3 * Copyright IBM Corporation, 2012
4 * Author Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
7 * Copyright (C) 2019 Red Hat, Inc.
8 * Author: Giuseppe Scrivano <gscrivan@redhat.com>
10 * This program is free software; you can redistribute it and/or modify it
11 * under the terms of version 2.1 of the GNU Lesser General Public License
12 * as published by the Free Software Foundation.
14 * This program is distributed in the hope that it would be useful, but
15 * WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
20 #include <linux/cgroup.h>
21 #include <linux/page_counter.h>
22 #include <linux/slab.h>
23 #include <linux/hugetlb.h>
24 #include <linux/hugetlb_cgroup.h>
26 #define MEMFILE_PRIVATE(x, val) (((x) << 16) | (val))
27 #define MEMFILE_IDX(val) (((val) >> 16) & 0xffff)
28 #define MEMFILE_ATTR(val) ((val) & 0xffff)
30 static struct hugetlb_cgroup
*root_h_cgroup __read_mostly
;
32 static inline struct page_counter
*
33 __hugetlb_cgroup_counter_from_cgroup(struct hugetlb_cgroup
*h_cg
, int idx
,
37 return &h_cg
->rsvd_hugepage
[idx
];
38 return &h_cg
->hugepage
[idx
];
41 static inline struct page_counter
*
42 hugetlb_cgroup_counter_from_cgroup(struct hugetlb_cgroup
*h_cg
, int idx
)
44 return __hugetlb_cgroup_counter_from_cgroup(h_cg
, idx
, false);
47 static inline struct page_counter
*
48 hugetlb_cgroup_counter_from_cgroup_rsvd(struct hugetlb_cgroup
*h_cg
, int idx
)
50 return __hugetlb_cgroup_counter_from_cgroup(h_cg
, idx
, true);
54 struct hugetlb_cgroup
*hugetlb_cgroup_from_css(struct cgroup_subsys_state
*s
)
56 return s
? container_of(s
, struct hugetlb_cgroup
, css
) : NULL
;
60 struct hugetlb_cgroup
*hugetlb_cgroup_from_task(struct task_struct
*task
)
62 return hugetlb_cgroup_from_css(task_css(task
, hugetlb_cgrp_id
));
65 static inline bool hugetlb_cgroup_is_root(struct hugetlb_cgroup
*h_cg
)
67 return (h_cg
== root_h_cgroup
);
70 static inline struct hugetlb_cgroup
*
71 parent_hugetlb_cgroup(struct hugetlb_cgroup
*h_cg
)
73 return hugetlb_cgroup_from_css(h_cg
->css
.parent
);
76 static inline bool hugetlb_cgroup_have_usage(struct hugetlb_cgroup
*h_cg
)
81 if (page_counter_read(
82 hugetlb_cgroup_counter_from_cgroup(h_cg
, hstate_index(h
))))
88 static void hugetlb_cgroup_init(struct hugetlb_cgroup
*h_cgroup
,
89 struct hugetlb_cgroup
*parent_h_cgroup
)
93 for (idx
= 0; idx
< HUGE_MAX_HSTATE
; idx
++) {
94 struct page_counter
*fault_parent
= NULL
;
95 struct page_counter
*rsvd_parent
= NULL
;
99 if (parent_h_cgroup
) {
100 fault_parent
= hugetlb_cgroup_counter_from_cgroup(
