numa 自动balance 的bug分析
关于migrate_swap() 和 active_balance()之间的hardlock
背景：这个是在3.10.0-957.el7.x86_64 遇到的一例crash
下面列一下我们是怎么排查并解这个问题的。
一、故障现象
Oppo云智能监控发现机器down机：
KERNEL: /usr/lib/debug/lib/modules/3.10.0-957.el7.x86_64/vmlinux
....
PANIC: "Kernel panic - not syncing: Hard LOCKUP"
PID: 14
COMMAND: "migration/1"
TASK: ffff8f1bf6bb9040 [THREAD_INFO: ffff8f1bf6bc4000]
CPU: 1
STATE: TASK_INTERRUPTIBLE (PANIC)
crash> bt
PID: 14 TASK: ffff8f1bf6bb9040 CPU: 1 COMMAND: "migration/1"
#0 [ffff8f4afbe089f0] machine_kexec at ffffffff83863674
#1 [ffff8f4afbe08a50] __crash_kexec at ffffffff8391ce12
#2 [ffff8f4afbe08b20] panic at ffffffff83f5b4db
#3 [ffff8f4afbe08ba0] nmi_panic at ffffffff8389739f
#4 [ffff8f4afbe08bb0] watchdog_overflow_callback at ffffffff83949241
#5 [ffff8f4afbe08bc8] __perf_event_overflow at ffffffff839a1027
#6 [ffff8f4afbe08c00] perf_event_overflow at ffffffff839aa694
#7 [ffff8f4afbe08c10] intel_pmu_handle_irq at ffffffff8380a6b0
#8 [ffff8f4afbe08e38] perf_event_nmi_handler at ffffffff83f6b031
#9 [ffff8f4afbe08e58] nmi_handle at ffffffff83f6c8fc
#10 [ffff8f4afbe08eb0] do_nmi at ffffffff83f6cbd8
#11 [ffff8f4afbe08ef0] end_repeat_nmi at ffffffff83f6bd69
[exception RIP: native_queued_spin_lock_slowpath+462]
RIP: ffffffff839121ae RSP: ffff8f1bf6bc7c50 RFLAGS: 00000002
RAX: 0000000000000001 RBX: 0000000000000082 RCX: 0000000000000001
RDX: 0000000000000101 RSI: 0000000000000001 RDI: ffff8f1afdf55fe8---锁
RBP: ffff8f1bf6bc7c50 R8: 0000000000000101 R9: 0000000000000400
R10: 000000000000499e R11: 000000000000499f R12: ffff8f1afdf55fe8
R13: ffff8f1bf5150000 R14: ffff8f1afdf5b488 R15: ffff8f1bf5187818
ORIG_RAX: ffffffffffffffff CS: 0010 SS: 0018
--- <NMI exception stack> ---
#12 [ffff8f1bf6bc7c50] native_queued_spin_lock_slowpath at ffffffff839121ae
#13 [ffff8f1bf6bc7c58] queued_spin_lock_slowpath at ffffffff83f5bf4b
#14 [ffff8f1bf6bc7c68] _raw_spin_lock_irqsave at ffffffff83f6a487
#15 [ffff8f1bf6bc7c80] cpu_stop_queue_work at ffffffff8392fc70
#16 [ffff8f1bf6bc7cb0] stop_one_cpu_nowait at ffffffff83930450
#17 [ffff8f1bf6bc7cc0] load_balance at ffffffff838e4c6e
#18 [ffff8f1bf6bc7da8] idle_balance at ffffffff838e5451
#19 [ffff8f1bf6bc7e00] __schedule at ffffffff83f67b14
#20 [ffff8f1bf6bc7e88] schedule at ffffffff83f67bc9
#21 [ffff8f1bf6bc7e98] smpboot_thread_fn at ffffffff838ca562
#22 [ffff8f1bf6bc7ec8] kthread at ffffffff838c1c31
#23 [ffff8f1bf6bc7f50] ret_from_fork_nospec_begin at ffffffff83f74c1d
crash>
二、故障现象分析
hardlock一般是由于关中断时间过长，从堆栈看，上面的"migration/1" 进程在抢spinlock，
由于_raw_spin_lock_irqsave 会先调用 arch_local_irq_disable,然后再去拿锁，而
arch_local_irq_disable 是常见的关中断函数,下面分析这个进程想要拿的锁被谁拿着。
x86架构下，native_queued_spin_lock_slowpath的rdi就是存放锁地址的
crash> arch_spinlock_t ffff8f1afdf55fe8
struct arch_spinlock_t {
val = {
counter = 257
}
}
下面，我们需要了解，这个是一把什么锁。
从调用链分析 idle_balance-->load_balance-->stop_one_cpu_nowait-->cpu_stop_queue_work
反汇编 cpu_stop_queue_work 拿锁阻塞的代码：
crash> dis -l ffffffff8392fc70
/usr/src/debug/kernel-3.10.0-957.el7/linux-3.10.0-957.el7.x86_64/kernel/stop_machine.c: 91
0xffffffff8392fc70 <cpu_stop_queue_work+48>: cmpb $0x0,0xc(%rbx)
