sysfs sched
分析下 linux/kernel/sched/debug.c 中的内容
/proc/schedstat && /proc/pid/schedstat
实现文件 : sched/stat.c ，一共也就 200 行
- https://docs.kernel.org/scheduler/sched-stats.html 解释 /proc/schedstat 和 /proc/pid/schedstat
- /proc/schedstat 对应函数 show_schedstat ，关联的 config 为 CONFIG_SCHEDSTATS
- /proc/pid/schedstat 对应函数 proc_pid_schedstat ，关联 config 为 CONFIG_SCHED_INFO
- schedstat 可以通过内核启动参数关闭或者 /proc/sys/kernel/sched_schedstats，参考 setup_schedstats 和 sysctl_schedstats
/proc/pid/schedstat
/proc/pid/schedstat 的含义:
time spent on the cpu (in nanoseconds)
time spent waiting on a runqueue (in nanoseconds)
# of timeslices run on this cpu
struct sched_info {
#ifdef CONFIG_SCHED_INFO
/* Cumulative counters: */
/* # of times we have run on this CPU: */
unsigned long pcount;
/* Time spent waiting on a runqueue: */
unsigned long long run_delay;
/* Timestamps: */
/* When did we last run on a CPU? */
unsigned long long last_arrival;
/* When were we last queued to run? */
unsigned long long last_queued;
#endif /* CONFIG_SCHED_INFO */
};
- 无法理解什么叫做 of times we have run on this CPU ，这个参数和 cpu 运行是没有关系的吧
[ ] 密切关联的
config TASK_DELAY_ACCT
bool "Enable per-task delay accounting"
depends on TASKSTATS
select SCHED_INFO
help
Collect information on time spent by a task waiting for system
resources like cpu, synchronous block I/O completion and swapping
in pages. Such statistics can help in setting a task's priorities
relative to other tasks for cpu, io, rss limits etc.
Say N if unsure.
[ ] 回答这个问题
https://stackoverflow.com/questions/52736873/understanding-scheduling-and-proc-pid-sched
/proc/pid/sched
fs/proc/base.c 中 sched_show -> proc_sched_show_task 关联的 config 为 CONFIG_SCHED_DEBUG 。其中的内容是受到 CONFIG_SCHEDSTATS 影响的。
实现的函数在:
- proc_sched_show_task
- print_cfs_stats
- 以及 dump struct sched_statistics 中内容
- show_numa_stats
qemu (11176, #threads: 124)
-------------------------------------------------------------------
se.exec_start : 140935799.916922
se.vruntime : 10268.049254
se.sum_exec_runtime : 43129.162737
se.nr_migrations : 3645
nr_switches : 218495
nr_voluntary_switches : 218380
nr_involuntary_switches : 115
se.load.weight : 1048576
se.runnable_weight : 1048576
se.avg.load_sum : 19
se.avg.runnable_load_sum : 19
se.avg.util_sum : 19456
se.avg.load_avg : 0
se.avg.runnable_load_avg : 0
se.avg.util_avg : 0
se.avg.last_update_time : 140935799916544
se.avg.util_est.ewma : 10
se.avg.util_est.enqueued : 0
policy : 0
prio : 120
clock-delta : 30
mm->numa_scan_seq : 61
numa_pages_migrated : 724
numa_preferred_nid : 5
total_numa_faults : 3440
current_node=5, numa_group_id=11176
numa_faults node=0 task_private=1 task_shared=1 group_private=1 group_shared=1
numa_faults node=1 task_private=1 task_shared=1 group_private=1 group_shared=1
numa_faults node=2 task_private=1 task_shared=1 group_private=1 group_shared=1
numa_faults node=3 task_private=1 task_shared=0 group_private=1 group_shared=0