101 parent_h_cgroup
, idx
);
102 rsvd_parent
= hugetlb_cgroup_counter_from_cgroup_rsvd(
103 parent_h_cgroup
, idx
);
105 page_counter_init(hugetlb_cgroup_counter_from_cgroup(h_cgroup
,
109 hugetlb_cgroup_counter_from_cgroup_rsvd(h_cgroup
, idx
),
112 limit
= round_down(PAGE_COUNTER_MAX
,
113 pages_per_huge_page(&hstates
[idx
]));
115 ret
= page_counter_set_max(
116 hugetlb_cgroup_counter_from_cgroup(h_cgroup
, idx
),
119 ret
= page_counter_set_max(
120 hugetlb_cgroup_counter_from_cgroup_rsvd(h_cgroup
, idx
),
126 static void hugetlb_cgroup_free(struct hugetlb_cgroup
*h_cgroup
)
131 kfree(h_cgroup
->nodeinfo
[node
]);
135 static struct cgroup_subsys_state
*
136 hugetlb_cgroup_css_alloc(struct cgroup_subsys_state
*parent_css
)
138 struct hugetlb_cgroup
*parent_h_cgroup
= hugetlb_cgroup_from_css(parent_css
);
139 struct hugetlb_cgroup
*h_cgroup
;
142 h_cgroup
= kzalloc(struct_size(h_cgroup
, nodeinfo
, nr_node_ids
),
146 return ERR_PTR(-ENOMEM
);
148 if (!parent_h_cgroup
)
149 root_h_cgroup
= h_cgroup
;
152 * TODO: this routine can waste much memory for nodes which will
153 * never be onlined. It's better to use memory hotplug callback
156 for_each_node(node
) {
157 /* Set node_to_alloc to NUMA_NO_NODE for offline nodes. */
159 node_state(node
, N_NORMAL_MEMORY
) ? node
: NUMA_NO_NODE
;
160 h_cgroup
->nodeinfo
[node
] =
161 kzalloc_node(sizeof(struct hugetlb_cgroup_per_node
),
162 GFP_KERNEL
, node_to_alloc
);
163 if (!h_cgroup
->nodeinfo
[node
])
164 goto fail_alloc_nodeinfo
;
167 hugetlb_cgroup_init(h_cgroup
, parent_h_cgroup
);
168 return &h_cgroup
->css
;
171 hugetlb_cgroup_free(h_cgroup
);
172 return ERR_PTR(-ENOMEM
);
175 static void hugetlb_cgroup_css_free(struct cgroup_subsys_state
*css
)
177 hugetlb_cgroup_free(hugetlb_cgroup_from_css(css
));
181 * Should be called with hugetlb_lock held.
182 * Since we are holding hugetlb_lock, pages cannot get moved from
183 * active list or uncharged from the cgroup, So no need to get
184 * page reference and test for page active here. This function
187 static void hugetlb_cgroup_move_parent(int idx
, struct hugetlb_cgroup
*h_cg
,
190 unsigned int nr_pages
;
191 struct page_counter
*counter
;
192 struct hugetlb_cgroup
*page_hcg
;
193 struct hugetlb_cgroup
*parent
= parent_hugetlb_cgroup(h_cg
);
194 struct folio
*folio
= page_folio(page
);
196 page_hcg
= hugetlb_cgroup_from_folio(folio
);
198 * We can have pages in active list without any cgroup
199 * ie, hugepage with less than 3 pages. We can safely
200 * ignore those pages.