85 static void cpu_stop_queue_work(unsigned int cpu, struct cpu_stop_work *work)
86 {
87 struct cpu_stopper *stopper = &per_cpu(cpu_stopper, cpu);
88 unsigned long flags;
89
90 spin_lock_irqsave(&stopper->lock, flags);---所以是卡在拿这把锁
91 if (stopper->enabled)
92 __cpu_stop_queue_work(stopper, work);
93 else
94 cpu_stop_signal_done(work->done, false);
95 spin_unlock_irqrestore(&stopper->lock, flags);
96 }
看起来 需要根据cpu号，来获取对应的percpu变量 cpu_stopper,这个入参在 load_balance 函数中找到的
最忙的rq，然后获取其对应的cpu号:
6545 static int load_balance(int this_cpu, struct rq *this_rq,
6546 struct sched_domain *sd, enum cpu_idle_type idle,
6547 int *should_balance)
6548 {
....
6735 if (active_balance) {
6736 stop_one_cpu_nowait(cpu_of(busiest),
6737 active_load_balance_cpu_stop, busiest,
6738 &busiest->active_balance_work);
6739 }
....
6781 }
crash> dis -l load_balance |grep stop_one_cpu_nowait -B 6
0xffffffff838e4c4d <load_balance+2045>: callq 0xffffffff83f6a0e0 <_raw_spin_unlock_irqrestore>
/usr/src/debug/kernel-3.10.0-957.el7/linux-3.10.0-957.el7.x86_64/kernel/sched/fair.c: 6736
0xffffffff838e4c52 <load_balance+2050>: mov 0x930(%rbx),%edi------------根据rbx可以取cpu号，rbx就是最忙的rq
0xffffffff838e4c58 <load_balance+2056>: lea 0x908(%rbx),%rcx
0xffffffff838e4c5f <load_balance+2063>: mov %rbx,%rdx
0xffffffff838e4c62 <load_balance+2066>: mov $0xffffffff838de690,%rsi
0xffffffff838e4c69 <load_balance+2073>: callq 0xffffffff83930420 <stop_one_cpu_nowait>
然后我们再栈中取的数据如下：
最忙的组是：
crash> rq.cpu ffff8f1afdf5ab80
cpu = 26
也就是说，1号cpu在等 percpu变量cpu_stopper 的26号cpu的锁。
然后我们搜索这把锁在其他哪个进程的栈中，找到了如下：
ffff8f4957fbfab0: ffff8f1afdf55fe8 --------这个在 355608 的栈中
crash> kmem ffff8f4957fbfab0
PID: 355608
COMMAND: "custom_exporter"
TASK: ffff8f4aea3a8000 [THREAD_INFO: ffff8f4957fbc000]
CPU: 26--------刚好也是运行在26号cpu的进程
STATE: TASK_RUNNING (ACTIVE)
下面，就需要分析，为什么位于26号cpu的进程 custom_exporter 会长时间拿着 ffff8f1afdf55fe8
我们来分析26号cpu的堆栈：
crash> bt -f 355608
PID: 355608 TASK: ffff8f4aea3a8000 CPU: 26 COMMAND: "custom_exporter"
.....
#3 [ffff8f1afdf48ef0] end_repeat_nmi at ffffffff83f6bd69
[exception RIP: try_to_wake_up+114]
RIP: ffffffff838d63d2 RSP: ffff8f4957fbfa30 RFLAGS: 00000002
RAX: 0000000000000001 RBX: ffff8f1bf6bb9844 RCX: 0000000000000000
RDX: 0000000000000001 RSI: 0000000000000003 RDI: ffff8f1bf6bb9844
RBP: ffff8f4957fbfa70 R8: ffff8f4afbe15ff0 R9: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
R13: ffff8f1bf6bb9040 R14: 0000000000000000 R15: 0000000000000003
ORIG_RAX: ffffffffffffffff CS: 0010 SS: 0000
--- <NMI exception stack> ---
#4 [ffff8f4957fbfa30] try_to_wake_up at ffffffff838d63d2
ffff8f4957fbfa38: 000000000001ab80 0000000000000086
ffff8f4957fbfa48: ffff8f4afbe15fe0 ffff8f4957fbfb48
ffff8f4957fbfa58: 0000000000000001 ffff8f4afbe15fe0
ffff8f4957fbfa68: ffff8f1afdf55fe0 ffff8f4957fbfa80
ffff8f4957fbfa78: ffffffff838d6705
#5 [ffff8f4957fbfa78] wake_up_process at ffffffff838d6705
ffff8f4957fbfa80: ffff8f4957fbfa98 ffffffff8392fc05
#6 [ffff8f4957fbfa88] __cpu_stop_queue_work at ffffffff8392fc05
ffff8f4957fbfa90: 000000000000001a ffff8f4957fbfbb0
ffff8f4957fbfaa0: ffffffff8393037a
#7 [ffff8f4957fbfaa0] stop_two_cpus at ffffffff8393037a
.....
