very high memory usage due to kernfs_node_cache slabs
https://pms.uniontech.com/bug-view-238303.html
bug环境
1 2 uname -a
1 2 3 4 5 6 7 8 apt policy systemd
查看当前用户密码，方便切root：
1 cat /var/log/cloud-init.log | grep ctyun
初步分析
Slab
/proc/meminfo
1 2 3 4 cat /proc/meminfo | grep Slab: -A2
slabtop
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 slabtop -s c -o | head -n20
slub_debug
打开下方slub_debug相关内核编译选项：
1 2 3 4 5 6 7 CONFIG_SLUB=y
在grub中增加参数：
追踪kernfs_node_cache分配释放：
1 echo 1 > /sys/kernel/slab/kernfs_node_cache/trace && sleep 60 && echo 0 > /sys/kernel/slab/kernfs_node_cache/trace
1 /sys/kernel/slab/kernfs_node_cache/group
在/var/log/kern.log中查看kernfs_node_cache alloc:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 1673 2024-01-24 12:40:55 0000000g-A8cUta6pu5 kernel: [ 201.452633] TRACE kernfs_node_cache alloc 0x00000000a05f4917 inuse=35 fp=0x (null)
1 2 grep "kernfs_node_cache alloc" kern.log | wc -l
1 2 3 ./scripts/faddr2line vmlinux vfs_mkdir+0xfc/0x1c0
1 2 3 ./scripts/faddr2line vmlinux kernfs_iop_mkdir+0x58/0x88
1 2 3 4 ./scripts/faddr2line vmlinux cgroup_mkdir+0x2f0/0x4e8
在/var/log/kern.log中查看kernfs_node_cache free:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 3106 2024-01-24 12:40:56 0000000g-A8cUta6pu5 kernel: [ 203.212900] TRACE kernfs_node_cache free 0x00000000e2ea365c inuse=34 fp=0x00000000a8805aea
1 2 grep "kernfs_node_cache free" kern.log | wc -l
trace-bpfcc
1 2 sudo apt install systemd-dbgsymsource systemd=241.61-deepin1
1 trace-bpfcc -tKU 'r::kernfs_new_node "%llx", retval'
从kernfs_new_node.log内容来看，主要有内核线程kworker和systemd在调用kernfs_new_node。
kworker
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 3.796820 47 47 kworker/3:1 kernfs_new_node
在 Linux 内核中，错误代码 -14 对应的含义是 EFAULT，表示发生了“无效地址”（Bad address）错误。通常情况下，EFAULT 错误表示系统调用中传递的参数指针指向的地址无效，无法访问或者无法写入。
sysfs_slab_add
tyy-1053 sysfs_slab_add 参数追踪：
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 trace-bpfcc -t -Ilinux/slab.h -Ilinux/slub_def.h 'sysfs_slab_add(struct kmem_cache *s) "%llx" s->memcg_params.root_cache' 96.61296 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001f10d078096.61350 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001f10d040096.61372 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001f10d388096.61390 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001f10d2e0096.61434 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001f131b50096.61457 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001f0a3a70096.61485 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001f131b88096.61504 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001ff68478096.61535 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001ff7d4b0096.61571 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001ff7d558096.61589 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001ff7d670096.61608 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001ff7d750096.61624 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001ff7d478096.61962 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001f10d2a8096.61981 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001ff68440096.62001 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001ff68718096.62299 23371 23371 kworker/2 :1 sysfs_slab_add ffff8001ff687500