202 if (!page_hcg
|| page_hcg
!= h_cg
)
205 nr_pages
= compound_nr(page
);
207 parent
= root_h_cgroup
;
208 /* root has no limit */
209 page_counter_charge(&parent
->hugepage
[idx
], nr_pages
);
211 counter
= &h_cg
->hugepage
[idx
];
212 /* Take the pages off the local counter */
213 page_counter_cancel(counter
, nr_pages
);
215 set_hugetlb_cgroup(folio
, parent
);
221 * Force the hugetlb cgroup to empty the hugetlb resources by moving them to
224 static void hugetlb_cgroup_css_offline(struct cgroup_subsys_state
*css
)
226 struct hugetlb_cgroup
*h_cg
= hugetlb_cgroup_from_css(css
);
232 spin_lock_irq(&hugetlb_lock
);
233 list_for_each_entry(page
, &h
->hugepage_activelist
, lru
)
234 hugetlb_cgroup_move_parent(hstate_index(h
), h_cg
, page
);
236 spin_unlock_irq(&hugetlb_lock
);
239 } while (hugetlb_cgroup_have_usage(h_cg
));
242 static inline void hugetlb_event(struct hugetlb_cgroup
*hugetlb
, int idx
,
243 enum hugetlb_memory_event event
)
245 atomic_long_inc(&hugetlb
->events_local
[idx
][event
]);
246 cgroup_file_notify(&hugetlb
->events_local_file
[idx
]);
249 atomic_long_inc(&hugetlb
->events
[idx
][event
]);
250 cgroup_file_notify(&hugetlb
->events_file
[idx
]);
251 } while ((hugetlb
= parent_hugetlb_cgroup(hugetlb
)) &&
252 !hugetlb_cgroup_is_root(hugetlb
));
255 static int __hugetlb_cgroup_charge_cgroup(int idx
, unsigned long nr_pages
,
256 struct hugetlb_cgroup
**ptr
,
260 struct page_counter
*counter
;
261 struct hugetlb_cgroup
*h_cg
= NULL
;
263 if (hugetlb_cgroup_disabled())
267 h_cg
= hugetlb_cgroup_from_task(current
);
268 if (!css_tryget(&h_cg
->css
)) {
274 if (!page_counter_try_charge(
275 __hugetlb_cgroup_counter_from_cgroup(h_cg
, idx
, rsvd
),
276 nr_pages
, &counter
)) {
278 hugetlb_event(h_cg
, idx
, HUGETLB_MAX
);
282 /* Reservations take a reference to the css because they do not get
292 int hugetlb_cgroup_charge_cgroup(int idx
, unsigned long nr_pages
,
293 struct hugetlb_cgroup
**ptr
)
295 return __hugetlb_cgroup_charge_cgroup(idx
, nr_pages
, ptr
, false);
298 int hugetlb_cgroup_charge_cgroup_rsvd(int idx
, unsigned long nr_pages
,
299 struct hugetlb_cgroup
**ptr
)
301 return __hugetlb_cgroup_charge_cgroup(idx
, nr_pages
, ptr
, true);
304 /* Should be called with hugetlb_lock held */
305 static void __hugetlb_cgroup_commit_charge(int idx
, unsigned long nr_pages
,
306 struct hugetlb_cgroup
*h_cg
,
307 struct folio
*folio
, bool rsvd
)
309 if (hugetlb_cgroup_disabled() || !h_cg
)
311 lockdep_assert_held(&hugetlb_lock
);
312 __set_hugetlb_cgroup(folio
, h_cg
, rsvd
);
314 unsigned long usage
=
315 h_cg
->nodeinfo
[folio_nid(folio
)]->usage
[idx
];
317 * This write is not atomic due to fetching usage and writing
318 * to it, but that's fine because we call this with
319 * hugetlb_lock held anyway.
321 WRITE_ONCE(h_cg
->nodeinfo
[folio_nid(folio
)]->usage
[idx
],
326 void hugetlb_cgroup_commit_charge(int idx
, unsigned long nr_pages
,
327 struct hugetlb_cgroup
*h_cg
,
330 __hugetlb_cgroup_commit_charge(idx
, nr_pages
, h_cg
, folio
, false);
333 void hugetlb_cgroup_commit_charge_rsvd(int idx
, unsigned long nr_pages
,
334 struct hugetlb_cgroup
*h_cg
,
337 __hugetlb_cgroup_commit_charge(idx
, nr_pages
, h_cg
, folio
, true);
341 * Should be called with hugetlb_lock held
343 static void __hugetlb_cgroup_uncharge_folio(int idx
, unsigned long nr_pages
,
344 struct folio
*folio
, bool rsvd
)
346 struct hugetlb_cgroup
*h_cg
;
348 if (hugetlb_cgroup_disabled())
350 lockdep_assert_held(&hugetlb_lock
);
351 h_cg
= __hugetlb_cgroup_from_folio(folio
, rsvd
);
354 __set_hugetlb_cgroup(folio
, NULL
, rsvd
);
356 page_counter_uncharge(__hugetlb_cgroup_counter_from_cgroup(h_cg
, idx
,
363 unsigned long usage
=
364 h_cg
->nodeinfo
[folio_nid(folio
)]->usage
[idx
];
366 * This write is not atomic due to fetching usage and writing
367 * to it, but that's fine because we call this with
368 * hugetlb_lock held anyway.
