Pull cgroup updates from Tejun Heo: - Add deprecation info messages to cgroup1-only features - rstat updates including a bug fix and breaking up a critical section to reduce interrupt latency impact - Other misc and doc updates * tag 'cgroup-for-6.15' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup: cgroup: rstat: Cleanup flushing functions and locking cgroup/rstat: avoid disabling irqs for O(num_cpu) mm: Fix a build breakage in memcontrol-v1.c blk-cgroup: Simplify policy files registration cgroup: Update file naming comment cgroup: Add deprecation message to legacy freezer controller mm: Add transformation message for per-memcg swappiness RFC cgroup/cpuset-v1: Add deprecation messages to sched_relax_domain_level cgroup/cpuset-v1: Add deprecation messages to memory_migrate cgroup/cpuset-v1: Add deprecation messages to mem_exclusive and mem_hardwall cgroup: Print message when /proc/cgroups is read on v2-only system cgroup/blkio: Add deprecation messages to reset_stats cgroup/cpuset-v1: Add deprecation messages to memory_spread_page and memory_spread_slab cgroup/cpuset-v1: Add deprecation messages to sched_load_balance and memory_pressure_enabled cgroup, docs: Be explicit about independence of RT_GROUP_SCHED and non-cpu controllers cgroup/rstat: Fix forceidle time in cpu.stat cgroup/misc: Remove unused misc_cg_res_total_usage cgroup/cpuset: Move procfs cpuset attribute under cgroup-v1.c cgroup: update comment about dropping cgroup kn refs
Unified · Split
| @@ -125,3 +125,7 @@ to unfreeze all tasks in the container:: · 124 unmodified lines | ||
| 125 | 125 | |
| 126 | 126 | This is the basic mechanism which should do the right thing for user space task |
| 127 | 127 | in a simple scenario. |
| 128 | + | |
| 129 | +This freezer implementation is affected by shortcomings (see commit | |
| 130 | +76f969e8948d8 ("cgroup: cgroup v2 freezer")) and cgroup v2 freezer is | |
| 131 | +recommended. | |
| @@ -90,6 +90,7 @@ Brief summary of control files. · 89 unmodified lines | ||
| 90 | 90 | used. |
| 91 | 91 | memory.swappiness set/show swappiness parameter of vmscan |
| 92 | 92 | (See sysctl's vm.swappiness) |
| 93 | + Per memcg knob does not exist in cgroup v2. | |
| 93 | 94 | memory.move_charge_at_immigrate This knob is deprecated. |
| 94 | 95 | memory.oom_control set/show oom controls. |
| 95 | 96 | This knob is deprecated and shouldn't be |
| @@ -1076,15 +1076,20 @@ cpufreq governor about the minimum desired frequency which should always be · 1075 unmodified lines | ||
| 1076 | 1076 | provided by a CPU, as well as the maximum desired frequency, which should not |
| 1077 | 1077 | be exceeded by a CPU. |
| 1078 | 1078 | |
| 1079 | −WARNING: cgroup2 doesn't yet support control of realtime processes. For | |
| 1080 | −a kernel built with the CONFIG_RT_GROUP_SCHED option enabled for group | |
| 1081 | −scheduling of realtime processes, the cpu controller can only be enabled | |
| 1082 | −when all RT processes are in the root cgroup. This limitation does | |
| 1083 | −not apply if CONFIG_RT_GROUP_SCHED is disabled. Be aware that system | |
| 1084 | −management software may already have placed RT processes into nonroot | |
| 1085 | −cgroups during the system boot process, and these processes may need | |
| 1086 | −to be moved to the root cgroup before the cpu controller can be enabled | |
| 1087 | −with a CONFIG_RT_GROUP_SCHED enabled kernel. | |
| 1079 | +WARNING: cgroup2 cpu controller doesn't yet fully support the control of | |
| 1080 | +realtime processes. For a kernel built with the CONFIG_RT_GROUP_SCHED option | |
