1 ================== 1 ==================
2 HugeTLB Controller 2 HugeTLB Controller
3 ================== 3 ==================
4 4
5 HugeTLB controller can be created by first mou 5 HugeTLB controller can be created by first mounting the cgroup filesystem.
6 6
7 # mount -t cgroup -o hugetlb none /sys/fs/cgro 7 # mount -t cgroup -o hugetlb none /sys/fs/cgroup
8 8
9 With the above step, the initial or the parent 9 With the above step, the initial or the parent HugeTLB group becomes
10 visible at /sys/fs/cgroup. At bootup, this gro 10 visible at /sys/fs/cgroup. At bootup, this group includes all the tasks in
11 the system. /sys/fs/cgroup/tasks lists the tas 11 the system. /sys/fs/cgroup/tasks lists the tasks in this cgroup.
12 12
13 New groups can be created under the parent gro 13 New groups can be created under the parent group /sys/fs/cgroup::
14 14
15 # cd /sys/fs/cgroup 15 # cd /sys/fs/cgroup
16 # mkdir g1 16 # mkdir g1
17 # echo $$ > g1/tasks 17 # echo $$ > g1/tasks
18 18
19 The above steps create a new group g1 and move 19 The above steps create a new group g1 and move the current shell
20 process (bash) into it. 20 process (bash) into it.
21 21
22 Brief summary of control files:: 22 Brief summary of control files::
23 23
24 hugetlb.<hugepagesize>.rsvd.limit_in_bytes 24 hugetlb.<hugepagesize>.rsvd.limit_in_bytes # set/show limit of "hugepagesize" hugetlb reservations
25 hugetlb.<hugepagesize>.rsvd.max_usage_in_byte 25 hugetlb.<hugepagesize>.rsvd.max_usage_in_bytes # show max "hugepagesize" hugetlb reservations and no-reserve faults
26 hugetlb.<hugepagesize>.rsvd.usage_in_bytes 26 hugetlb.<hugepagesize>.rsvd.usage_in_bytes # show current reservations and no-reserve faults for "hugepagesize" hugetlb
27 hugetlb.<hugepagesize>.rsvd.failcnt 27 hugetlb.<hugepagesize>.rsvd.failcnt # show the number of allocation failure due to HugeTLB reservation limit
28 hugetlb.<hugepagesize>.limit_in_bytes 28 hugetlb.<hugepagesize>.limit_in_bytes # set/show limit of "hugepagesize" hugetlb faults
29 hugetlb.<hugepagesize>.max_usage_in_bytes 29 hugetlb.<hugepagesize>.max_usage_in_bytes # show max "hugepagesize" hugetlb usage recorded
30 hugetlb.<hugepagesize>.usage_in_bytes 30 hugetlb.<hugepagesize>.usage_in_bytes # show current usage for "hugepagesize" hugetlb
31 hugetlb.<hugepagesize>.failcnt 31 hugetlb.<hugepagesize>.failcnt # show the number of allocation failure due to HugeTLB usage limit
32 hugetlb.<hugepagesize>.numa_stat 32 hugetlb.<hugepagesize>.numa_stat # show the numa information of the hugetlb memory charged to this cgroup
33 33
34 For a system supporting three hugepage sizes ( 34 For a system supporting three hugepage sizes (64k, 32M and 1G), the control
35 files include:: 35 files include::
36 36
37 hugetlb.1GB.limit_in_bytes 37 hugetlb.1GB.limit_in_bytes
38 hugetlb.1GB.max_usage_in_bytes 38 hugetlb.1GB.max_usage_in_bytes
39 hugetlb.1GB.numa_stat 39 hugetlb.1GB.numa_stat