370 WRITE_ONCE(h_cg
->nodeinfo
[folio_nid(folio
)]->usage
[idx
],
375 void hugetlb_cgroup_uncharge_folio(int idx
, unsigned long nr_pages
,
378 __hugetlb_cgroup_uncharge_folio(idx
, nr_pages
, folio
, false);
381 void hugetlb_cgroup_uncharge_folio_rsvd(int idx
, unsigned long nr_pages
,
384 __hugetlb_cgroup_uncharge_folio(idx
, nr_pages
, folio
, true);
387 static void __hugetlb_cgroup_uncharge_cgroup(int idx
, unsigned long nr_pages
,
388 struct hugetlb_cgroup
*h_cg
,
391 if (hugetlb_cgroup_disabled() || !h_cg
)
394 page_counter_uncharge(__hugetlb_cgroup_counter_from_cgroup(h_cg
, idx
,
402 void hugetlb_cgroup_uncharge_cgroup(int idx
, unsigned long nr_pages
,
403 struct hugetlb_cgroup
*h_cg
)
405 __hugetlb_cgroup_uncharge_cgroup(idx
, nr_pages
, h_cg
, false);
408 void hugetlb_cgroup_uncharge_cgroup_rsvd(int idx
, unsigned long nr_pages
,
409 struct hugetlb_cgroup
*h_cg
)
411 __hugetlb_cgroup_uncharge_cgroup(idx
, nr_pages
, h_cg
, true);
414 void hugetlb_cgroup_uncharge_counter(struct resv_map
*resv
, unsigned long start
,
417 if (hugetlb_cgroup_disabled() || !resv
|| !resv
->reservation_counter
||
421 page_counter_uncharge(resv
->reservation_counter
,
422 (end
- start
) * resv
->pages_per_hpage
);
426 void hugetlb_cgroup_uncharge_file_region(struct resv_map
*resv
,
427 struct file_region
*rg
,
428 unsigned long nr_pages
,
431 if (hugetlb_cgroup_disabled() || !resv
|| !rg
|| !nr_pages
)
434 if (rg
->reservation_counter
&& resv
->pages_per_hpage
&&
435 !resv
->reservation_counter
) {
436 page_counter_uncharge(rg
->reservation_counter
,
437 nr_pages
* resv
->pages_per_hpage
);
439 * Only do css_put(rg->css) when we delete the entire region
440 * because one file_region must hold exactly one css reference.
458 static int hugetlb_cgroup_read_numa_stat(struct seq_file
*seq
, void *dummy
)
461 struct cftype
*cft
= seq_cft(seq
);
462 int idx
= MEMFILE_IDX(cft
->private);
463 bool legacy
= MEMFILE_ATTR(cft
->private);
464 struct hugetlb_cgroup
*h_cg
= hugetlb_cgroup_from_css(seq_css(seq
));
465 struct cgroup_subsys_state
*css
;
469 /* Add up usage across all nodes for the non-hierarchical total. */
471 for_each_node_state(nid
, N_MEMORY
)
472 usage
+= READ_ONCE(h_cg
->nodeinfo
[nid
]->usage
[idx
]);
473 seq_printf(seq
, "total=%lu", usage
* PAGE_SIZE
);
475 /* Simply print the per-node usage for the non-hierarchical total. */
476 for_each_node_state(nid
, N_MEMORY
)
477 seq_printf(seq
, " N%d=%lu", nid
,
478 READ_ONCE(h_cg
->nodeinfo
[nid
]->usage
[idx
]) *
484 * The hierarchical total is pretty much the value recorded by the
485 * counter, so use that.