可以看到root_cache不是null。
ret_from_fork+0x10
1 2 3 ./scripts/faddr2line vmlinux ret_from_fork+0x10arch /arm64/kernel/entry.S:1063
vim arch/arm64/kernel/entry.S +1063 1 2 3 4 5 6 7 8 9 10 11 12 1055 1058 ENTRY(ret_from_fork)1059 bl schedule_tail 1060 cbz x19, 1f 1061 mov x0, x20 1062 blr x19 1063 1 : get_thread_info tsk 1064 b ret_to_user 1065 ENDPROC(ret_from_fork) 1066 NOKPROBE(ret_from_fork)
kthread+0x128
1 2 3 ./scripts/faddr2line vmlinux kthread+0x128
// vim kernel/kthread.c +246 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 205 static int kthread (void *_create) 207 208 struct kthread_create_info *create =209 int (*threadfn)(void *data) = create->threadfn; 210 void *data = create->data; 211 struct completion *done ;212 struct kthread *self ;213 int ret; 214 215 self = kzalloc(sizeof (*self), GFP_KERNEL); 216 set_kthread_struct(self); 217 218 219 done = xchg(&create->done, NULL ); 220 if (!done) { 221 kfree(create); 222 do_exit(-EINTR); 223 }224 225 if (!self) { 226 create->result = ERR_PTR(-ENOMEM); 227 complete(done); 228 do_exit(-ENOMEM); 229 }230 231 self->data = data; 232 init_completion(&self->exited); 233 init_completion(&self->parked); 234 current->vfork_done = &self->exited; 235 236 237 __set_current_state(TASK_UNINTERRUPTIBLE); 238 create->result = current; 239 complete(done); 240 schedule(); 241 242 ret = -EINTR; 243 if (!test_bit(KTHREAD_SHOULD_STOP, &self->flags)) { 244 cgroup_kthread_ready(); 245 __kthread_parkme(self); 246 ret = threadfn(data); 247 }248 do_exit(ret); 249 }
worker_thread+0x48
1 2 3 4 5 ./scripts/faddr2line vmlinux worker_thread+0x48
// vim kernel/workqueue.c +2297 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 2226 2236 static int worker_thread (void *__worker) 2238 struct worker *worker =2239 struct worker_pool *pool =2240 2241 2242 set_pf_worker(true );2243 woke_up:2244 spin_lock_irq(&pool->lock); 2245 2246 2247 if (unlikely(worker->flags & WORKER_DIE)) {2248 spin_unlock_irq(&pool->lock); 2249 WARN_ON_ONCE(!list_empty(&worker->entry));2250 set_pf_worker(false );2251 2252 set_task_comm(worker->task, "kworker/dying" );2253 ida_simple_remove(&pool->worker_ida, worker->id);2254 worker_detach_from_pool(worker);2255 kfree(worker);2256 return 0 ;2257 }2258 2259 worker_leave_idle(worker); 2260 recheck:2261 2262 if (!need_more_worker(pool)) 2263 goto sleep; 2264 2265 2266 if (unlikely(!may_start_working(pool)) && manage_workers(worker)) 2267 goto recheck; 2268 2269 2273 WARN_ON_ONCE(!list_empty(&worker->scheduled));2274 2275 2279 worker_clr_flags(worker, WORKER_PREP | WORKER_REBOUND);2280 2281 do {2282 struct work_struct *work =2283 list_first_entry(&pool->worklist,2284 struct work_struct, entry);2285 2286 pool->watchdog_ts = jiffies;2287 2288 if (likely(!(*work_data_bits(work) & WORK_STRUCT_LINKED))) {2289 2290 process_one_work(worker, work);2291 if (unlikely(!list_empty(&worker->scheduled)))2292 process_scheduled_works(worker);2293 } else {2294 move_linked_works(work, &worker->scheduled, NULL );2295 process_scheduled_works(worker);2296 }2297 } while (keep_working(pool));2298 2299 worker_set_flags(worker, WORKER_PREP);2300 sleep:2301 2305 worker_enter_idle(worker); 2306 __set_current_state(TASK_IDLE); 2307 spin_unlock_irq(&pool->lock); 2308 schedule(); 2309 goto woke_up; 2310 }