40 hugetlb.1GB.usage_in_bytes 40 hugetlb.1GB.usage_in_bytes
41 hugetlb.1GB.failcnt 41 hugetlb.1GB.failcnt
42 hugetlb.1GB.rsvd.limit_in_bytes 42 hugetlb.1GB.rsvd.limit_in_bytes
43 hugetlb.1GB.rsvd.max_usage_in_bytes 43 hugetlb.1GB.rsvd.max_usage_in_bytes
44 hugetlb.1GB.rsvd.usage_in_bytes 44 hugetlb.1GB.rsvd.usage_in_bytes
45 hugetlb.1GB.rsvd.failcnt 45 hugetlb.1GB.rsvd.failcnt
46 hugetlb.64KB.limit_in_bytes 46 hugetlb.64KB.limit_in_bytes
47 hugetlb.64KB.max_usage_in_bytes 47 hugetlb.64KB.max_usage_in_bytes
48 hugetlb.64KB.numa_stat 48 hugetlb.64KB.numa_stat
49 hugetlb.64KB.usage_in_bytes 49 hugetlb.64KB.usage_in_bytes
50 hugetlb.64KB.failcnt 50 hugetlb.64KB.failcnt
51 hugetlb.64KB.rsvd.limit_in_bytes 51 hugetlb.64KB.rsvd.limit_in_bytes
52 hugetlb.64KB.rsvd.max_usage_in_bytes 52 hugetlb.64KB.rsvd.max_usage_in_bytes
53 hugetlb.64KB.rsvd.usage_in_bytes 53 hugetlb.64KB.rsvd.usage_in_bytes
54 hugetlb.64KB.rsvd.failcnt 54 hugetlb.64KB.rsvd.failcnt
55 hugetlb.32MB.limit_in_bytes 55 hugetlb.32MB.limit_in_bytes
56 hugetlb.32MB.max_usage_in_bytes 56 hugetlb.32MB.max_usage_in_bytes
57 hugetlb.32MB.numa_stat 57 hugetlb.32MB.numa_stat
58 hugetlb.32MB.usage_in_bytes 58 hugetlb.32MB.usage_in_bytes
59 hugetlb.32MB.failcnt 59 hugetlb.32MB.failcnt
60 hugetlb.32MB.rsvd.limit_in_bytes 60 hugetlb.32MB.rsvd.limit_in_bytes
61 hugetlb.32MB.rsvd.max_usage_in_bytes 61 hugetlb.32MB.rsvd.max_usage_in_bytes
62 hugetlb.32MB.rsvd.usage_in_bytes 62 hugetlb.32MB.rsvd.usage_in_bytes
63 hugetlb.32MB.rsvd.failcnt 63 hugetlb.32MB.rsvd.failcnt
64 64
65 65
66 1. Page fault accounting 66 1. Page fault accounting
67 67
68 :: 68 ::
69 69
70 hugetlb.<hugepagesize>.limit_in_bytes 70 hugetlb.<hugepagesize>.limit_in_bytes
71 hugetlb.<hugepagesize>.max_usage_in_bytes 71 hugetlb.<hugepagesize>.max_usage_in_bytes
72 hugetlb.<hugepagesize>.usage_in_bytes 72 hugetlb.<hugepagesize>.usage_in_bytes
73 hugetlb.<hugepagesize>.failcnt 73 hugetlb.<hugepagesize>.failcnt
74 74
75 The HugeTLB controller allows users to limit t 75 The HugeTLB controller allows users to limit the HugeTLB usage (page fault) per
76 control group and enforces the limit during pa 76 control group and enforces the limit during page fault. Since HugeTLB
77 doesn't support page reclaim, enforcing the li 77 doesn't support page reclaim, enforcing the limit at page fault time implies
78 that, the application will get SIGBUS signal i 78 that, the application will get SIGBUS signal if it tries to fault in HugeTLB
79 pages beyond its limit. Therefore the applicat 79 pages beyond its limit. Therefore the application needs to know exactly how many
80 HugeTLB pages it uses before hand, and the sys 80 HugeTLB pages it uses before hand, and the sysadmin needs to make sure that
81 there are enough available on the machine for 81 there are enough available on the machine for all the users to avoid processes
82 getting SIGBUS. 82 getting SIGBUS.