487 seq_printf(seq
, "%stotal=%lu", legacy
? "hierarchical_" : "",
488 page_counter_read(&h_cg
->hugepage
[idx
]) * PAGE_SIZE
);
491 * For each node, transverse the css tree to obtain the hierarchical
494 for_each_node_state(nid
, N_MEMORY
) {
497 css_for_each_descendant_pre(css
, &h_cg
->css
) {
498 usage
+= READ_ONCE(hugetlb_cgroup_from_css(css
)
503 seq_printf(seq
, " N%d=%lu", nid
, usage
* PAGE_SIZE
);
511 static u64
hugetlb_cgroup_read_u64(struct cgroup_subsys_state
*css
,
514 struct page_counter
*counter
;
515 struct page_counter
*rsvd_counter
;
516 struct hugetlb_cgroup
*h_cg
= hugetlb_cgroup_from_css(css
);
518 counter
= &h_cg
->hugepage
[MEMFILE_IDX(cft
->private)];
519 rsvd_counter
= &h_cg
->rsvd_hugepage
[MEMFILE_IDX(cft
->private)];
521 switch (MEMFILE_ATTR(cft
->private)) {
523 return (u64
)page_counter_read(counter
) * PAGE_SIZE
;
525 return (u64
)page_counter_read(rsvd_counter
) * PAGE_SIZE
;
527 return (u64
)counter
->max
* PAGE_SIZE
;
529 return (u64
)rsvd_counter
->max
* PAGE_SIZE
;
531 return (u64
)counter
->watermark
* PAGE_SIZE
;
532 case RES_RSVD_MAX_USAGE
:
533 return (u64
)rsvd_counter
->watermark
* PAGE_SIZE
;
535 return counter
->failcnt
;
536 case RES_RSVD_FAILCNT
:
537 return rsvd_counter
->failcnt
;
543 static int hugetlb_cgroup_read_u64_max(struct seq_file
*seq
, void *v
)
547 struct cftype
*cft
= seq_cft(seq
);
549 struct page_counter
*counter
;
550 struct hugetlb_cgroup
*h_cg
= hugetlb_cgroup_from_css(seq_css(seq
));
552 idx
= MEMFILE_IDX(cft
->private);
553 counter
= &h_cg
->hugepage
[idx
];
555 limit
= round_down(PAGE_COUNTER_MAX
,
556 pages_per_huge_page(&hstates
[idx
]));
558 switch (MEMFILE_ATTR(cft
->private)) {
560 counter
= &h_cg
->rsvd_hugepage
[idx
];
563 val
= (u64
)page_counter_read(counter
);
564 seq_printf(seq
, "%llu\n", val
* PAGE_SIZE
);
567 counter
= &h_cg
->rsvd_hugepage
[idx
];
570 val
= (u64
)counter
->max
;
572 seq_puts(seq
, "max\n");
574 seq_printf(seq
, "%llu\n", val
* PAGE_SIZE
);
583 static DEFINE_MUTEX(hugetlb_limit_mutex
);
585 static ssize_t
hugetlb_cgroup_write(struct kernfs_open_file
*of
,
586 char *buf
, size_t nbytes
, loff_t off
,
590 unsigned long nr_pages
;
591 struct hugetlb_cgroup
*h_cg
= hugetlb_cgroup_from_css(of_css(of
));
594 if (hugetlb_cgroup_is_root(h_cg
)) /* Can't set limit on root */
598 ret
= page_counter_memparse(buf
, max
, &nr_pages
);
602 idx
= MEMFILE_IDX(of_cft(of
)->private);
603 nr_pages
= round_down(nr_pages
, pages_per_huge_page(&hstates
[idx
]));
605 switch (MEMFILE_ATTR(of_cft(of
)->private)) {
610 mutex_lock(&hugetlb_limit_mutex
);
611 ret
= page_counter_set_max(
612 __hugetlb_cgroup_counter_from_cgroup(h_cg
, idx
, rsvd
),
614 mutex_unlock(&hugetlb_limit_mutex
);
620 return ret
?: nbytes
;
623 static ssize_t
hugetlb_cgroup_write_legacy(struct kernfs_open_file
*of
,
624 char *buf
, size_t nbytes
, loff_t off
)
626 return hugetlb_cgroup_write(of
, buf
, nbytes
, off
, "-1");
629 static ssize_t