create_worker
// vim kernel/workqueue.c +1813 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 1779 * create_worker - 创建一个新的工作队列工作者 1780 * @pool: 新工作者将属于的工作池 1781 * 1782 * 创建并启动一个新的工作者，它附加到 @pool。 1783 * 1784 * 上下文： 1785 * 可能睡眠。进行 GFP_KERNEL 分配。 1786 * 1787 * 返回值： 1788 * 指向新创建的工作者的指针。 1789 */ 1790 static struct worker *create_worker (struct worker_pool *pool) 1792 struct worker *worker =NULL ; 1793 int id = -1 ; 1794 char id_buf[16 ]; 1795 1796 1797 id = ida_simple_get(&pool->worker_ida, 0 , 0 , GFP_KERNEL); 1798 if (id < 0 ) 1799 goto fail; 1800 1801 worker = alloc_worker(pool->node); 1802 if (!worker) 1803 goto fail; 1804 1805 worker->id = id; 1806 1807 if (pool->cpu >= 0 ) 1808 snprintf (id_buf, sizeof (id_buf), "%d:%d%s" , pool->cpu, id, 1809 pool->attrs->nice < 0 ? "H" : "" ); 1810 else 1811 snprintf (id_buf, sizeof (id_buf), "u%d:%d" , pool->id, id); 1812 1813 worker->task = kthread_create_on_node(worker_thread, worker, pool->node,1814 "kworker/%s" , id_buf); 1815 if (IS_ERR(worker->task)) 1816 goto fail; 1817 1818 set_user_nice(worker->task, pool->attrs->nice); 1819 kthread_bind_mask(worker->task, pool->attrs->cpumask); 1820 1821 1822 worker_attach_to_pool(worker, pool); 1823 1824 1825 spin_lock_irq(&pool->lock); 1826 worker->pool->nr_workers++; 1827 worker_enter_idle(worker); 1828 wake_up_process(worker->task); 1829 spin_unlock_irq(&pool->lock); 1830 1831 return worker; 1832 1833 fail:1834 if (id >= 0 )1835 ida_simple_remove(&pool->worker_ida, id); 1836 kfree(worker); 1837 return NULL ; 1838 }
kthread_create_on_node
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 347 369 struct task_struct *kthread_create_on_node (int (*threadfn)(void *data), 370 void *data, int node, 371 const char namefmt[], 372 ...) 374 struct task_struct *task ;375 va_list args; 376 377 va_start(args, namefmt); 378 task = __kthread_create_on_node(threadfn, data, node, namefmt, args); 379 va_end(args); 380 381 return task; 382 } 383 EXPORT_SYMBOL(kthread_create_on_node);
maybe_create_worker
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 1943 1958 static void maybe_create_worker (struct worker_pool *pool) releases (&pool->lock) acquires (&pool->lock) 1962 restart:1963 spin_unlock_irq(&pool->lock); 1964 1965 1966 mod_timer(&pool->mayday_timer, jiffies + MAYDAY_INITIAL_TIMEOUT);1967 1968 while (true ) {1969 if (create_worker(pool) || !need_to_create_worker(pool)) 1970 break ;1971 1972 schedule_timeout_interruptible(CREATE_COOLDOWN); 1973 1974 if (!need_to_create_worker(pool)) 1975 break ;1976 }1977 1978 del_timer_sync(&pool->mayday_timer); 1979 spin_lock_irq(&pool->lock); 1980 1984 if (need_to_create_worker(pool)) 1985 goto restart; 1986 }
process_one_work+0x1e8
1 2 3 4 5 6 ./scripts/faddr2line vmlinux process_one_work+0x1e8arch /arm64/include/asm/jump_label.h:20