| 1081 | +enabled for group scheduling of realtime processes, the cpu controller can only | |
| 1082 | +be enabled when all RT processes are in the root cgroup. Be aware that system | |
| 1083 | +management software may already have placed RT processes into non-root cgroups | |
| 1084 | +during the system boot process, and these processes may need to be moved to the | |
| 1085 | +root cgroup before the cpu controller can be enabled with a | |
| 1086 | +CONFIG_RT_GROUP_SCHED enabled kernel. | |
| 1087 | + | |
| 1088 | +With CONFIG_RT_GROUP_SCHED disabled, this limitation does not apply and some of | |
| 1089 | +the interface files either affect realtime processes or account for them. See | |
| 1090 | +the following section for details. Only the cpu controller is affected by | |
| 1091 | +CONFIG_RT_GROUP_SCHED. Other controllers can be used for the resource control of | |
| 1092 | +realtime processes irrespective of CONFIG_RT_GROUP_SCHED. | |
| 1088 | 1093 | |
| 1089 | 1094 | |
| 1090 | 1095 | CPU Interface Files |
| @@ -659,6 +659,7 @@ static int blkcg_reset_stats(struct cgroup_subsys_state *css, · 658 unmodified lines | ||
| 659 | 659 | struct blkcg_gq *blkg; |
| 660 | 660 | int i; |
| 661 | 661 | |
| 662 | + pr_info_once("blkio.%s is deprecated\n", cftype->name); | |
| 662 | 663 | mutex_lock(&blkcg_pol_mutex); |
| 663 | 664 | spin_lock_irq(&blkcg->lock); |
| 664 | 665 | |
| @@ -1770,12 +1771,15 @@ int blkcg_policy_register(struct blkcg_policy *pol) · 1105 unmodified lines | ||
| 1770 | 1771 | mutex_unlock(&blkcg_pol_mutex); |
| 1771 | 1772 | |
| 1772 | 1773 | /* everything is in place, add intf files for the new policy */ |
| 1773 | − if (pol->dfl_cftypes) | |
| 1774 | + if (pol->dfl_cftypes == pol->legacy_cftypes) { | |
| 1775 | + WARN_ON(cgroup_add_cftypes(&io_cgrp_subsys, | |
| 1776 | + pol->dfl_cftypes)); | |
| 1777 | + } else { | |
| 1774 | 1778 | WARN_ON(cgroup_add_dfl_cftypes(&io_cgrp_subsys, |
| 1775 | 1779 | pol->dfl_cftypes)); |
| 1776 | − if (pol->legacy_cftypes) | |
| 1777 | 1780 | WARN_ON(cgroup_add_legacy_cftypes(&io_cgrp_subsys, |
| 1778 | 1781 | pol->legacy_cftypes)); |
| 1782 | + } | |
| 1779 | 1783 | mutex_unlock(&blkcg_pol_register_mutex); |
| 1780 | 1784 | return 0; |
| 1781 | 1785 | |
| @@ -113,27 +113,18 @@ static void ioprio_free_cpd(struct blkcg_policy_data *cpd) · 112 unmodified lines | ||
| 113 | 113 | kfree(blkcg); |
| 114 | 114 | } |
| 115 | 115 | |
| 116 | −#define IOPRIO_ATTRS \ | |
| 117 | − { \ | |
| 118 | − .name = "prio.class", \ | |
| 119 | − .seq_show = ioprio_show_prio_policy, \ | |
| 120 | − .write = ioprio_set_prio_policy, \ | |
| 121 | − }, \ | |
| 122 | − { } /* sentinel */ | |
| 123 | − | |
| 124 | −/* cgroup v2 attributes */ | |
| 125 | 116 | static struct cftype ioprio_files[] = { |
| 126 | − IOPRIO_ATTRS | |
| 127 | −}; | |
| 128 | − | |
| 129 | −/* cgroup v1 attributes */ | |
| 130 | −static struct cftype ioprio_legacy_files[] = { | |
| 131 | − IOPRIO_ATTRS | |
| 117 | + { | |
| 118 | + .name = "prio.class", | |
| 119 | + .seq_show = ioprio_show_prio_policy, | |
| 120 | + .write = ioprio_set_prio_policy, | |
| 121 | + }, | |
| 122 | + { } /* sentinel */ | |
| 132 | 123 | }; |
| 133 | 124 | |
| 134 | 125 | static struct blkcg_policy ioprio_policy = { |
| 135 | 126 | .dfl_cftypes = ioprio_files, |
| 136 | − .legacy_cftypes = ioprio_legacy_files, | |
| 127 | + .legacy_cftypes = ioprio_files, | |
| 137 | 128 | |
| 138 | 129 | .cpd_alloc_fn = ioprio_alloc_cpd, |
| 139 | 130 | .cpd_free_fn = ioprio_free_cpd, |
| @@ -619,9 +619,8 @@ struct cgroup_root { · 618 unmodified lines | ||
| 619 | 619 | */ |
