1 .. SPDX-License-Identifier: GPL-2.0
7 :Author: David Rientjes <rientjes@cs.washington.edu>
9 Using numa=fake and CPUSets for Resource Management
11 This document describes how the numa=fake x86_64 command-line option can be used
12 in conjunction with cpusets for coarse memory management. Using this feature,
13 you can create fake NUMA nodes that represent contiguous chunks of memory and
14 assign them to cpusets and their attached tasks. This is a way of limiting the
15 amount of system memory that are available to a certain class of tasks.
17 For more information on the features of cpusets, see
18 Documentation/admin-guide/cgroup-v1/cpusets.rst.
19 There are a number of different configurations you can use for your needs. For
20 more information on the numa=fake command line option and its various ways of
21 configuring fake nodes, see Documentation/x86/x86_64/boot-options.rst.
23 For the purposes of this introduction, we'll assume a very primitive NUMA
24 emulation setup of "numa=fake=4*512,". This will split our system memory into
25 four equal chunks of 512M each that we can now use to assign to cpusets. As
26 you become more familiar with using this combination for resource control,
27 you'll determine a better setup to minimize the number of nodes you have to deal
30 A machine may be split as follows with "numa=fake=4*512," as reported by dmesg::
32 Faking node 0 at 0000000000000000-0000000020000000 (512MB)
33 Faking node 1 at 0000000020000000-0000000040000000 (512MB)
34 Faking node 2 at 0000000040000000-0000000060000000 (512MB)
35 Faking node 3 at 0000000060000000-0000000080000000 (512MB)
37 On node 0 totalpages: 130975
38 On node 1 totalpages: 131072
39 On node 2 totalpages: 131072
40 On node 3 totalpages: 131072
42 Now following the instructions for mounting the cpusets filesystem from
43 Documentation/admin-guide/cgroup-v1/cpusets.rst, you can assign fake nodes (i.e. contiguous memory
44 address spaces) to individual cpusets::
46 [root@xroads /]# mkdir exampleset
47 [root@xroads /]# mount -t cpuset none exampleset
48 [root@xroads /]# mkdir exampleset/ddset
49 [root@xroads /]# cd exampleset/ddset
50 [root@xroads /exampleset/ddset]# echo 0-1 > cpus
51 [root@xroads /exampleset/ddset]# echo 0-1 > mems
53 Now this cpuset, 'ddset', will only allowed access to fake nodes 0 and 1 for
54 memory allocations (1G).
56 You can now assign tasks to these cpusets to limit the memory resources
57 available to them according to the fake nodes assigned as mems::
59 [root@xroads /exampleset/ddset]# echo $$ > tasks
60 [root@xroads /exampleset/ddset]# dd if=/dev/zero of=tmp bs=1024 count=1G
63 Notice the difference between the system memory usage as reported by
64 /proc/meminfo between the restricted cpuset case above and the unrestricted
65 case (i.e. running the same 'dd' command without assigning it to a fake NUMA
68 ======== ============ ==========
69 Name Unrestricted Restricted
70 ======== ============ ==========
71 MemTotal 3091900 kB 3091900 kB
72 MemFree 42113 kB 1513236 kB
73 ======== ============ ==========
75 This allows for coarse memory management for the tasks you assign to particular
76 cpusets. Since cpusets can form a hierarchy, you can create some pretty
77 interesting combinations of use-cases for various classes of tasks for your
78 memory management needs.