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 2032 2042 static void process_one_work (struct worker *worker, struct work_struct *work) releases (&pool->lock) acquires (&pool->lock) 2046 struct pool_workqueue *pwq =2047 struct worker_pool *pool =2048 bool cpu_intensive = pwq->wq->flags & WQ_CPU_INTENSIVE; 2049 int work_color;2050 struct worker *collision ;2051 #ifdef CONFIG_LOCKDEP 2052 2056 struct lockdep_map lockdep_map ;2057 2058 lockdep_copy_map(&lockdep_map, &work->lockdep_map);2059 #endif 2060 2061 WARN_ON_ONCE(!(pool->flags & POOL_DISASSOCIATED) &&2062 raw_smp_processor_id() != pool->cpu);2063 2064 2068 collision = find_worker_executing_work(pool, work);2069 if (unlikely(collision)) {2070 move_linked_works(work, &collision->scheduled, NULL ); 2071 return ;2072 }2073 2074 2075 debug_work_deactivate(work); 2076 hash_add(pool->busy_hash, &worker->hentry, (unsigned long )work); 2077 worker->current_work = work; 2078 worker->current_func = work->func; 2079 worker->current_pwq = pwq; 2080 work_color = get_work_color(work); 2081 2082 2085 strscpy(worker->desc, pwq->wq->name, WORKER_DESC_LEN); 2086 2087 list_del_init(&work->entry); 2088 2089 2093 if (unlikely(cpu_intensive))2094 worker_set_flags(worker, WORKER_CPU_INTENSIVE); 2095 2096 2100 if (need_more_worker(pool))2101 wake_up_worker(pool); 2102 2103 2107 set_work_pool_and_clear_pending(work, pool->id); 2108 2109 spin_unlock_irq(&pool->lock); 2110 2111 lock_map_acquire(&pwq->wq->lockdep_map); 2112 lock_map_acquire(&lockdep_map); 2113 2127 lockdep_invariant_state(true ); 2128 trace_workqueue_execute_start(work); 2129 worker->current_func(work); 2130 2133 trace_workqueue_execute_end(work); 2134 lock_map_release(&lockdep_map); 2135 lock_map_release(&pwq->wq->lockdep_map); 2136 2137 if (unlikely(in_atomic() || lockdep_depth(current) > 0 )) {2138 pr_err("BUG: workqueue leaked lock or atomic: %s/0x%08x/%d\n" 2139 " last function: %pf\n" ,2140 current->comm, preempt_count(), task_pid_nr(current),2141 worker->current_func);2142 debug_show_held_locks(current);2143 dump_stack();2144 }2145 2146 2149 cond_resched(); 2150 2151 spin_lock_irq(&pool->lock); 2152 2153 2154 if (unlikely(cpu_intensive))2155 worker_clr_flags(worker, WORKER_CPU_INTENSIVE); 2156 2157 2158 hash_del(&worker->hentry); 2159 worker->current_work = NULL ; 2160 worker->current_func = NULL ; 2161 worker->current_pwq = NULL ; 2162 pwq_dec_nr_in_flight(pwq, work_color); 2163 }
memcg_kmem_cache_create_func+0x1c
1 2 3 4 ./scripts/faddr2line vmlinux memcg_kmem_cache_create_func+0x1c
1 2 3 4 5 6 7 8 9 10 11 12 2500 static void memcg_kmem_cache_create_func (struct work_struct *w) 2502 struct memcg_kmem_cache_create_work *cw =2503 container_of(w, struct memcg_kmem_cache_create_work, work); 2504 struct mem_cgroup *memcg =2505 struct kmem_cache *cachep =2506 2507 memcg_create_kmem_cache(memcg, cachep); 2508 2509 css_put(&memcg->css); 2510 kfree(cw); 2511 }
__memcg_schedule_kmem_cache_create
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 2513 2517 static void __memcg_schedule_kmem_cache_create(struct mem_cgroup *memcg,2518 struct kmem_cache *cachep)2519 {2520 struct memcg_kmem_cache_create_work *cw ;2521 2522 cw = kmalloc(sizeof (*cw), GFP_NOWAIT | __GFP_NOWARN); 2523 if (!cw) 2524 return ; 2525 2526 css_get(&memcg->css); 2527 2528 cw->memcg = memcg; 2529 cw->cachep = cachep; 2530 INIT_WORK(&cw->work, memcg_kmem_cache_create_func); 2531 2532 queue_work(memcg_kmem_cache_wq, &cw->work); 2533 }
__memcg_schedule_kmem_cache_create
memcg_schedule_kmem_cache_create
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 2534 static void memcg_schedule_kmem_cache_create (struct mem_cgroup *memcg, 2535 struct kmem_cache *cachep)2537 2553 current->memcg_kmem_skip_account = 1 ; 2554 __memcg_schedule_kmem_cache_create(memcg, cachep); 2555 current->memcg_kmem_skip_account = 0 ; 2556 }