ffff8f4957fbfbb8: ffffffff838d3867
#8 [ffff8f4957fbfbb8] migrate_swap at ffffffff838d3867
ffff8f4957fbfbc0: ffff8f4aea3a8000 ffff8f1ae77dc100 -------栈中的 migration_swap_arg
ffff8f4957fbfbd0: 000000010000001a 0000000080490f7c
ffff8f4957fbfbe0: ffff8f4aea3a8000 ffff8f4957fbfc30
ffff8f4957fbfbf0: 0000000000000076 0000000000000076
ffff8f4957fbfc00: 0000000000000371 ffff8f4957fbfce8
ffff8f4957fbfc10: ffffffff838dd0ba
#9 [ffff8f4957fbfc10] task_numa_migrate at ffffffff838dd0ba
ffff8f4957fbfc18: ffff8f1afc121f40 000000000000001a
ffff8f4957fbfc28: 0000000000000371 ffff8f4aea3a8000 ---这里ffff8f4957fbfc30 就是 task_numa_env 的存放在栈中的地址
ffff8f4957fbfc38: 000000000000001a 000000010000003f
ffff8f4957fbfc48: 000000000000000b 000000000000022c
ffff8f4957fbfc58: 00000000000049a0 0000000000000012
ffff8f4957fbfc68: 0000000000000001 0000000000000003
ffff8f4957fbfc78: 000000000000006f 000000000000499f
ffff8f4957fbfc88: 0000000000000012 0000000000000001
ffff8f4957fbfc98: 0000000000000070 ffff8f1ae77dc100
ffff8f4957fbfca8: 00000000000002fb 0000000000000001
ffff8f4957fbfcb8: 0000000080490f7c ffff8f4aea3a8000 ---rbx压栈在此，所以这个就是current
ffff8f4957fbfcc8: 0000000000017a48 0000000000001818
ffff8f4957fbfcd8: 0000000000000018 ffff8f4957fbfe20
ffff8f4957fbfce8: ffff8f4957fbfcf8 ffffffff838dd4d3
#10 [ffff8f4957fbfcf0] numa_migrate_preferred at ffffffff838dd4d3
ffff8f4957fbfcf8: ffff8f4957fbfd88 ffffffff838df5b0
.....
crash>
crash>
整体上看，26号上的cpu也正在进行numa的balance动作，简单展开介绍一下numa在balance下的动作
在 task_tick_fair 函数中：
static void task_tick_fair(struct rq *rq, struct task_struct *curr, int queued)
{
struct cfs_rq *cfs_rq;
struct sched_entity *se = &curr->se;
for_each_sched_entity(se) {
cfs_rq = cfs_rq_of(se);
entity_tick(cfs_rq, se, queued);
}
if (numabalancing_enabled)----------如果开启numabalancing，则会调用task_tick_numa
task_tick_numa(rq, curr);
update_rq_runnable_avg(rq, 1);
}
而 task_tick_numa 会根据扫描情况，将当前进程需要numa_balance的时候推送到一个work中。
通过调用change_prot_numa将所有映射到VMA的PTE页表项该为PAGE_NONE，使得下次进程访问页表的时候
产生缺页中断，handle_pte_fault 函数 会由于缺页中断的机会来根据numa 选择更好的node，具体不再展开。
在 26号cpu的调用链中，stop_two_cpus-->cpu_stop_queue_two_works-->__cpu_stop_queue_work 函数
由于 cpu_stop_queue_two_works 被内联了，但是 cpu_stop_queue_two_works 调用 __cpu_stop_queue_work
有两次，所以需要根据压栈地址判断当前是哪次调用出现问题。
227 static int cpu_stop_queue_two_works(int cpu1, struct cpu_stop_work *work1,
228 int cpu2, struct cpu_stop_work *work2)
229 {
230 struct cpu_stopper *stopper1 = per_cpu_ptr(&cpu_stopper, cpu1);
231 struct cpu_stopper *stopper2 = per_cpu_ptr(&cpu_stopper, cpu2);
232 int err;
233
234 lg_double_lock(&stop_cpus_lock, cpu1, cpu2);
235 spin_lock_irq(&stopper1->lock);---注意到这里已经持有了stopper1的锁
236 spin_lock_nested(&stopper2->lock, SINGLE_DEPTH_NESTING);
.....
