸh¯h MÓhj—  hžhubhÆ)”}”(hŒîThe sched_setaffinity calls can also be done at the shell prompt using
SGI's runon or Robert Love's taskset.  The mbind and set_mempolicy
calls can be done at the shell prompt using the numactl command
(part of Andi Kleen's numa package).”h]”hŒôThe sched_setaffinity calls can also be done at the shell prompt using
SGIâ€™s runon or Robert Loveâ€™s taskset.  The mbind and set_mempolicy
calls can be done at the shell prompt using the numactl command
(part of Andi Kleenâ€™s numa package).”…””}”(hjˆ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MÛhj—  hžhubeh}”(h]”Œhow-do-i-use-cpusets”ah ]”h"]”Œ1.9 how do i use cpusets ?”ah$]”h&]”uh1h°hjÚ  hžhhŸh¯h Miubeh}”(h]”Œid2”ah ]”h"]”Œ
1. cpusets”ah$]”h&]”uh1h°hh²hžhhŸh¯h K&ubh±)”}”(hhh]”(h¶)”}”(hŒ2. Usage Examples and Syntax”h]”hŒ2. Usage Examples and Syntax”…””}”(hj©  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhj¦  hžhhŸh¯h Máubh±)”}”(hhh]”(h¶)”}”(hŒ2.1 Basic Usage”h]”hŒ2.1 Basic Usage”…””}”(hjº  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhj·  hžhhŸh¯h MäubhÆ)”}”(hŒYCreating, modifying, using the cpusets can be done through the cpuset
virtual filesystem.”h]”hŒYCreating, modifying, using the cpusets can be done through the cpuset
virtual filesystem.”…””}”(hjÈ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mæhj·  hžhubhÆ)”}”(hŒKTo mount it, type:
# mount -t cgroup -o cpuset cpuset /sys/fs/cgroup/cpuset”h]”hŒKTo mount it, type:
# mount -t cgroup -o cpuset cpuset /sys/fs/cgroup/cpuset”…””}”(hjÖ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Méhj·  hžhubhÆ)”}”(hŒ¾Then under /sys/fs/cgroup/cpuset you can find a tree that corresponds to the
tree of the cpusets in the system. For instance, /sys/fs/cgroup/cpuset
is the cpuset that holds the whole system.”h]”hŒ¾Then under /sys/fs/cgroup/cpuset you can find a tree that corresponds to the
tree of the cpusets in the system. For instance, /sys/fs/cgroup/cpuset
is the cpuset that holds the whole system.”…””}”(hjä  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mìhj·  hžhubhÆ)”}”(hŒ@If you want to create a new cpuset under /sys/fs/cgroup/cpuset::”h]”hŒ?If you want to create a new cpuset under /sys/fs/cgroup/cpuset:”…””}”(hjò  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mðhj·  hžhubj  )”}”(hŒ,# cd /sys/fs/cgroup/cpuset
# mkdir my_cpuset”h]”hŒ,# cd /sys/fs/cgroup/cpuset
# mkdir my_cpuset”…””}”hj   sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h Mòhj·  hžhubhÆ)”}”(hŒ/Now you want to do something with this cpuset::”h]”hŒ.Now you want to do something with this cpuset:”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mõhj·  hžhubj  )”}”(hŒ# cd my_cpuset”h]”hŒ# cd my_cpuset”…””}”hj  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h M÷hj·  hžhubhÆ)”}”(hŒ.In this directory you can find several files::”h]”hŒ-In this directory you can find several files:”…””}”(hj*  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mùhj·  hžhubj  )”}”(hXe  # ls
cgroup.clone_children  cpuset.memory_pressure
cgroup.event_control   cpuset.memory_spread_page
cgroup.procs           cpuset.memory_spread_slab
cpuset.cpu_exclusive   cpuset.mems
cpuset.cpus            cpuset.sched_load_balance
cpuset.mem_exclusive   cpuset.sched_relax_domain_level
cpuset.mem_hardwall    notify_on_release
cpuset.memory_migrate  tasks”h]”hXe  # ls
cgroup.clone_children  cpuset.memory_pressure
cgroup.event_control   cpuset.memory_spread_page
cgroup.procs           cpuset.memory_spread_slab
cpuset.cpu_exclusive   cpuset.mems
cpuset.cpus            cpuset.sched_load_balance
cpuset.mem_exclusive   cpuset.sched_relax_domain_level
cpuset.mem_hardwall    notify_on_release