| 620 | 620 | struct cftype { |
| 621 | 621 | /* |
| 622 | − * By convention, the name should begin with the name of the | |
| 623 | − * subsystem, followed by a period. Zero length string indicates | |
| 624 | − * end of cftype array. | |
| 622 | + * Name of the subsystem is prepended in cgroup_file_name(). | |
| 623 | + * Zero length string indicates end of cftype array. | |
| 625 | 624 | */ |
| 626 | 625 | char name[MAX_CFTYPE_NAME]; |
| 627 | 626 | unsigned long private; |
| @@ -113,6 +113,7 @@ int cgroup_transfer_tasks(struct cgroup *to, struct cgroup *from); · 112 unmodified lines | ||
| 113 | 113 | |
| 114 | 114 | int cgroup_add_dfl_cftypes(struct cgroup_subsys *ss, struct cftype *cfts); |
| 115 | 115 | int cgroup_add_legacy_cftypes(struct cgroup_subsys *ss, struct cftype *cfts); |
| 116 | +int cgroup_add_cftypes(struct cgroup_subsys *ss, struct cftype *cfts); | |
| 116 | 117 | int cgroup_rm_cftypes(struct cftype *cfts); |
| 117 | 118 | void cgroup_file_notify(struct cgroup_file *cfile); |
| 118 | 119 | void cgroup_file_show(struct cgroup_file *cfile, bool show); |
| @@ -689,8 +690,6 @@ static inline void cgroup_path_from_kernfs_id(u64 id, char *buf, size_t buflen) · 570 unmodified lines | ||
| 689 | 690 | */ |
| 690 | 691 | void cgroup_rstat_updated(struct cgroup *cgrp, int cpu); |
| 691 | 692 | void cgroup_rstat_flush(struct cgroup *cgrp); |
| 692 | −void cgroup_rstat_flush_hold(struct cgroup *cgrp); | |
| 693 | −void cgroup_rstat_flush_release(struct cgroup *cgrp); | |
| 694 | 693 | |
| 695 | 694 | /* |
| 696 | 695 | * Basic resource stats. |
| @@ -60,7 +60,6 @@ struct misc_cg { · 59 unmodified lines | ||
| 60 | 60 | struct misc_res res[MISC_CG_RES_TYPES]; |
| 61 | 61 | }; |
| 62 | 62 | |
| 63 | −u64 misc_cg_res_total_usage(enum misc_res_type type); | |
| 64 | 63 | int misc_cg_set_capacity(enum misc_res_type type, u64 capacity); |
| 65 | 64 | int misc_cg_try_charge(enum misc_res_type type, struct misc_cg *cg, u64 amount); |
| 66 | 65 | void misc_cg_uncharge(enum misc_res_type type, struct misc_cg *cg, u64 amount); |
| @@ -104,11 +103,6 @@ static inline void put_misc_cg(struct misc_cg *cg) · 37 unmodified lines | ||
| 104 | 103 | |
| 105 | 104 | #else /* !CONFIG_CGROUP_MISC */ |
| 106 | 105 | |
| 107 | −static inline u64 misc_cg_res_total_usage(enum misc_res_type type) | |
| 108 | −{ | |
| 109 | − return 0; | |
| 110 | −} | |
| 111 | − | |
| 112 | 106 | static inline int misc_cg_set_capacity(enum misc_res_type type, u64 capacity) |
| 113 | 107 | { |
| 114 | 108 | return 0; |
| @@ -1198,7 +1198,8 @@ config CPUSETS_V1 · 1197 unmodified lines | ||
| 1198 | 1198 | help |
| 1199 | 1199 | Legacy cgroup v1 cpusets controller which has been deprecated by |
| 1200 | 1200 | cgroup v2 implementation. The v1 is there for legacy applications |
| 1201 | − which haven't migrated to the new cgroup v2 interface yet. If you | |
| 1201 | + which haven't migrated to the new cgroup v2 interface yet. Legacy | |
| 1202 | + interface includes cpuset filesystem and /proc/<pid>/cpuset. If you | |
| 1202 | 1203 | do not have any such application then you are completely fine leaving |
| 1203 | 1204 | this option disabled. |
| 1204 | 1205 | |
| @@ -1206,7 +1207,7 @@ config CPUSETS_V1 · 1 unmodified line | ||
| 1206 | 1207 | |
| 1207 | 1208 | config PROC_PID_CPUSET |
| 1208 | 1209 | bool "Include legacy /proc/<pid>/cpuset file" |
| 1209 | − depends on CPUSETS | |
| 1210 | + depends on CPUSETS_V1 | |
| 1210 | 1211 | default y |
| 1211 | 1212 | |
| 1212 | 1213 | config CGROUP_DEVICE |
| @@ -168,6 +168,7 @@ struct cgroup_mgctx { · 167 unmodified lines | ||