hugetlb_cgroup_write_dfl(struct kernfs_open_file
*of
,
630 char *buf
, size_t nbytes
, loff_t off
)
632 return hugetlb_cgroup_write(of
, buf
, nbytes
, off
, "max");
635 static ssize_t
hugetlb_cgroup_reset(struct kernfs_open_file
*of
,
636 char *buf
, size_t nbytes
, loff_t off
)
639 struct page_counter
*counter
, *rsvd_counter
;
640 struct hugetlb_cgroup
*h_cg
= hugetlb_cgroup_from_css(of_css(of
));
642 counter
= &h_cg
->hugepage
[MEMFILE_IDX(of_cft(of
)->private)];
643 rsvd_counter
= &h_cg
->rsvd_hugepage
[MEMFILE_IDX(of_cft(of
)->private)];
645 switch (MEMFILE_ATTR(of_cft(of
)->private)) {
647 page_counter_reset_watermark(counter
);
649 case RES_RSVD_MAX_USAGE
:
650 page_counter_reset_watermark(rsvd_counter
);
653 counter
->failcnt
= 0;
655 case RES_RSVD_FAILCNT
:
656 rsvd_counter
->failcnt
= 0;
662 return ret
?: nbytes
;
665 static char *mem_fmt(char *buf
, int size
, unsigned long hsize
)
668 snprintf(buf
, size
, "%luGB", hsize
/ SZ_1G
);
669 else if (hsize
>= SZ_1M
)
670 snprintf(buf
, size
, "%luMB", hsize
/ SZ_1M
);
672 snprintf(buf
, size
, "%luKB", hsize
/ SZ_1K
);
676 static int __hugetlb_events_show(struct seq_file
*seq
, bool local
)
680 struct cftype
*cft
= seq_cft(seq
);
681 struct hugetlb_cgroup
*h_cg
= hugetlb_cgroup_from_css(seq_css(seq
));
683 idx
= MEMFILE_IDX(cft
->private);
686 max
= atomic_long_read(&h_cg
->events_local
[idx
][HUGETLB_MAX
]);
688 max
= atomic_long_read(&h_cg
->events
[idx
][HUGETLB_MAX
]);
690 seq_printf(seq
, "max %lu\n", max
);
695 static int hugetlb_events_show(struct seq_file
*seq
, void *v
)
697 return __hugetlb_events_show(seq
, false);
700 static int hugetlb_events_local_show(struct seq_file
*seq
, void *v
)
702 return __hugetlb_events_show(seq
, true);
705 static void __init
__hugetlb_cgroup_file_dfl_init(int idx
)
709 struct hstate
*h
= &hstates
[idx
];
711 /* format the size */
712 mem_fmt(buf
, sizeof(buf
), huge_page_size(h
));
714 /* Add the limit file */
715 cft
= &h
->cgroup_files_dfl
[0];
716 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.max", buf
);
717 cft
->private = MEMFILE_PRIVATE(idx
, RES_LIMIT
);
718 cft
->seq_show
= hugetlb_cgroup_read_u64_max
;
719 cft
->write
= hugetlb_cgroup_write_dfl
;
720 cft
->flags
= CFTYPE_NOT_ON_ROOT
;
722 /* Add the reservation limit file */
723 cft
= &h
->cgroup_files_dfl
[1];
724 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.rsvd.max", buf
);
725 cft
->private = MEMFILE_PRIVATE(idx
, RES_RSVD_LIMIT
);
726 cft
->seq_show
= hugetlb_cgroup_read_u64_max
;
727 cft
->write
= hugetlb_cgroup_write_dfl
;
728 cft
->flags
= CFTYPE_NOT_ON_ROOT
;
730 /* Add the current usage file */
731 cft
= &h
->cgroup_files_dfl
[2];
732 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.current", buf
);
733 cft
->private = MEMFILE_PRIVATE(idx
, RES_USAGE
);
734 cft
->seq_show
= hugetlb_cgroup_read_u64_max
;
735 cft
->flags
= CFTYPE_NOT_ON_ROOT
;
737 /* Add the current reservation usage file */