cpuset.memory_migrate  tasks”…””}”hj8  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h Mûhj·  hžhubhÆ)”}”(hŒ×Reading them will give you information about the state of this cpuset:
the CPUs and Memory Nodes it can use, the processes that are using
it, its properties.  By writing to these files you can manipulate
the cpuset.”h]”hŒ×Reading them will give you information about the state of this cpuset:
the CPUs and Memory Nodes it can use, the processes that are using
it, its properties.  By writing to these files you can manipulate
the cpuset.”…””}”(hjF  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj·  hžhubhÆ)”}”(hŒSet some flags::”h]”hŒSet some flags:”…””}”(hjT  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M
hj·  hžhubj  )”}”(hŒ$# /bin/echo 1 > cpuset.cpu_exclusive”h]”hŒ$# /bin/echo 1 > cpuset.cpu_exclusive”…””}”hjb  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h Mhj·  hžhubhÆ)”}”(hŒAdd some cpus::”h]”hŒAdd some cpus:”…””}”(hjp  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj·  hžhubj  )”}”(hŒ# /bin/echo 0-7 > cpuset.cpus”h]”hŒ# /bin/echo 0-7 > cpuset.cpus”…””}”hj~  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h Mhj·  hžhubhÆ)”}”(hŒAdd some mems::”h]”hŒAdd some mems:”…””}”(hjŒ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj·  hžhubj  )”}”(hŒ# /bin/echo 0-7 > cpuset.mems”h]”hŒ# /bin/echo 0-7 > cpuset.mems”…””}”hjš  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h Mhj·  hžhubhÆ)”}”(hŒ&Now attach your shell to this cpuset::”h]”hŒ%Now attach your shell to this cpuset:”…””}”(hj¨  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj·  hžhubj  )”}”(hŒ# /bin/echo $$ > tasks”h]”hŒ# /bin/echo $$ > tasks”…””}”hj¶  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h Mhj·  hžhubhÆ)”}”(hŒQYou can also create cpusets inside your cpuset by using mkdir in this
directory::”h]”hŒPYou can also create cpusets inside your cpuset by using mkdir in this
directory:”…””}”(hjÄ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj·  hžhubj  )”}”(hŒ# mkdir my_sub_cs”h]”hŒ# mkdir my_sub_cs”…””}”hjÒ  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h Mhj·  hžhubhÆ)”}”(hŒ$To remove a cpuset, just use rmdir::”h]”hŒ#To remove a cpuset, just use rmdir:”…””}”(hjà  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mhj·  hžhubj  )”}”(hŒ# rmdir my_sub_cs”h]”hŒ# rmdir my_sub_cs”…””}”hjî  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h M!hj·  hžhubhÆ)”}”(hŒWThis will fail if the cpuset is in use (has cpusets inside, or has
processes attached).”h]”hŒWThis will fail if the cpuset is in use (has cpusets inside, or has
processes attached).”…””}”(hjü  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M#hj·  hžhubhÆ)”}”(hŒgNote that for legacy reasons, the "cpuset" filesystem exists as a
wrapper around the cgroup filesystem.”h]”hŒkNote that for legacy reasons, the â€œcpusetâ€ filesystem exists as a
wrapper around the cgroup filesystem.”…””}”(hj
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M&hj·  hžhubhÆ)”}”(hŒThe command::”h]”hŒThe command:”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M)hj·  hžhubj  )”}”(hŒ'mount -t cpuset X /sys/fs/cgroup/cpuset”h]”hŒ'mount -t cpuset X /sys/fs/cgroup/cpuset”…””}”hj&  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h M+hj·  hžhubhÆ)”}”(hŒis equivalent to::”h]”hŒis equivalent to:”…””}”(hj4  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M-hj·  hžhubj  )”}”(hŒmount -t cgroup -ocpuset,noprefix X /sys/fs/cgroup/cpuset
echo "/sbin/cpuset_release_agent" > /sys/fs/cgroup/cpuset/release_agent”h]”hŒmount -t cgroup -ocpuset,noprefix X /sys/fs/cgroup/cpuset