| 168 | 168 | |
| 169 | 169 | extern struct cgroup_subsys *cgroup_subsys[]; |
| 170 | 170 | extern struct list_head cgroup_roots; |
| 171 | +extern bool cgrp_dfl_visible; | |
| 171 | 172 | |
| 172 | 173 | /* iterate across the hierarchies */ |
| 173 | 174 | #define for_each_root(root) \ |
| @@ -673,6 +673,7 @@ struct cftype cgroup1_base_files[] = { · 672 unmodified lines | ||
| 673 | 673 | int proc_cgroupstats_show(struct seq_file *m, void *v) |
| 674 | 674 | { |
| 675 | 675 | struct cgroup_subsys *ss; |
| 676 | + bool cgrp_v1_visible = false; | |
| 676 | 677 | int i; |
| 677 | 678 | |
| 678 | 679 | seq_puts(m, "#subsys_name\thierarchy\tnum_cgroups\tenabled\n"); |
| @@ -684,12 +685,18 @@ int proc_cgroupstats_show(struct seq_file *m, void *v) · 5 unmodified lines | ||
| 684 | 685 | for_each_subsys(ss, i) { |
| 685 | 686 | if (cgroup1_subsys_absent(ss)) |
| 686 | 687 | continue; |
| 688 | + cgrp_v1_visible |= ss->root != &cgrp_dfl_root; | |
| 689 | + | |
| 687 | 690 | seq_printf(m, "%s\t%d\t%d\t%d\n", |
| 688 | 691 | ss->legacy_name, ss->root->hierarchy_id, |
| 689 | 692 | atomic_read(&ss->root->nr_cgrps), |
| 690 | 693 | cgroup_ssid_enabled(i)); |
| 691 | 694 | } |
| 692 | 695 | |
| 696 | + if (cgrp_dfl_visible && !cgrp_v1_visible) | |
| 697 | + pr_info_once("/proc/cgroups lists only v1 controllers, use cgroup.controllers of root cgroup for v2 info\n"); | |
| 698 | + | |
| 699 | + | |
| 693 | 700 | return 0; |
| 694 | 701 | } |
| 695 | 702 | |
| @@ -171,7 +171,7 @@ EXPORT_SYMBOL_GPL(cgrp_dfl_root); · 170 unmodified lines | ||
| 171 | 171 | * The default hierarchy always exists but is hidden until mounted for the |
| 172 | 172 | * first time. This is for backward compatibility. |
| 173 | 173 | */ |
| 174 | −static bool cgrp_dfl_visible; | |
| 174 | +bool cgrp_dfl_visible; | |
| 175 | 175 | |
| 176 | 176 | /* some controllers are not supported in the default hierarchy */ |
| 177 | 177 | static u16 cgrp_dfl_inhibit_ss_mask; |
| @@ -4447,7 +4447,7 @@ int cgroup_rm_cftypes(struct cftype *cfts) · 4269 unmodified lines | ||
| 4447 | 4447 | * function currently returns 0 as long as @cfts registration is successful |
| 4448 | 4448 | * even if some file creation attempts on existing cgroups fail. |
| 4449 | 4449 | */ |
| 4450 | −static int cgroup_add_cftypes(struct cgroup_subsys *ss, struct cftype *cfts) | |
| 4450 | +int cgroup_add_cftypes(struct cgroup_subsys *ss, struct cftype *cfts) | |
| 4451 | 4451 | { |
| 4452 | 4452 | int ret; |
| 4453 | 4453 | |
| @@ -5831,7 +5831,7 @@ int cgroup_mkdir(struct kernfs_node *parent_kn, const char *name, umode_t mode) · 1377 unmodified lines | ||
| 5831 | 5831 | } |
| 5832 | 5832 | |
| 5833 | 5833 | /* |
| 5834 | − * This extra ref will be put in cgroup_free_fn() and guarantees | |
| 5834 | + * This extra ref will be put in css_free_rwork_fn() and guarantees | |
| 5835 | 5835 | * that @cgrp->kn is always accessible. |
| 5836 | 5836 | */ |
| 5837 | 5837 | kernfs_get(cgrp->kn); |
| @@ -1,5 +1,6 @@ | ||
| 1 | 1 | // SPDX-License-Identifier: GPL-2.0-or-later |
| 2 | 2 | |
| 3 | +#include "cgroup-internal.h" | |
| 3 | 4 | #include "cpuset-internal.h" |
| 4 | 5 | |
| 5 | 6 | /* |
| @@ -175,6 +176,7 @@ static int cpuset_write_s64(struct cgroup_subsys_state *css, struct cftype *cft, · 169 unmodified lines | ||
| 175 | 176 | |
| 176 | 177 | switch (type) { |
| 177 | 178 | case FILE_SCHED_RELAX_DOMAIN_LEVEL: |
| 179 | + pr_info_once("cpuset.%s is deprecated\n", cft->name); | |
| 178 | 180 | retval = update_relax_domain_level(cs, val); |
| 179 | 181 | break; |
| 180 | 182 | default: |