memcg_kmem_get_cache
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 2560 2571 struct kmem_cache *memcg_kmem_get_cache (struct kmem_cache *cachep) 2573 struct mem_cgroup *memcg ;2574 struct kmem_cache *memcg_cachep ;2575 int kmemcg_id; 2576 2577 VM_BUG_ON(!is_root_cache(cachep)); 2578 2579 if (memcg_kmem_bypass()) 2580 return cachep; 2581 2582 if (current->memcg_kmem_skip_account) 2583 return cachep; 2584 2585 memcg = get_mem_cgroup_from_current(); 2586 kmemcg_id = READ_ONCE(memcg->kmemcg_id); 2587 if (kmemcg_id < 0 ) 2588 goto out; 2589 2590 memcg_cachep = cache_from_memcg_idx(cachep, kmemcg_id); 2591 if (likely(memcg_cachep)) 2592 return memcg_cachep; 2593 2594 2602 memcg_schedule_kmem_cache_create(memcg, cachep); 2603 out:2604 css_put(&memcg->css); 2605 return cachep; 2606 }
用bpf追踪一下哪些进程会调用memcg_kmem_get_cache：
1 trace-bpfcc -t memcg_kmem_get_cache | tee memcg_kmem_get_cache1.log
输出结果如下：
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 TIME PID TID COMM FUNC
貌似每个进程都会调用memcg_kmem_get_cache函数，也就意味着每次分配都会在内存控制组缓存中创建一个新的缓存。
slab_pre_alloc_hook
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 414 static inline struct kmem_cache *slab_pre_alloc_hook (struct kmem_cache *s, 415 gfp_t flags)417 flags &= gfp_allowed_mask; 418 419 fs_reclaim_acquire(flags); 420 fs_reclaim_release(flags); 421 422 might_sleep_if(gfpflags_allow_blocking(flags)); 423 424 if (should_failslab(s, flags)) 425 return NULL ; 426 427 if (memcg_kmem_enabled() && 428 ((flags & __GFP_ACCOUNT) || (s->flags & SLAB_ACCOUNT))) 429 return memcg_kmem_get_cache(s); 430 431 return s; 432 }
此处好像有个递归？？？
1 2 3 4 5 slab_pre_alloc_hook
这个递归解释了memcg_schedule_kmem_cache_create函数注释中提到的可能存在递归的原因。
修复方案
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 From dfd4043beff95e15bad4207cc9db9b29940e20d4 Mon Sep 17 00 :00 :00 2001 16 Apr 2024 11 :12 :58 +0800 "slabtop -s c -o" , kernfs_node_cache is ranked first-4.19 .y. Althoughlog /kern.log :0x0 /0x190 0x14 /0x20 0xa8 /0xcc 0x58 /0x188 .34 +0x31c /0x388 0x210 /0x278 0x60 /0x1f8 0x24 /0x48 0x30 /0x88 0x2f0 /0x4e8 0x58 /0x88 0xfc /0x1c0 0xec /0x100 0x1c /0x28 0x90 /0x178 0x9c /0xa8 0x8 /0xc 3.796820 47 47 kworker/3 :1 kernfs_new_node0x0 [kernel]0x9c [kernel]0x104 [kernel]0x14 [kernel] # 上游4.19 .306 这里没走到0xb8 [kernel] # 上游4.19 .306 这里走到了0x128 [kernel]0xcc [kernel]0xf8 [kernel]0x1c [kernel]0x1e8 [kernel]0x48 [kernel]0x128 [kernel]0x10 [kernel]-14 1519956. .93343e5 100644 -147 ,6 +147 ,7 @@if (root_cache) {NULL ;
linux-4.19.y上*(s->memcg_params.root_cache->memcg_kset) = 0， uos-arm-kernel-1053-tyy上*(s->memcg_params.root_cache->memcg_kset)是个随机值。
kmem_cache通过双向链表管理，kmem_cache父节点的memcg_kset会传导给子节点，故第一个非根节点的kset需要初始化为null。
本次问题虽然修复了，但是感觉uos-arm-kernel-1053-tyy上存在一个CVE。
修复后跟上游对比
systemd
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 TIME PID TID COMM FUNC mkdir +0x14 [libc-2.28.so]
1 2 bpftrace -l | grep cgroup_mkdir
1 trace-bpfcc -tKU cgroup_mkdir | tee cgroup_mkdir.log
vim cgroup_mkdir.log:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 TIME PID TID COMM FUNC