echo "/sbin/cpuset_release_agent" > /sys/fs/cgroup/cpuset/release_agent”…””}”hjB  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h M/hj·  hžhubeh}”(h]”Œbasic-usage”ah ]”h"]”Œ2.1 basic usage”ah$]”h&]”uh1h°hj¦  hžhhŸh¯h Mäubh±)”}”(hhh]”(h¶)”}”(hŒ2.2 Adding/removing cpus”h]”hŒ2.2 Adding/removing cpus”…””}”(hj[  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhjX  hžhhŸh¯h M3ubhÆ)”}”(hŒXThis is the syntax to use when writing in the cpus or mems files
in cpuset directories::”h]”hŒWThis is the syntax to use when writing in the cpus or mems files
in cpuset directories:”…””}”(hji  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M5hjX  hžhubj  )”}”(hŒ# /bin/echo 1-4 > cpuset.cpus         -> set cpus list to cpus 1,2,3,4
# /bin/echo 1,2,3,4 > cpuset.cpus     -> set cpus list to cpus 1,2,3,4”h]”hŒ# /bin/echo 1-4 > cpuset.cpus         -> set cpus list to cpus 1,2,3,4
# /bin/echo 1,2,3,4 > cpuset.cpus     -> set cpus list to cpus 1,2,3,4”…””}”hjw  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h M8hjX  hžhubhÆ)”}”(hŒrTo add a CPU to a cpuset, write the new list of CPUs including the
CPU to be added. To add 6 to the above cpuset::”h]”hŒqTo add a CPU to a cpuset, write the new list of CPUs including the
CPU to be added. To add 6 to the above cpuset:”…””}”(hj…  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M;hjX  hžhubj  )”}”(hŒH# /bin/echo 1-4,6 > cpuset.cpus       -> set cpus list to cpus 1,2,3,4,6”h]”hŒH# /bin/echo 1-4,6 > cpuset.cpus       -> set cpus list to cpus 1,2,3,4,6”…””}”hj“  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h M>hjX  hžhubhÆ)”}”(hŒbSimilarly to remove a CPU from a cpuset, write the new list of CPUs
without the CPU to be removed.”h]”hŒbSimilarly to remove a CPU from a cpuset, write the new list of CPUs
without the CPU to be removed.”…””}”(hj¡  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h M@hjX  hžhubhÆ)”}”(hŒTo remove all the CPUs::”h]”hŒTo remove all the CPUs:”…””}”(hj¯  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MChjX  hžhubj  )”}”(hŒ8# /bin/echo "" > cpuset.cpus          -> clear cpus list”h]”hŒ8# /bin/echo "" > cpuset.cpus          -> clear cpus list”…””}”hj½  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h MEhjX  hžhubeh}”(h]”Œadding-removing-cpus”ah ]”h"]”Œ2.2 adding/removing cpus”ah$]”h&]”uh1h°hj¦  hžhhŸh¯h M3ubh±)”}”(hhh]”(h¶)”}”(hŒ2.3 Setting flags”h]”hŒ2.3 Setting flags”…””}”(hjÖ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhjÓ  hžhhŸh¯h MHubhÆ)”}”(hŒThe syntax is very simple::”h]”hŒThe syntax is very simple:”…””}”(hjä  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MJhjÓ  hžhubj  )”}”(hŒ“# /bin/echo 1 > cpuset.cpu_exclusive  -> set flag 'cpuset.cpu_exclusive'
# /bin/echo 0 > cpuset.cpu_exclusive  -> unset flag 'cpuset.cpu_exclusive'”h]”hŒ“# /bin/echo 1 > cpuset.cpu_exclusive  -> set flag 'cpuset.cpu_exclusive'
# /bin/echo 0 > cpuset.cpu_exclusive  -> unset flag 'cpuset.cpu_exclusive'”…””}”hjò  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h MLhjÓ  hžhubeh}”(h]”Œsetting-flags”ah ]”h"]”Œ2.3 setting flags”ah$]”h&]”uh1h°hj¦  hžhhŸh¯h MHubh±)”}”(hhh]”(h¶)”}”(hŒ2.4 Attaching processes”h]”hŒ2.4 Attaching processes”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhj  hžhhŸh¯h MPubj  )”}”(hŒ# /bin/echo PID > tasks”h]”hŒ# /bin/echo PID > tasks”…””}”hj  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h MThj  hžhubhÆ)”}”(hŒ‘Note that it is PID, not PIDs. You can only attach ONE task at a time.
If you have several tasks to attach, you have to do it one after another::”h]”hŒNote that it is PID, not PIDs. You can only attach ONE task at a time.
If you have several tasks to attach, you have to do it one after another:”…””}”(hj'  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h MVhj  hžhubj  )”}”(hŒT# /bin/echo PID1 > tasks
# /bin/echo PID2 > tasks
      ...
# /bin/echo PIDn > tasks”h]”hŒT# /bin/echo PID1 > tasks
# /bin/echo PID2 > tasks
      ...