| @@ -373,6 +375,46 @@ out: · 192 unmodified lines | ||
| 373 | 375 | return ret; |
| 374 | 376 | } |
| 375 | 377 | |
| 378 | +#ifdef CONFIG_PROC_PID_CPUSET | |
| 379 | +/* | |
| 380 | + * proc_cpuset_show() | |
| 381 | + * - Print tasks cpuset path into seq_file. | |
| 382 | + * - Used for /proc/<pid>/cpuset. | |
| 383 | + */ | |
| 384 | +int proc_cpuset_show(struct seq_file *m, struct pid_namespace *ns, | |
| 385 | + struct pid *pid, struct task_struct *tsk) | |
| 386 | +{ | |
| 387 | + char *buf; | |
| 388 | + struct cgroup_subsys_state *css; | |
| 389 | + int retval; | |
| 390 | + | |
| 391 | + retval = -ENOMEM; | |
| 392 | + buf = kmalloc(PATH_MAX, GFP_KERNEL); | |
| 393 | + if (!buf) | |
| 394 | + goto out; | |
| 395 | + | |
| 396 | + rcu_read_lock(); | |
| 397 | + spin_lock_irq(&css_set_lock); | |
| 398 | + css = task_css(tsk, cpuset_cgrp_id); | |
| 399 | + retval = cgroup_path_ns_locked(css->cgroup, buf, PATH_MAX, | |
| 400 | + current->nsproxy->cgroup_ns); | |
| 401 | + spin_unlock_irq(&css_set_lock); | |
| 402 | + rcu_read_unlock(); | |
| 403 | + | |
| 404 | + if (retval == -E2BIG) | |
| 405 | + retval = -ENAMETOOLONG; | |
| 406 | + if (retval < 0) | |
| 407 | + goto out_free; | |
| 408 | + seq_puts(m, buf); | |
| 409 | + seq_putc(m, '\n'); | |
| 410 | + retval = 0; | |
| 411 | +out_free: | |
| 412 | + kfree(buf); | |
| 413 | +out: | |
| 414 | + return retval; | |
| 415 | +} | |
| 416 | +#endif /* CONFIG_PROC_PID_CPUSET */ | |
| 417 | + | |
| 376 | 418 | static u64 cpuset_read_u64(struct cgroup_subsys_state *css, struct cftype *cft) |
| 377 | 419 | { |
| 378 | 420 | struct cpuset *cs = css_cs(css); |
| @@ -424,24 +466,31 @@ static int cpuset_write_u64(struct cgroup_subsys_state *css, struct cftype *cft, · 45 unmodified lines | ||
| 424 | 466 | retval = cpuset_update_flag(CS_CPU_EXCLUSIVE, cs, val); |
| 425 | 467 | break; |
| 426 | 468 | case FILE_MEM_EXCLUSIVE: |
| 469 | + pr_info_once("cpuset.%s is deprecated\n", cft->name); | |
| 427 | 470 | retval = cpuset_update_flag(CS_MEM_EXCLUSIVE, cs, val); |
| 428 | 471 | break; |
| 429 | 472 | case FILE_MEM_HARDWALL: |
| 473 | + pr_info_once("cpuset.%s is deprecated\n", cft->name); | |
| 430 | 474 | retval = cpuset_update_flag(CS_MEM_HARDWALL, cs, val); |
| 431 | 475 | break; |
| 432 | 476 | case FILE_SCHED_LOAD_BALANCE: |
| 477 | + pr_info_once("cpuset.%s is deprecated, use cpuset.cpus.partition instead\n", cft->name); | |
| 433 | 478 | retval = cpuset_update_flag(CS_SCHED_LOAD_BALANCE, cs, val); |
| 434 | 479 | break; |
| 435 | 480 | case FILE_MEMORY_MIGRATE: |
| 481 | + pr_info_once("cpuset.%s is deprecated\n", cft->name); | |
| 436 | 482 | retval = cpuset_update_flag(CS_MEMORY_MIGRATE, cs, val); |
| 437 | 483 | break; |
| 438 | 484 | case FILE_MEMORY_PRESSURE_ENABLED: |
| 485 | + pr_info_once("cpuset.%s is deprecated, use memory.pressure with CONFIG_PSI instead\n", cft->name); | |
| 439 | 486 | cpuset_memory_pressure_enabled = !!val; |
| 440 | 487 | break; |
| 441 | 488 | case FILE_SPREAD_PAGE: |
| 489 | + pr_info_once("cpuset.%s is deprecated\n", cft->name); | |
| 442 | 490 | retval = cpuset_update_flag(CS_SPREAD_PAGE, cs, val); |
| 443 | 491 | break; |
| 444 | 492 | case FILE_SPREAD_SLAB: |
| 493 | + pr_warn_once("cpuset.%s is deprecated\n", cft->name); | |
| 445 | 494 | retval = cpuset_update_flag(CS_SPREAD_SLAB, cs, val); |
| 446 | 495 | break; |
| 447 | 496 | default: |
| @@ -21,7 +21,6 @@ · 20 unmodified lines | ||
| 21 | 21 | * License. See the file COPYING in the main directory of the Linux |
| 22 | 22 | * distribution for more details. |
| 23 | 23 | */ |
