Alibaba Cloud Linux 2系统的ECS实例中Page Fault异常导致系统宕机

问题描述

在符合如下条件的Alibaba Cloud Linux 2实例中，系统运行时出现系统宕机问题。

镜像：Alibaba Cloud Linux 2.1903 LTS 64位。
内核：kernel-4.19.91-23.al7及之前的内核版本。

系统宕机，且出现如下调用栈信息。

[  332.057218] watchdog: BUG: soft lockup - CPU#7 stuck for 11s! [split_v2:28356]
[  332.057219]  mousedev isst_if_common hid_generic usbhid
[  332.057223] CPU: 3 PID: 28336 Comm: split_v2 Kdump: loaded Not tainted 4.19.91-19.1.al7.x86_64 #1
[  332.057507] Kernel panic - not syncing: softlockup: hung tasks
[  332.057508] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[  332.057510] CPU: 6 PID: 28355 Comm: split_v2 Kdump: loaded Tainted: G             L    4.19.91-19.1.al7.x86_64 #1
[  332.057513]  cp_new_stat+0x13d/0x160
[  332.057514] RDX: 000000c000100000 RSI: 000000c000100000 RDI: 0000000000000019
[  332.057515] Call Trace:
[  332.057516] Hardware name: Alibaba Cloud Alibaba Cloud ECS, BIOS 8a46cfe 04/01/2014
[  332.057518]  __se_sys_newfstat+0x2e/0x40
[  332.057518] Call Trace:
[  332.057519] Code: 89 d1 c1 e9 03 83 e2 07 f3 48 a5 89 d1 f3 a4 31 c0 0f 01 ca c3 0f 1f 80 00 00 00 00 0f 01 cb 83 fa 40 0f 82 70 ff ff ff 89 d1 <f3> a4 31 c0 0f 01 ca c3 66 2e 0f 1f 84 00 00 00 00 00 0f 01 cb 83
[  332.057521] RBP: 00007eff1201bf10 R08: 00007eff1201c700 R09: 00007eff1201c700
[  332.057523]  do_syscall_64+0x5b/0x1b0
[  332.057524]  <IRQ>
[  332.057525] RSP: 0018:ffffa389886efde8 EFLAGS: 00050206
[  332.057529]  dump_stack+0x66/0x8b
[  332.057531] R10: 00007eff1201c9d0 R11: 0000000000000246 R12: 0000000000000000
[  332.057534]  panic+0xd8/0x24c
[  332.057535] RAX: 000000c000100090 RBX: ffffa389886efea8 RCX: 0000000000000090
[  332.057536] R13: 0000000000801000 R14: 0000000000000000 R15: 00007eff1201c700
[  332.057539]  __do_page_fault+0x11d/0x470
[  332.057540]  ? 0xffffffffc0477000
[  332.057541] RDX: 0000000000000090 RSI: ffffa389886efdf8 RDI: 000000c000100000
[  332.057552]  watchdog_timer_fn+0x253/0x260
[  332.057555]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[  332.057556]  ? softlockup_fn+0x40/0x40
[  332.057557] RBP: 000000c000100000 R08: 0000000000000000 R09: 0000000000000000
[  332.057559]  __hrtimer_run_queues+0xeb/0x250
[  332.057560] R10: ffff8bfb1690a310 R11: ffff8bfb1f01a6c8 R12: ffff8bfaee04df00
[  332.057562]  hrtimer_interrupt+0x122/0x270
[  332.057563] RIP: 0033:0x7eff1b11e3a4
[  332.057564] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
[  332.057566]  smp_apic_timer_interrupt+0x6a/0x140
[  332.057568]  do_page_fault+0x32/0x140
[  332.057570]  apic_timer_interrupt+0xf/0x20
[  332.057572]  _copy_to_user+0x22/0x30
[  332.057573] Code: 00 f7 d8 64 89 02 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 83 ff 01 89 f0 77 19 48 63 f8 48 89 d6 b8 05 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 18 f3 c3 66 90 48 8b 05 99 7a 2d 00 64 c7 00
[  332.057574]  </IRQ>