83 83
84 84
85 2. Reservation accounting 85 2. Reservation accounting
86 86
87 :: 87 ::
88 88
89 hugetlb.<hugepagesize>.rsvd.limit_in_bytes 89 hugetlb.<hugepagesize>.rsvd.limit_in_bytes
90 hugetlb.<hugepagesize>.rsvd.max_usage_in_byt 90 hugetlb.<hugepagesize>.rsvd.max_usage_in_bytes
91 hugetlb.<hugepagesize>.rsvd.usage_in_bytes 91 hugetlb.<hugepagesize>.rsvd.usage_in_bytes
92 hugetlb.<hugepagesize>.rsvd.failcnt 92 hugetlb.<hugepagesize>.rsvd.failcnt
93 93
94 The HugeTLB controller allows to limit the Hug 94 The HugeTLB controller allows to limit the HugeTLB reservations per control
95 group and enforces the controller limit at res 95 group and enforces the controller limit at reservation time and at the fault of
96 HugeTLB memory for which no reservation exists 96 HugeTLB memory for which no reservation exists. Since reservation limits are
97 enforced at reservation time (on mmap or shget 97 enforced at reservation time (on mmap or shget), reservation limits never causes
98 the application to get SIGBUS signal if the me 98 the application to get SIGBUS signal if the memory was reserved before hand. For
99 MAP_NORESERVE allocations, the reservation lim 99 MAP_NORESERVE allocations, the reservation limit behaves the same as the fault
100 limit, enforcing memory usage at fault time an 100 limit, enforcing memory usage at fault time and causing the application to
101 receive a SIGBUS if it's crossing its limit. 101 receive a SIGBUS if it's crossing its limit.
102 102
103 Reservation limits are superior to page fault 103 Reservation limits are superior to page fault limits described above, since
104 reservation limits are enforced at reservation 104 reservation limits are enforced at reservation time (on mmap or shget), and
105 never causes the application to get SIGBUS sig 105 never causes the application to get SIGBUS signal if the memory was reserved
106 before hand. This allows for easier fallback t 106 before hand. This allows for easier fallback to alternatives such as
107 non-HugeTLB memory for example. In the case of 107 non-HugeTLB memory for example. In the case of page fault accounting, it's very
108 hard to avoid processes getting SIGBUS since t 108 hard to avoid processes getting SIGBUS since the sysadmin needs precisely know
109 the HugeTLB usage of all the tasks in the syst 109 the HugeTLB usage of all the tasks in the system and make sure there is enough
110 pages to satisfy all requests. Avoiding tasks 110 pages to satisfy all requests. Avoiding tasks getting SIGBUS on overcommited
111 systems is practically impossible with page fa 111 systems is practically impossible with page fault accounting.
112 112
113 113
114 3. Caveats with shared memory 114 3. Caveats with shared memory
115 115
116 For shared HugeTLB memory, both HugeTLB reserv 116 For shared HugeTLB memory, both HugeTLB reservation and page faults are charged
117 to the first task that causes the memory to be 117 to the first task that causes the memory to be reserved or faulted, and all
118 subsequent uses of this reserved or faulted me 118 subsequent uses of this reserved or faulted memory is done without charging.
119 119
120 Shared HugeTLB memory is only uncharged when i 120 Shared HugeTLB memory is only uncharged when it is unreserved or deallocated.
121 This is usually when the HugeTLB file is delet 121 This is usually when the HugeTLB file is deleted, and not when the task that
122 caused the reservation or fault has exited. 122 caused the reservation or fault has exited.
123 123
124 124
125 4. Caveats with HugeTLB cgroup offline. 125 4. Caveats with HugeTLB cgroup offline.
126 126
127 When a HugeTLB cgroup goes offline with some r 127 When a HugeTLB cgroup goes offline with some reservations or faults still
128 charged to it, the behavior is as follows: 128 charged to it, the behavior is as follows:
129 129
130 - The fault charges are charged to the parent 130 - The fault charges are charged to the parent HugeTLB cgroup (reparented),
131 - the reservation charges remain on the offlin 131 - the reservation charges remain on the offline HugeTLB cgroup.
132 132
133 This means that if a HugeTLB cgroup gets offli 133 This means that if a HugeTLB cgroup gets offlined while there is still HugeTLB
134 reservations charged to it, that cgroup persis 134 reservations charged to it, that cgroup persists as a zombie until all HugeTLB
135 reservations are uncharged. HugeTLB reservatio 135 reservations are uncharged. HugeTLB reservations behave in this manner to match
136 the memory controller whose cgroups also persi 136 the memory controller whose cgroups also persist as zombie until all charged
137 memory is uncharged. Also, the tracking of Hug 137 memory is uncharged. Also, the tracking of HugeTLB reservations is a bit more
138 complex compared to the tracking of HugeTLB fa 138 complex compared to the tracking of HugeTLB faults, so it is significantly
139 harder to reparent reservations at offline tim 139 harder to reparent reservations at offline time.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.