| 24 | −#include "cgroup-internal.h" | |
| 25 | 24 | #include "cpuset-internal.h" |
| 26 | 25 | |
| 27 | 26 | #include <linux/init.h> |
| @@ -4244,50 +4243,6 @@ void cpuset_print_current_mems_allowed(void) · 4216 unmodified lines | ||
| 4244 | 4243 | rcu_read_unlock(); |
| 4245 | 4244 | } |
| 4246 | 4245 | |
| 4247 | −#ifdef CONFIG_PROC_PID_CPUSET | |
| 4248 | −/* | |
| 4249 | − * proc_cpuset_show() | |
| 4250 | − * - Print tasks cpuset path into seq_file. | |
| 4251 | − * - Used for /proc/<pid>/cpuset. | |
| 4252 | − * - No need to task_lock(tsk) on this tsk->cpuset reference, as it | |
| 4253 | − * doesn't really matter if tsk->cpuset changes after we read it, | |
| 4254 | − * and we take cpuset_mutex, keeping cpuset_attach() from changing it | |
| 4255 | − * anyway. | |
| 4256 | − */ | |
| 4257 | −int proc_cpuset_show(struct seq_file *m, struct pid_namespace *ns, | |
| 4258 | − struct pid *pid, struct task_struct *tsk) | |
| 4259 | −{ | |
| 4260 | − char *buf; | |
| 4261 | − struct cgroup_subsys_state *css; | |
| 4262 | − int retval; | |
| 4263 | − | |
| 4264 | − retval = -ENOMEM; | |
| 4265 | − buf = kmalloc(PATH_MAX, GFP_KERNEL); | |
| 4266 | − if (!buf) | |
| 4267 | − goto out; | |
| 4268 | − | |
| 4269 | − rcu_read_lock(); | |
| 4270 | − spin_lock_irq(&css_set_lock); | |
| 4271 | − css = task_css(tsk, cpuset_cgrp_id); | |
| 4272 | − retval = cgroup_path_ns_locked(css->cgroup, buf, PATH_MAX, | |
| 4273 | − current->nsproxy->cgroup_ns); | |
| 4274 | − spin_unlock_irq(&css_set_lock); | |
| 4275 | − rcu_read_unlock(); | |
| 4276 | − | |
| 4277 | − if (retval == -E2BIG) | |
| 4278 | − retval = -ENAMETOOLONG; | |
| 4279 | − if (retval < 0) | |
| 4280 | − goto out_free; | |
| 4281 | − seq_puts(m, buf); | |
| 4282 | − seq_putc(m, '\n'); | |
| 4283 | − retval = 0; | |
| 4284 | −out_free: | |
| 4285 | − kfree(buf); | |
| 4286 | −out: | |
| 4287 | − return retval; | |
| 4288 | −} | |
| 4289 | −#endif /* CONFIG_PROC_PID_CPUSET */ | |
| 4290 | − | |
| 4291 | 4246 | /* Display task mems_allowed in /proc/<pid>/status file. */ |
| 4292 | 4247 | void cpuset_task_status_allowed(struct seq_file *m, struct task_struct *task) |
| 4293 | 4248 | { |
| @@ -430,9 +430,11 @@ static ssize_t freezer_write(struct kernfs_open_file *of, · 429 unmodified lines | ||
| 430 | 430 | |
| 431 | 431 | if (strcmp(buf, freezer_state_strs(0)) == 0) |
| 432 | 432 | freeze = false; |
| 433 | − else if (strcmp(buf, freezer_state_strs(CGROUP_FROZEN)) == 0) | |
| 433 | + else if (strcmp(buf, freezer_state_strs(CGROUP_FROZEN)) == 0) { | |
| 434 | + pr_info_once("Freezing with imperfect legacy cgroup freezer. " | |
| 435 | + "See cgroup.freeze of cgroup v2\n"); | |
| 434 | 436 | freeze = true; |
| 435 | − else | |
| 437 | + } else | |
| 436 | 438 | return -EINVAL; |
| 437 | 439 | |
| 438 | 440 | freezer_change_state(css_freezer(of_css(of)), freeze); |
| @@ -67,22 +67,6 @@ static inline bool valid_type(enum misc_res_type type) · 66 unmodified lines | ||
| 67 | 67 | return type >= 0 && type < MISC_CG_RES_TYPES; |
| 68 | 68 | } |
| 69 | 69 | |
| 70 | −/** | |
| 71 | − * misc_cg_res_total_usage() - Get the current total usage of the resource. | |
| 72 | − * @type: misc res type. | |
| 73 | − * | |
| 74 | − * Context: Any context. | |
| 75 | − * Return: Current total usage of the resource. | |
| 76 | − */ | |
| 77 | −u64 misc_cg_res_total_usage(enum misc_res_type type) | |
| 78 | −{ | |
| 79 | − if (valid_type(type)) | |
| 80 | − return atomic64_read(&root_cg.res[type].usage); | |
| 81 | − | |
| 82 | − return 0; | |
| 83 | −} | |
| 84 | −EXPORT_SYMBOL_GPL(misc_cg_res_total_usage); | |