243 __cpu_stop_queue_work(stopper1, work1);
244 __cpu_stop_queue_work(stopper2, work2);
.....
251 }
根据压栈的地址:
#5 [ffff8f4957fbfa78] wake_up_process at ffffffff838d6705
ffff8f4957fbfa80: ffff8f4957fbfa98 ffffffff8392fc05
#6 [ffff8f4957fbfa88] __cpu_stop_queue_work at ffffffff8392fc05
ffff8f4957fbfa90: 000000000000001a ffff8f4957fbfbb0
ffff8f4957fbfaa0: ffffffff8393037a
#7 [ffff8f4957fbfaa0] stop_two_cpus at ffffffff8393037a
ffff8f4957fbfaa8: 0000000100000001 ffff8f1afdf55fe8
crash> dis -l ffffffff8393037a 2
/usr/src/debug/kernel-3.10.0-957.el7/linux-3.10.0-957.el7.x86_64/kernel/stop_machine.c: 244
0xffffffff8393037a <stop_two_cpus+394>: lea 0x48(%rsp),%rsi
0xffffffff8393037f <stop_two_cpus+399>: mov %r15,%rdi
说明压栈的是244行的地址，也就是说目前调用的是243行的 __cpu_stop_queue_work。
然后分析对应的入参：
crash> task_numa_env ffff8f4957fbfc30
struct task_numa_env {
p = 0xffff8f4aea3a8000,
src_cpu = 26,
src_nid = 0,
dst_cpu = 63,
dst_nid = 1,
src_stats = {
nr_running = 11,
load = 556, ---load高
compute_capacity = 18848, ---容量相当
task_capacity = 18,
has_free_capacity = 1
},
dst_stats = {
nr_running = 3,
load = 111, ---load低，且容量相当，要迁移过来
compute_capacity = 18847, ---容量相当
task_capacity = 18,
has_free_capacity = 1
},
imbalance_pct = 112,
idx = 0,
best_task = 0xffff8f1ae77dc100, ---要对调的task，是通过 task_numa_find_cpu-->task_numa_compare-->task_numa_assign 来获取的
best_imp = 763,
best_cpu = 1---最佳的swap的对象对于1号cpu
}
crash> migration_swap_arg ffff8f4957fbfbc0
struct migration_swap_arg {
src_task = 0xffff8f4aea3a8000,
dst_task = 0xffff8f1ae77dc100,
src_cpu = 26,
dst_cpu = 1-----选择的dst cpu为1
}
根据 cpu_stop_queue_two_works 的代码，它在持有 cpu_stopper:26号cpu锁的情况下，去
调用try_to_wake_up ，wake的对象是 用来migrate的 kworker。
static void __cpu_stop_queue_work(struct cpu_stopper *stopper,
struct cpu_stop_work *work)
{
list_add_tail(&work->list, &stopper->works);
wake_up_process(stopper->thread);//其实一般就是唤醒 migration
}
由于最佳的cpu对象为1，所以需要cpu上的migrate来拉取进程。
crash> p cpu_stopper:1
per_cpu(cpu_stopper, 1) = $33 = {
thread = 0xffff8f1bf6bb9040, ----需要唤醒的目的task
lock = {
{
rlock = {
raw_lock = {
val = {
counter = 1
}
}
}
}
},
enabled = true,
works = {
next = 0xffff8f4957fbfac0,
prev = 0xffff8f4957fbfac0
},
stop_work = {
list = {
next = 0xffff8f4afbe16000,
prev = 0xffff8f4afbe16000
},
fn = 0xffffffff83952100,
arg = 0x0,
done = 0xffff8f1ae3647c08
}
}
crash> kmem 0xffff8f1bf6bb9040
CACHE NAME OBJSIZE ALLOCATED TOTAL SLABS SSIZE
ffff8eecffc05f00 task_struct 4152 1604 2219 317 32k
SLAB MEMORY NODE TOTAL ALLOCATED FREE
fffff26501daee00 ffff8f1bf6bb8000 1 7 7 0
FREE / [ALLOCATED]
[ffff8f1bf6bb9040]
PID: 14
COMMAND: "migration/1"--------------目的task就是对应的cpu上的migration
TASK: ffff8f1bf6bb9040 [THREAD_INFO: ffff8f1bf6bc4000]
CPU: 1
STATE: TASK_INTERRUPTIBLE (PANIC)