[  332.057575] RSP: 002b:00007eff1181aed8 EFLAGS: 00000246
[  332.057578] RIP: 0010:__do_page_fault+0x227/0x470
[  332.057579]  ORIG_RAX: 0000000000000005
[  332.057580] Code: 00 48 83 c4 30 5b 5d 41 5c 41 5d 41 5e 41 5f c3 f6 85 91 00 00 00 02 41 bf 14 00 00 00 0f 84 c5 fe ff ff fb 66 0f 1f 44 00 00 <e9> b9 fe ff ff f6 85 88 00 00 00 03 75 0d f6 85 92 00 00 00 04 0f
[  332.057582]  cp_new_stat+0x13d/0x160
[  332.057583] RSP: 0018:ffffa389886f7ca0 EFLAGS: 00000202 ORIG_RAX: ffffffffffffff13
[  332.057585]  __se_sys_newfstat+0x2e/0x40
[  332.057586] RAX: 0000000000000000 RBX: 0000000000000002 RCX: ffffffff93a00ae0
[  332.057587] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007eff1b11e3a4
[  332.057588] RDX: 0000000000000000 RSI: 0000000000000002 RDI: ffffa389886f7d38
[  332.057589]  do_syscall_64+0x5b/0x1b0
[  332.057590] RBP: ffffa389886f7d38 R08: 0000000000000000 R09: 0000000000000000
[  332.057591] RDX: 000000c000100000 RSI: 000000c000100000 RDI: 0000000000000009
[  332.057592] R10: 0000000000000000 R11: 0000000000000000 R12: 000000c000100000
[  332.057594]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[  332.057595] R13: ffff8bfb168bd940 R14: ffff8bfaee04af80 R15: 0000000000000014
[  332.057597] RIP: 0033:0x7eff1b11e3a4
[  332.057599]  async_page_fault+0x1e/0x30
[  332.057601]  ? restore_regs_and_return_to_kernel+0x25/0x25
[  332.057602] RBP: 00007eff1181af10 R08: 00007eff1181b700 R09: 00007eff1181b700
[  332.057602] Code: 00 f7 d8 64 89 02 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 83 ff 01 89 f0 77 19 48 63 f8 48 89 d6 b8 05 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 18 f3 c3 66 90 48 8b 05 99 7a 2d 00 64 c7 00
[  332.057604]  do_page_fault+0x32/0x140
[  332.057606] RIP: 0010:copy_user_enhanced_fast_string+0xe/0x20
[  332.057607] R10: 00007eff1181b9d0 R11: 0000000000000246 R12: 0000000000000000
[  332.057608] Code: 89 d1 c1 e9 03 83 e2 07 f3 48 a5 89 d1 f3 a4 31 c0 0f 01 ca c3 0f 1f 80 00 00 00 00 0f 01 cb 83 fa 40 0f 82 70 ff ff ff 89 d1 <f3> a4 31 c0 0f 01 ca c3 66 2e 0f 1f 84 00 00 00 00 00 0f 01 cb 83
[  332.057609]  async_page_fault+0x1e/0x30
[  332.057610] R13: 0000000000801000 R14: 0000000000000000 R15: 00007eff1181b700
[  332.057612] RIP: 0010:copy_user_enhanced_fast_string+0xe/0x20
[  332.057613] RSP: 002b:00007eff08808ed8 EFLAGS: 00000246
[  332.057614] Code: 89 d1 c1 e9 03 83 e2 07 f3 48 a5 89 d1 f3 a4 31 c0 0f 01 ca c3 0f 1f 80 00 00 00 00 0f 01 cb 83 fa 40 0f 82 70 ff ff ff 89 d1 <f3> a4 31 c0 0f 01 ca c3 66 2e 0f 1f 84 00 00 00 00 00 0f 01 cb 83
[  332.057615]  ORIG_RAX: 0000000000000005
[  332.057616] RSP: 0018:ffffa389886f7de8 EFLAGS: 00050206
[  332.057617] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007eff1b11e3a4
[  332.057618] RAX: 000000c000100090 RBX: ffffa389886f7ea8 RCX: 0000000000000090