# /bin/echo PIDn > tasks”…””}”hj5  sbah}”(h]”h ]”h"]”h$]”h&]”jØ  jÙ  uh1j  hŸh¯h MYhj  hžhubeh}”(h]”Œattaching-processes”ah ]”h"]”Œ2.4 attaching processes”ah$]”h&]”uh1h°hj¦  hžhhŸh¯h MPubeh}”(h]”Œusage-examples-and-syntax”ah ]”h"]”Œ2. usage examples and syntax”ah$]”h&]”uh1h°hh²hžhhŸh¯h Máubh±)”}”(hhh]”(h¶)”}”(hŒ3. Questions”h]”hŒ3. Questions”…””}”(hjV  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhjS  hžhhŸh¯h M`ubjè  )”}”(hhh]”(jí  )”}”(hŒ%Q:
what's up with this '/bin/echo' ?
”h]”(jó  )”}”(hŒQ:”h]”hŒQ:”…””}”(hjk  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jò  hŸh¯h Mchjg  ubj  )”}”(hhh]”hÆ)”}”(hŒ!what's up with this '/bin/echo' ?”h]”hŒ'whatâ€™s up with this â€˜/bin/echoâ€™ ?”…””}”(hj|  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mchjy  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j  hjg  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jì  hŸh¯h Mchjd  ubjí  )”}”(hŒ»A:
bash's builtin 'echo' command does not check calls to write() against
errors. If you use it in the cpuset file system, you won't be
able to tell whether a command succeeded or failed.
”h]”(jó  )”}”(hŒA:”h]”hŒA:”…””}”(hjš  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jò  hŸh¯h Mhhj–  ubj  )”}”(hhh]”hÆ)”}”(hŒ·bash's builtin 'echo' command does not check calls to write() against
errors. If you use it in the cpuset file system, you won't be
able to tell whether a command succeeded or failed.”h]”hŒ¿bashâ€™s builtin â€˜echoâ€™ command does not check calls to write() against
errors. If you use it in the cpuset file system, you wonâ€™t be
able to tell whether a command succeeded or failed.”…””}”(hj«  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mfhj¨  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j  hj–  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jì  hŸh¯h Mhhjd  hžhubjí  )”}”(hŒNQ:
When I attach processes, only the first of the line gets really attached !
”h]”(jó  )”}”(hŒQ:”h]”hŒQ:”…””}”(hjÉ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jò  hŸh¯h MkhjÅ  ubj  )”}”(hhh]”hÆ)”}”(hŒJWhen I attach processes, only the first of the line gets really attached !”h]”hŒJWhen I attach processes, only the first of the line gets really attached !”…””}”(hjÚ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mkhj×  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j  hjÅ  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jì  hŸh¯h Mkhjd  hžhubjí  )”}”(hŒ_A:
We can only return one error code per call to write(). So you should also
put only ONE pid.
”h]”(jó  )”}”(hŒA:”h]”hŒA:”…””}”(hjø  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jò  hŸh¯h Mohjô  ubj  )”}”(hhh]”hÆ)”}”(hŒ[We can only return one error code per call to write(). So you should also
put only ONE pid.”h]”hŒ[We can only return one error code per call to write(). So you should also
put only ONE pid.”…””}”(hj	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mnhj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1j  hjô  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jì  hŸh¯h Mohjd  hžhubeh}”(h]”h ]”h"]”h$]”h&]”uh1jç  hjS  hžhhŸh¯h Nubeh}”(h]”Œ	questions”ah ]”h"]”Œ3. questions”ah$]”h&]”uh1h°hh²hžhhŸh¯h M`ubh±)”}”(hhh]”(h¶)”}”(hŒ
4. Contact”h]”hŒ