PAGE PHYSICAL MAPPING INDEX CNT FLAGS
fffff26501daee40 3076bb9000 0 0 0 6fffff00008000 tail
现在的问题是，虽然我们知道了当前cpu26号进程在拿了锁的情况下去唤醒1号cpu上的migrate进程，
那么为什么会迟迟不释放锁，导致1号cpu因为等待该锁时间过长而触发了hardlock的panic呢？
下面就分析，为什么它持锁的时间这么长：
#3 [ffff8f1afdf48ef0] end_repeat_nmi at ffffffff83f6bd69
[exception RIP: try_to_wake_up+114]
RIP: ffffffff838d63d2 RSP: ffff8f4957fbfa30 RFLAGS: 00000002
RAX: 0000000000000001 RBX: ffff8f1bf6bb9844 RCX: 0000000000000000
RDX: 0000000000000001 RSI: 0000000000000003 RDI: ffff8f1bf6bb9844
RBP: ffff8f4957fbfa70 R8: ffff8f4afbe15ff0 R9: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
R13: ffff8f1bf6bb9040 R14: 0000000000000000 R15: 0000000000000003
ORIG_RAX: ffffffffffffffff CS: 0010 SS: 0000
--- <NMI exception stack> ---
#4 [ffff8f4957fbfa30] try_to_wake_up at ffffffff838d63d2
ffff8f4957fbfa38: 000000000001ab80 0000000000000086
ffff8f4957fbfa48: ffff8f4afbe15fe0 ffff8f4957fbfb48
ffff8f4957fbfa58: 0000000000000001 ffff8f4afbe15fe0
ffff8f4957fbfa68: ffff8f1afdf55fe0 ffff8f4957fbfa80
crash> dis -l ffffffff838d63d2
/usr/src/debug/kernel-3.10.0-957.el7/linux-3.10.0-957.el7.x86_64/kernel/sched/core.c: 1790
0xffffffff838d63d2 <try_to_wake_up+114>: mov 0x28(%r13),%eax
1721 static int
1722 try_to_wake_up(struct task_struct *p, unsigned int state, int wake_flags)
1723 {
.....
1787 * If the owning (remote) cpu is still in the middle of schedule() with
1788 * this task as prev, wait until its done referencing the task.
1789 */
1790 while (p->on_cpu)---------原来循环在此
1791 cpu_relax();
.....
1814 return success;
1815 }
我们用一个简单的图来表示一下这个hardlock：
CPU1 CPU26
schedule(.prev=migrate/1) <fault>
pick_next_task() ...
idle_balance() migrate_swap()
active_balance() stop_two_cpus()
spin_lock(stopper0->lock)
spin_lock(stopper1->lock)
try_to_wake_up
pause() -- waits for schedule()
stop_one_cpu(1)
spin_lock(stopper26->lock) -- waits for stopper lock
查看上游的补丁，
static void __cpu_stop_queue_work(struct cpu_stopper *stopper,
- struct cpu_stop_work *work)
+ struct cpu_stop_work *work,
+ struct wake_q_head *wakeq)
{
list_add_tail(&work->list, &stopper->works);
- wake_up_process(stopper->thread);
+ wake_q_add(wakeq, stopper->thread);
}
三、故障复现
1.由于这个是一个race condition导致的hardlock，逻辑上分析已经没有问题了，就没有花时间去复现，
该环境运行一个dpdk的node，不过为了性能设置了只在一个numa节点上运行，可以频繁造成numa的不均衡，所以要复现的同学，
可以参考单numa节点上运行dpdk来复现，会概率大一些。
四、故障规避或解决
我们的解决方案是：
1.关闭numa的自动balance，echo 0 > /proc/sys/kernel/numa_balancing。或者直接在grub中增加numa=off
2.手工合入 linux社区的 0b26351b910f 补丁
3.这个补丁在centos的 3.10.0-974.el7 合入了:
[kernel] stop_machine, sched: Fix migrate_swap() vs. active_balance() deadlock (Phil Auld) [1557061]
同时红帽又反向合入到了3.10.0-957.27.2.el7.x86_64，所以把centos内核升级到 3.10.0-957.27.2.el7.x86_64也是
一种选择。
水平有限，如果有错误，请帮忙提醒我。如果您觉得本文对您有帮助，可以点击下面的 推荐 支持一下我。版权所有，需要转发请带上本文源地址，博客一直在更新，欢迎 关注 。