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1626 static int unit_create_cgroup ( 1627 Unit *u,1628 CGroupMask target_mask,1629 CGroupMask enable_mask,1630 ManagerState state)1631 1632 bool created;1633 int r;1634 1635 assert(u);1636 1637 if (!UNIT_HAS_CGROUP_CONTEXT(u))1638 return 0 ;1639 1640 1641 r = unit_pick_cgroup_path(u);1642 if (r < 0 )1643 return r;1644 1645 1646 r = cg_create_everywhere(u->manager->cgroup_supported, target_mask, u->cgroup_path);1647 if (r < 0 )1648 return log_unit_error_errno(u, r, "Failed to create cgroup %s: %m" , u->cgroup_path);1649 created = r;
systemd log
1 2 grep -i failed syslog | grep systemd | wc -l
1 grep -i failed syslog | grep systemd | grep .service | grep -v "Failed to start" > systemd.log
deepin-anything-monitor.service
1 2 grep deepin-anything-monitor.service systemd.log | wc -l
1 2 3 4 5 6 systemctl status deepin-anything-monitor.service
logrotate.service
1 2 grep logrotate.service systemd.log | wc -l
1 2 3 4 5 2024-01-01 00:00:12 0000000g-A8cUta6pu5 systemd[1]: logrotate.service: Failed with result 'exit-code' .'exit-code' .'exit-code' .'exit-code' .'exit-code' .
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 systemctl status logrotate.servicelog files"/var/log/apt/term.log" because parent directory has insecure permissions (It's world writable or writable by group which is not "root") Set "su 1月 25 09:00:04 0000000g-A8cUta6pu5 logrotate[7067]: error: skipping "/var/log/apt/history.log" because parent directory has insecure permissions (It' s world writable or writable by group which is not "root" ) Set "/var/log/mirror/printer/printer.log" because parent directory has insecure permissions (It's world writable or writable by group which is not " 1月 25 09:00:04 0000000g-A8cUta6pu5 logrotate[7067]: error: skipping "/var/log/cups/access_log" because parent directory has insecure permissions (It' s world writable or writable by group which is not "root" ) Set "/var/log/cups/error_log" because parent directory has insecure permissions (It's world writable or writable by group which is not "root") Set " 1月 25 09:00:04 0000000g-A8cUta6pu5 logrotate[7067]: error: skipping "/var/log/mirror/ctyunInstall/ctyunInstall.log" because parent directory has insecure permissions (It' s world writable or writable by group whic'exit-code' .log files.
avahi-daemon.service
1 2 uptime
1 2 grep dbus-org.freedesktop.Avahi.service systemd.log | wc -l
1 2 3 4 2024-01-16 05:45:03 0000000g-A8cUta6pu5 dbus-daemon[539]: [system] Activation via systemd failed for unit 'dbus-org.freedesktop.Avahi.service' : Unit dbus-org.freedesktop.Avahi.service not found.for unit 'dbus-org.freedesktop.Avahi.service' : Unit dbus-org.freedesktop.Avahi.service not found.for unit 'dbus-org.freedesktop.Avahi.service' : Unit dbus-org.freedesktop.Avahi.service not found.for unit 'dbus-org.freedesktop.Avahi.service' : Unit dbus-org.freedesktop.Avahi.service not found.
5S重启一次，本次系统启动也就18个半钟头，跑了143146次。
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 systemctl cat avahi-daemon.service
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 apt show avahi-daemonfor Multicast DNS Service Discovery.local network with no specific configuration. Forprint to, files to look at and people to talk to.which represents your machine
解决方案
1 2 3 sudo apt reinstall avahi-daemon
More
patch