4. Contact”…””}”(hj4  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hµhj1  hžhhŸh¯h MrubhÆ)”}”(hŒ)Web: http://www.bullopensource.org/cpuset”h]”(hŒWeb: ”…””}”(hjB  hžhhŸNh NubhÞ)”}”(hŒ$http://www.bullopensource.org/cpuset”h]”hŒ$http://www.bullopensource.org/cpuset”…””}”(hjJ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”Œrefuri”jL  uh1hÝhjB  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÅhŸh¯h Mthj1  hžhubeh}”(h]”Œcontact”ah ]”h"]”Œ
4. contact”ah$]”h&]”uh1h°hh²hžhhŸh¯h Mrubeh}”(h]”(h®Œid1”eh ]”h"]”Œcpusets”ah$]”Œcpusets”ah&]”uh1h°hhhžhhŸh¯h KŒ
referenced”KŒexpect_referenced_by_name”}”jl  h£sŒexpect_referenced_by_id”}”h®h£subeh}”(h]”h ]”h"]”h$]”h&]”Œsource”h¯uh1hŒcurrent_source”NŒcurrent_line”NŒsettings”Œdocutils.frontend”ŒValues”“”)”}”(hµNŒ	generator”NŒ	datestamp”NŒsource_link”NŒ
source_url”NŒtoc_backlinks”jM  Œfootnote_backlinks”KŒsectnum_xform”KŒstrip_comments”NŒstrip_elements_with_classes”NŒstrip_classes”NŒreport_level”KŒ
halt_level”KŒexit_status_level”KŒdebug”NŒwarning_stream”NŒ	traceback”ˆŒinput_encoding”Œ	utf-8-sig”Œinput_encoding_error_handler”Œstrict”Œoutput_encoding”Œutf-8”Œoutput_encoding_error_handler”j—  Œerror_encoding”Œutf-8”Œerror_encoding_error_handler”Œbackslashreplace”Œlanguage_code”Œen”Œrecord_dependencies”NŒconfig”NŒ	id_prefix”hŒauto_id_prefix”Œid”Œdump_settings”NŒdump_internals”NŒdump_transforms”NŒdump_pseudo_xml”NŒexpose_internals”NŒstrict_visitor”NŒ_disable_config”NŒ_source”h¯Œ_destination”NŒ_config_files”]”Œ7/var/lib/git/docbuild/linux/Documentation/docutils.conf”aŒfile_insertion_enabled”ˆŒraw_enabled”KŒline_length_limit”M'Œpep_references”NŒpep_base_url”Œhttps://peps.python.org/”Œpep_file_url_template”Œpep-%04d”Œrfc_references”NŒrfc_base_url”Œ&https://datatracker.ietf.org/doc/html/”Œ	tab_width”KŒtrim_footnote_reference_space”‰Œsyntax_highlight”Œlong”Œsmart_quotes”ˆŒsmartquotes_locales”]”Œcharacter_level_inline_markup”‰Œdoctitle_xform”‰Œdocinfo_xform”KŒsectsubtitle_xform”‰Œimage_loading”Œlink”Œembed_stylesheet”‰Œcloak_email_addresses”ˆŒsection_self_link”‰Œenv”NubŒreporter”NŒindirect_targets”]”Œsubstitution_defs”}”Œsubstitution_names”}”Œrefnames”}”Œrefids”}”h®]”h£asŒnameids”}”(jl  h®j£  j   jG  jD  j  j  jR  jO  j›  j˜  jŠ  j‡  jg	  jd	  jª
  j§
  j»  j¸  j”  j‘  j›  j˜  jP  jM  jU  jR  jÐ  jÍ  j  j  jH  jE  j.  j+  jd  ja  uŒ	nametypes”}”(jl  ˆj£  ‰jG  ‰j  ‰jR  ‰j›  ‰jŠ  ‰jg	  ‰jª
  ‰j»  ‰j”  ‰j›  ‰jP  ‰jU  ‰jÐ  ‰j  ‰jH  ‰j.  ‰jd  ‰uh}”(h®h²ji  h²j   jÚ  jD  jë  j  jJ  jO  j  j˜  jU  j‡  jž  jd	  j  j§
  jj	  j¸  j­
  j‘  j¾  j˜  j—  jM  j¦  jR  j·  jÍ  jX  j  jÓ  jE  j  j+  jS  ja  j1  uŒfootnote_refs”}”Œcitation_refs”}”Œautofootnotes”]”Œautofootnote_refs”]”Œsymbol_footnotes”]”Œsymbol_footnote_refs”]”Œ	footnotes”]”Œ	citations”]”Œautofootnote_start”KŒsymbol_footnote_start”K Œ
id_counter”Œcollections”ŒCounter”“”}”j¥  Ks…”R”Œparse_messages”]”hŒsystem_message”“”)”}”(hhh]”hÆ)”}”(hŒ*Duplicate implicit target name: "cpusets".”h]”hŒ.Duplicate implicit target name: â€œcpusetsâ€.”…””}”(hjÿ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhjü  ubah}”(h]”h ]”h"]”h$]”h&]”ji  aŒlevel”KŒtype”ŒINFO”Œsource”h¯Œline”Kuh1jú  hh²hžhhŸh¯h KubaŒtransform_messages”]”jû  )”}”(hhh]”hÆ)”}”(hhh]”hŒ-Hyperlink target "cpusets" is not referenced.”…””}”hj  sbah}”(h]”h ]”h"]”h$]”h&]”uh1hÅhj  ubah}”(h]”h ]”h"]”h$]”h&]”Œlevel”KŒtype”j  Œsource”h¯Œline”Kuh1jú  ubaŒtransformer”NŒinclude_log”]”Œ
decoration”Nhžhub.