738 cft
= &h
->cgroup_files_dfl
[3];
739 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.rsvd.current", buf
);
740 cft
->private = MEMFILE_PRIVATE(idx
, RES_RSVD_USAGE
);
741 cft
->seq_show
= hugetlb_cgroup_read_u64_max
;
742 cft
->flags
= CFTYPE_NOT_ON_ROOT
;
744 /* Add the events file */
745 cft
= &h
->cgroup_files_dfl
[4];
746 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.events", buf
);
747 cft
->private = MEMFILE_PRIVATE(idx
, 0);
748 cft
->seq_show
= hugetlb_events_show
;
749 cft
->file_offset
= offsetof(struct hugetlb_cgroup
, events_file
[idx
]);
750 cft
->flags
= CFTYPE_NOT_ON_ROOT
;
752 /* Add the events.local file */
753 cft
= &h
->cgroup_files_dfl
[5];
754 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.events.local", buf
);
755 cft
->private = MEMFILE_PRIVATE(idx
, 0);
756 cft
->seq_show
= hugetlb_events_local_show
;
757 cft
->file_offset
= offsetof(struct hugetlb_cgroup
,
758 events_local_file
[idx
]);
759 cft
->flags
= CFTYPE_NOT_ON_ROOT
;
761 /* Add the numa stat file */
762 cft
= &h
->cgroup_files_dfl
[6];
763 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.numa_stat", buf
);
764 cft
->private = MEMFILE_PRIVATE(idx
, 0);
765 cft
->seq_show
= hugetlb_cgroup_read_numa_stat
;
766 cft
->flags
= CFTYPE_NOT_ON_ROOT
;
768 /* NULL terminate the last cft */
769 cft
= &h
->cgroup_files_dfl
[7];
770 memset(cft
, 0, sizeof(*cft
));
772 WARN_ON(cgroup_add_dfl_cftypes(&hugetlb_cgrp_subsys
,
773 h
->cgroup_files_dfl
));
776 static void __init
__hugetlb_cgroup_file_legacy_init(int idx
)
780 struct hstate
*h
= &hstates
[idx
];
782 /* format the size */
783 mem_fmt(buf
, sizeof(buf
), huge_page_size(h
));
785 /* Add the limit file */
786 cft
= &h
->cgroup_files_legacy
[0];
787 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.limit_in_bytes", buf
);
788 cft
->private = MEMFILE_PRIVATE(idx
, RES_LIMIT
);
789 cft
->read_u64
= hugetlb_cgroup_read_u64
;
790 cft
->write
= hugetlb_cgroup_write_legacy
;
792 /* Add the reservation limit file */
793 cft
= &h
->cgroup_files_legacy
[1];
794 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.rsvd.limit_in_bytes", buf
);
795 cft
->private = MEMFILE_PRIVATE(idx
, RES_RSVD_LIMIT
);
796 cft
->read_u64
= hugetlb_cgroup_read_u64
;
797 cft
->write
= hugetlb_cgroup_write_legacy
;
799 /* Add the usage file */
800 cft
= &h
->cgroup_files_legacy
[2];
801 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.usage_in_bytes", buf
);
802 cft
->private = MEMFILE_PRIVATE(idx
, RES_USAGE
);
803 cft
->read_u64
= hugetlb_cgroup_read_u64
;
805 /* Add the reservation usage file */
806 cft
= &h
->cgroup_files_legacy
[3];
807 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.rsvd.usage_in_bytes", buf
);
808 cft
->private = MEMFILE_PRIVATE(idx
, RES_RSVD_USAGE
);
809 cft
->read_u64
= hugetlb_cgroup_read_u64
;
811 /* Add the MAX usage file */
812 cft
= &h
->cgroup_files_legacy
[4];
813 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.max_usage_in_bytes", buf