| 85 | − | |
| 86 | 70 | /** |
| 87 | 71 | * misc_cg_set_capacity() - Set the capacity of the misc cgroup res. |
| 88 | 72 | * @type: Type of the misc res. |
| @@ -299,40 +299,6 @@ static inline void __cgroup_rstat_unlock(struct cgroup *cgrp, int cpu_in_loop) · 298 unmodified lines | ||
| 299 | 299 | spin_unlock_irq(&cgroup_rstat_lock); |
| 300 | 300 | } |
| 301 | 301 | |
| 302 | −/* see cgroup_rstat_flush() */ | |
| 303 | −static void cgroup_rstat_flush_locked(struct cgroup *cgrp) | |
| 304 | − __releases(&cgroup_rstat_lock) __acquires(&cgroup_rstat_lock) | |
| 305 | −{ | |
| 306 | − int cpu; | |
| 307 | − | |
| 308 | − lockdep_assert_held(&cgroup_rstat_lock); | |
| 309 | − | |
| 310 | − for_each_possible_cpu(cpu) { | |
| 311 | − struct cgroup *pos = cgroup_rstat_updated_list(cgrp, cpu); | |
| 312 | − | |
| 313 | − for (; pos; pos = pos->rstat_flush_next) { | |
| 314 | − struct cgroup_subsys_state *css; | |
| 315 | − | |
| 316 | − cgroup_base_stat_flush(pos, cpu); | |
| 317 | − bpf_rstat_flush(pos, cgroup_parent(pos), cpu); | |
| 318 | − | |
| 319 | − rcu_read_lock(); | |
| 320 | − list_for_each_entry_rcu(css, &pos->rstat_css_list, | |
| 321 | − rstat_css_node) | |
| 322 | − css->ss->css_rstat_flush(css, cpu); | |
| 323 | − rcu_read_unlock(); | |
| 324 | − } | |
| 325 | − | |
| 326 | − /* play nice and yield if necessary */ | |
| 327 | − if (need_resched() || spin_needbreak(&cgroup_rstat_lock)) { | |
| 328 | − __cgroup_rstat_unlock(cgrp, cpu); | |
| 329 | − if (!cond_resched()) | |
| 330 | − cpu_relax(); | |
| 331 | − __cgroup_rstat_lock(cgrp, cpu); | |
| 332 | − } | |
| 333 | − } | |
| 334 | −} | |
| 335 | − | |
| 336 | 302 | /** |
| 337 | 303 | * cgroup_rstat_flush - flush stats in @cgrp's subtree |
| 338 | 304 | * @cgrp: target cgroup |
| @@ -348,38 +314,30 @@ static void cgroup_rstat_flush_locked(struct cgroup *cgrp) · 9 unmodified lines | ||
| 348 | 314 | */ |
| 349 | 315 | __bpf_kfunc void cgroup_rstat_flush(struct cgroup *cgrp) |
| 350 | 316 | { |
| 317 | + int cpu; | |
| 318 | + | |
| 351 | 319 | might_sleep(); |
| 320 | + for_each_possible_cpu(cpu) { | |
| 321 | + struct cgroup *pos = cgroup_rstat_updated_list(cgrp, cpu); | |
| 352 | 322 | |
| 353 | − __cgroup_rstat_lock(cgrp, -1); | |
| 354 | − cgroup_rstat_flush_locked(cgrp); | |
| 355 | − __cgroup_rstat_unlock(cgrp, -1); | |
| 356 | −} | |
| 323 | + /* Reacquire for each CPU to avoid disabling IRQs too long */ | |
| 324 | + __cgroup_rstat_lock(cgrp, cpu); | |
| 325 | + for (; pos; pos = pos->rstat_flush_next) { | |
| 326 | + struct cgroup_subsys_state *css; | |
| 357 | 327 | |
| 358 | −/** | |
| 359 | − * cgroup_rstat_flush_hold - flush stats in @cgrp's subtree and hold | |
| 360 | − * @cgrp: target cgroup | |
| 361 | − * | |
| 362 | − * Flush stats in @cgrp's subtree and prevent further flushes. Must be | |
| 363 | − * paired with cgroup_rstat_flush_release(). | |
| 364 | − * | |
| 365 | − * This function may block. | |
| 366 | − */ | |
| 367 | −void cgroup_rstat_flush_hold(struct cgroup *cgrp) | |
| 368 | − __acquires(&cgroup_rstat_lock) | |
| 369 | −{ | |
| 370 | − might_sleep(); | |
| 371 | − __cgroup_rstat_lock(cgrp, -1); | |
| 372 | − cgroup_rstat_flush_locked(cgrp); | |
| 373 | −} | |
| 328 | + cgroup_base_stat_flush(pos, cpu); | |
| 329 | + bpf_rstat_flush(pos, cgroup_parent(pos), cpu); | |
| 374 | 330 | |
| 375 | −/** | |
| 376 | − * cgroup_rstat_flush_release - release cgroup_rstat_flush_hold() | |
| 377 | − * @cgrp: cgroup used by tracepoint | |
| 378 | − */ | |
| 379 | −void cgroup_rstat_flush_release(struct cgroup *cgrp) | |