numa_faults node=4 task_private=1 task_shared=1 group_private=1 group_shared=1
numa_faults node=5 task_private=3427 task_shared=1 group_private=3427 group_shared=1
numa_faults node=6 task_private=1 task_shared=1 group_private=1 group_shared=1
numa_faults node=7 task_private=1 task_shared=0 group_private=1 group_shared=0
/proc/pid/status
实现函数 : proc_pid_status
voluntary_ctxt_switches: 0
nonvoluntary_ctxt_switches: 0
x86_Thread_features:
x86_Thread_features_locked:
/proc/stat
文档: https://man7.org/linux/man-pages/man5/proc_stat.5.html
可以通过 fs/proc/stat.c:show_stat 将各种时间的统计开始结束时间确定下
/* Copy values here to work around gcc-2.95.3, gcc-2.96 */
user = cpustat[CPUTIME_USER];
nice = cpustat[CPUTIME_NICE];
system = cpustat[CPUTIME_SYSTEM];
idle = get_idle_time(&kcpustat, i);
iowait = get_iowait_time(&kcpustat, i);
irq = cpustat[CPUTIME_IRQ];
softirq = cpustat[CPUTIME_SOFTIRQ];
steal = cpustat[CPUTIME_STEAL];
guest = cpustat[CPUTIME_GUEST];
guest_nice = cpustat[CPUTIME_GUEST_NICE];
通过 /proc/stat 来实现 top 1 ，展示各个 CPU 在各种任务上话费了多少时间。
debugfs
/sys/kernel/debug/sched
debug
展示每一个 cfs_rq 的信息，也就是 cgroup 层级 * cpu 数量
cfs_rq[31]:/test
.exec_clock : 0.000000
.MIN_vruntime : 0.000001
.min_vruntime : 69643.566731
.max_vruntime : 0.000001
.spread : 0.000000
.spread0 : -854710191.753634
.nr_spread_over : 0
.nr_running : 1
.h_nr_running : 1
.idle_nr_running : 0
.idle_h_nr_running : 0
.load : 1048576
.load_avg : 1024
.runnable_avg : 1024
.util_avg : 1019
.util_est_enqueued : 0
.removed.load_avg : 0
.removed.util_avg : 0
.removed.runnable_avg : 0
.tg_load_avg_contrib : 1024
.tg_load_avg : 32763
.throttled : 0
.throttle_count : 0
.se->exec_start : 1083661332.978182
.se->vruntime : 1142880506.076297
.se->sum_exec_runtime : 69682.607166
.se->load.weight : 32773
.se->avg.load_avg : 32
.se->avg.util_avg : 1019
.se->avg.runnable_avg : 1024
features
GENTLE_FAIR_SLEEPERS START_DEBIT NO_NEXT_BUDDY LAST_BUDDY CACHE_HOT_BUDDY WAKEUP_PREEMPTION NO_HRTICK NO_HRTICK_DL NO_DOUBLE_TICK NONTASK_CAPACITY TTWU_QUEUE NO_SIS_PROP SIS_UTIL NO_WARN_DOUBLE_CLOCK RT_PUSH_IPI NO_RT_RUNTIME_SHARE NO_LB_MIN ATTACH_AGE_LOAD WA_IDLE WA_WEIGHT WA_BIAS UTIL_EST UTIL_EST_FASTUP NO_LATENCY_WARN ALT_PERIOD BASE_SLICE
idle_min_granularity_ns
750000
latency_ns
latency_warn_ms
latency_warn_once
migration_cost_ns
min_granularity_ns
nr_migrate
preempt
tunable_scaling
verbose
wakeup_granularity_ns
echo w | sudo tee /proc/sysrq-trigger
- sysrq_sched_debug_show : echo t 进去
- sysrq_showstate_blocked_op : echo w 进去
/proc 创建目录和 /proc/pid/task 下的目录都是动态生成的
很显然，就是 thread group 组成 /proc :
/* for the /proc/TGID/task/ directories */
static int proc_task_readdir(struct file *file, struct dir_context *ctx)
忽然意识到，这个目录是查询的过程中动态展示出来的，参考
- proc_pid_readdir
- 只是有点奇怪的是，这个遍历效率很低啊
proc_pid_make_inode ，
#0 proc_pid_make_inode (sb=0xffff888103c28000, task=task@entry=0xffff8881109b8000, mode=33188) at fs/proc/base.c:1890