);
814 cft
->private = MEMFILE_PRIVATE(idx
, RES_MAX_USAGE
);
815 cft
->write
= hugetlb_cgroup_reset
;
816 cft
->read_u64
= hugetlb_cgroup_read_u64
;
818 /* Add the MAX reservation usage file */
819 cft
= &h
->cgroup_files_legacy
[5];
820 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.rsvd.max_usage_in_bytes", buf
);
821 cft
->private = MEMFILE_PRIVATE(idx
, RES_RSVD_MAX_USAGE
);
822 cft
->write
= hugetlb_cgroup_reset
;
823 cft
->read_u64
= hugetlb_cgroup_read_u64
;
825 /* Add the failcntfile */
826 cft
= &h
->cgroup_files_legacy
[6];
827 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.failcnt", buf
);
828 cft
->private = MEMFILE_PRIVATE(idx
, RES_FAILCNT
);
829 cft
->write
= hugetlb_cgroup_reset
;
830 cft
->read_u64
= hugetlb_cgroup_read_u64
;
832 /* Add the reservation failcntfile */
833 cft
= &h
->cgroup_files_legacy
[7];
834 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.rsvd.failcnt", buf
);
835 cft
->private = MEMFILE_PRIVATE(idx
, RES_RSVD_FAILCNT
);
836 cft
->write
= hugetlb_cgroup_reset
;
837 cft
->read_u64
= hugetlb_cgroup_read_u64
;
839 /* Add the numa stat file */
840 cft
= &h
->cgroup_files_legacy
[8];
841 snprintf(cft
->name
, MAX_CFTYPE_NAME
, "%s.numa_stat", buf
);
842 cft
->private = MEMFILE_PRIVATE(idx
, 1);
843 cft
->seq_show
= hugetlb_cgroup_read_numa_stat
;
845 /* NULL terminate the last cft */
846 cft
= &h
->cgroup_files_legacy
[9];
847 memset(cft
, 0, sizeof(*cft
));
849 WARN_ON(cgroup_add_legacy_cftypes(&hugetlb_cgrp_subsys
,
850 h
->cgroup_files_legacy
));
853 static void __init
__hugetlb_cgroup_file_init(int idx
)
855 __hugetlb_cgroup_file_dfl_init(idx
);
856 __hugetlb_cgroup_file_legacy_init(idx
);
859 void __init
hugetlb_cgroup_file_init(void)
864 __hugetlb_cgroup_file_init(hstate_index(h
));
868 * hugetlb_lock will make sure a parallel cgroup rmdir won't happen
869 * when we migrate hugepages
871 void hugetlb_cgroup_migrate(struct folio
*old_folio
, struct folio
*new_folio
)
873 struct hugetlb_cgroup
*h_cg
;
874 struct hugetlb_cgroup
*h_cg_rsvd
;
875 struct hstate
*h
= folio_hstate(old_folio
);
877 if (hugetlb_cgroup_disabled())
880 spin_lock_irq(&hugetlb_lock
);
881 h_cg
= hugetlb_cgroup_from_folio(old_folio
);
882 h_cg_rsvd
= hugetlb_cgroup_from_folio_rsvd(old_folio
);
883 set_hugetlb_cgroup(old_folio
, NULL
);
884 set_hugetlb_cgroup_rsvd(old_folio
, NULL
);
886 /* move the h_cg details to new cgroup */
887 set_hugetlb_cgroup(new_folio
, h_cg
);
888 set_hugetlb_cgroup_rsvd(new_folio
, h_cg_rsvd
);
889 list_move(&new_folio
->lru
, &h
->hugepage_activelist
);
890 spin_unlock_irq(&hugetlb_lock
);
894 static struct cftype hugetlb_files
[] = {
898 struct cgroup_subsys hugetlb_cgrp_subsys
= {
899 .css_alloc
= hugetlb_cgroup_css_alloc
,
900 .css_offline
= hugetlb_cgroup_css_offline
,
901 .css_free
= hugetlb_cgroup_css_free
,
902 .dfl_cftypes
= hugetlb_files
,
903 .legacy_cftypes
= hugetlb_files
,