| 380 | − __releases(&cgroup_rstat_lock) | |
| 381 | −{ | |
| 382 | − __cgroup_rstat_unlock(cgrp, -1); | |
| 331 | + rcu_read_lock(); | |
| 332 | + list_for_each_entry_rcu(css, &pos->rstat_css_list, | |
| 333 | + rstat_css_node) | |
| 334 | + css->ss->css_rstat_flush(css, cpu); | |
| 335 | + rcu_read_unlock(); | |
| 336 | + } | |
| 337 | + __cgroup_rstat_unlock(cgrp, cpu); | |
| 338 | + if (!cond_resched()) | |
| 339 | + cpu_relax(); | |
| 340 | + } | |
| 383 | 341 | } |
| 384 | 342 | |
| 385 | 343 | int cgroup_rstat_init(struct cgroup *cgrp) |
| @@ -612,36 +570,34 @@ static void cgroup_force_idle_show(struct seq_file *seq, struct cgroup_base_stat · 226 unmodified lines | ||
| 612 | 570 | void cgroup_base_stat_cputime_show(struct seq_file *seq) |
| 613 | 571 | { |
| 614 | 572 | struct cgroup *cgrp = seq_css(seq)->cgroup; |
| 615 | − u64 usage, utime, stime, ntime; | |
| 573 | + struct cgroup_base_stat bstat; | |
| 616 | 574 | |
| 617 | 575 | if (cgroup_parent(cgrp)) { |
| 618 | − cgroup_rstat_flush_hold(cgrp); | |
| 619 | − usage = cgrp->bstat.cputime.sum_exec_runtime; | |
| 576 | + cgroup_rstat_flush(cgrp); | |
| 577 | + __cgroup_rstat_lock(cgrp, -1); | |
| 578 | + bstat = cgrp->bstat; | |
| 620 | 579 | cputime_adjust(&cgrp->bstat.cputime, &cgrp->prev_cputime, |
| 621 | − &utime, &stime); | |
| 622 | − ntime = cgrp->bstat.ntime; | |
| 623 | − cgroup_rstat_flush_release(cgrp); | |
| 580 | + &bstat.cputime.utime, &bstat.cputime.stime); | |
| 581 | + __cgroup_rstat_unlock(cgrp, -1); | |
| 624 | 582 | } else { |
| 625 | − /* cgrp->bstat of root is not actually used, reuse it */ | |
| 626 | − root_cgroup_cputime(&cgrp->bstat); | |
| 627 | − usage = cgrp->bstat.cputime.sum_exec_runtime; | |
| 628 | − utime = cgrp->bstat.cputime.utime; | |
| 629 | − stime = cgrp->bstat.cputime.stime; | |
| 630 | − ntime = cgrp->bstat.ntime; | |
| 583 | + root_cgroup_cputime(&bstat); | |
| 631 | 584 | } |
| 632 | 585 | |
| 633 | − do_div(usage, NSEC_PER_USEC); | |
| 634 | − do_div(utime, NSEC_PER_USEC); | |
| 635 | − do_div(stime, NSEC_PER_USEC); | |
| 636 | − do_div(ntime, NSEC_PER_USEC); | |
| 586 | + do_div(bstat.cputime.sum_exec_runtime, NSEC_PER_USEC); | |
| 587 | + do_div(bstat.cputime.utime, NSEC_PER_USEC); | |
| 588 | + do_div(bstat.cputime.stime, NSEC_PER_USEC); | |
| 589 | + do_div(bstat.ntime, NSEC_PER_USEC); | |
| 637 | 590 | |
| 638 | 591 | seq_printf(seq, "usage_usec %llu\n" |
| 639 | 592 | "user_usec %llu\n" |
| 640 | 593 | "system_usec %llu\n" |
| 641 | 594 | "nice_usec %llu\n", |
| 642 | − usage, utime, stime, ntime); | |
| 595 | + bstat.cputime.sum_exec_runtime, | |
| 596 | + bstat.cputime.utime, | |
| 597 | + bstat.cputime.stime, | |
| 598 | + bstat.ntime); | |
| 643 | 599 | |
| 644 | − cgroup_force_idle_show(seq, &cgrp->bstat); | |
| 600 | + cgroup_force_idle_show(seq, &bstat); | |
| 645 | 601 | } |
| 646 | 602 | |
| 647 | 603 | /* Add bpf kfuncs for cgroup_rstat_updated() and cgroup_rstat_flush() */ |
| @@ -1855,9 +1855,11 @@ static int mem_cgroup_swappiness_write(struct cgroup_subsys_state *css, · 1854 unmodified lines | ||
| 1855 | 1855 | if (val > MAX_SWAPPINESS) |
| 1856 | 1856 | return -EINVAL; |
| 1857 | 1857 | |
| 1858 | − if (!mem_cgroup_is_root(memcg)) | |
| 1858 | + if (!mem_cgroup_is_root(memcg)) { | |
| 1859 | + pr_info_once("Per memcg swappiness does not exist in cgroup v2. " | |
| 1860 | + "See memory.reclaim or memory.swap.max there\n "); | |
| 1859 | 1861 | WRITE_ONCE(memcg->swappiness, val); |
| 1860 | − else | |
| 1862 | + } else | |
| 1861 | 1863 | WRITE_ONCE(vm_swappiness, val); |
| 1862 | 1864 | |
| 1863 | 1865 | return 0; |