#1 0xffffffff814c8f15 in proc_pident_instantiate (dentry=dentry@entry=0xffff88811da25680, task=task@entry=0xffff8881109b8000, ptr=ptr@entry=0xffffffff8264c0e8 <tgid_base_stuff+520>) at fs/proc/base.c:2643
#2 0xffffffff814c908e in proc_pident_lookup (end=0xffffffff8264c6b0, p=0xffffffff8264c0e8 <tgid_base_stuff+520>, dentry=0xffff88811da25680, dir=<optimized out>) at fs/proc/base.c:2679
#3 proc_tgid_base_lookup (dir=<optimized out>, dentry=0xffff88811da25680, flags=<optimized out>) at fs/proc/base.c:3378
#4 0xffffffff8143f15a in lookup_open (op=0xffffc90000017edc, op=0xffffc90000017edc, got_write=false, file=0xffff888106eceb00, nd=0xffffc90000017dc0) at fs/namei.c:3394
#5 open_last_lookups (op=0xffffc90000017edc, file=0xffff888106eceb00, nd=0xffffc90000017dc0) at fs/namei.c:3484
#6 path_openat (nd=nd@entry=0xffffc90000017dc0, op=op@entry=0xffffc90000017edc, flags=flags@entry=65) at fs/namei.c:3712
#7 0xffffffff81440d06 in do_filp_open (dfd=dfd@entry=-100, pathname=pathname@entry=0xffff888110a72000, op=op@entry=0xffffc90000017edc) at fs/namei.c:3742
#8 0xffffffff81427aea in do_sys_openat2 (dfd=-100, filename=<optimized out>, how=how@entry=0xffffc90000017f18) at fs/open.c:1356
#9 0xffffffff81427fe7 in do_sys_open (mode=<optimized out>, flags=<optimized out>, filename=<optimized out>, dfd=<optimized out>) at fs/open.c:1372
#10 __do_sys_openat (mode=<optimized out>, flags=<optimized out>, filename=<optimized out>, dfd=<optimized out>) at fs/open.c:1388
#11 __se_sys_openat (mode=<optimized out>, flags=<optimized out>, filename=<optimized out>, dfd=<optimized out>) at fs/open.c:1383
#12 __x64_sys_openat (regs=<optimized out>) at fs/open.c:1383
#13 0xffffffff822a0f0c in do_syscall_x64 (nr=<optimized out>, regs=0xffffc90000017f58) at arch/x86/entry/common.c:50
#14 do_syscall_64 (regs=0xffffc90000017f58, nr=<optimized out>) at arch/x86/entry/common.c:80
/proc/pid
主要看看其中的软链接是如何实现的?
/proc/pid/stat
获取一个 process 上次运行所在的 CPU :
awk '{print "PID: " $1, "Last CPU: " $39}' /proc/self/stat
/proc/pid/cmdline
proc_pid_cmdline_ops
实现的非常复杂，不是一行的事
/proc/pid/pwd
- proc_cwd_link
- get_fs_pwd
- path_get
- get_fs_pwd
然后获取:
struct task_struct {
// ...
/* Filesystem information: */
struct fs_struct *fs;
// ...
如果一个进程的 pwd 被删除了，在 /proc/10204/cwd 将是:
cwd -> '/root/gg (deleted)'
如果重新创建 /root/gg ，输出还是如此。
这个现象可以继续跟踪一下。
/proc/sys/kernel 和 sched 相关控制项目
https://docs.kernel.org/admin-guide/sysctl/kernel.html
- threads-max : 系统中到底可以有多少个 thread
- pid_max : 当 pid 到多少的时候，那么 pid 重新从最小的开始分配
- sched_cfs_bandwidth_slice_us
- sched_deadline_period_max_us
- sched_deadline_period_min_us
- sched_energy_aware : ???
- sched_rr_timeslice_ms
- sched_rt_period_us
- sched_rt_runtime_us
- sched_schedstats : 默认为 0
- sched_autogroup_enabled
- sched_util_clamp_max
- sched_util_clamp_min
- sched_util_clamp_min_rt_default
本站所有文章转发 CSDN 将按侵权追究法律责任，其它情况随意。