[  332.057619] RDX: 000000c000100000 RSI: 000000c000100000 RDI: 0000000000000024
[  332.057620] RDX: 0000000000000090 RSI: ffffa389886f7df8 RDI: 000000c000100000
[  332.057621] RSP: 0018:ffffa389886ffde8 EFLAGS: 00050206
[  332.057623] RBP: 000000c000100000 R08: 0000000000000000 R09: 0000000000000000
[  332.057624] RBP: 00007eff08808f10 R08: 00007eff08809700 R09: 00007eff08809700
[  332.057625] R10: ffff8bfb1690b810 R11: ffff8bfb1f01a6c8 R12: ffff8bfaee04af80
[  332.057626] R10: 00007eff088099d0 R11: 0000000000000246 R12: 0000000000000000
[  332.057627] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
[  332.057628] R13: 0000000000801000 R14: 0000000000000000 R15: 00007eff08809700
[  332.057630]  _copy_to_user+0x22/0x30
[  332.057631] RAX: 000000c000100090 RBX: ffffa389886ffea8 RCX: 0000000000000090
[  332.057632]  cp_new_stat+0x13d/0x160
[  332.057633] RDX: 0000000000000090 RSI: ffffa389886ffdf8 RDI: 000000c000100000
[  332.057634] RBP: 000000c000100000 R08: 0000000000000000 R09: 0000000000000000
[  332.057635]  __se_sys_newfstat+0x2e/0x40
[  332.057636] R10: ffff8bfb1690ad10 R11: ffff8bfb1f01a6c8 R12: ffff8bfaee048000
[  332.057637]  do_syscall_64+0x5b/0x1b0
[  332.057638] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
[  332.057640]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[  332.057642]  _copy_to_user+0x22/0x30
[  332.057643] RIP: 0033:0x7eff1b11e3a4
[  332.057645]  cp_new_stat+0x13d/0x160
[  332.057646] Code: 00 f7 d8 64 89 02 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 83 ff 01 89 f0 77 19 48 63 f8 48 89 d6 b8 05 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 18 f3 c3 66 90 48 8b 05 99 7a 2d 00 64 c7 00
[  332.057647]  __se_sys_newfstat+0x2e/0x40
[  332.057648] RSP: 002b:00007eff08007ed8 EFLAGS: 00000246 ORIG_RAX: 0000000000000005
[  332.057651]  do_syscall_64+0x5b/0x1b0
[  332.057652] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007eff1b11e3a4
[  332.057654]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[  332.057655] RDX: 000000c000100000 RSI: 000000c000100000 RDI: 000000000000002e
[  332.057656] RIP: 0033:0x7eff1b11e3a4
[  332.057657] RBP: 00007eff08007f10 R08: 00007eff08008700 R09: 00007eff08008700
[  332.057658] Code: 00 f7 d8 64 89 02 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 83 ff 01 89 f0 77 19 48 63 f8 48 89 d6 b8 05 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 18 f3 c3 66 90 48 8b 05 99 7a 2d 00 64 c7 00
[  332.057659] R10: 00007eff080089d0 R11: 0000000000000246 R12: 0000000000000000
[  332.057660] RSP: 002b:00007eff07806ed8 EFLAGS: 00000246 ORIG_RAX: 0000000000000005
[  332.057662] R13: 0000000000801000 R14: 0000000000000000 R15: 00007eff08008700
[  332.057663] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007eff1b11e3a4
[  332.057663] RDX: 000000c000100000 RSI: 000000c000100000 RDI: 000000000000001e
[  332.057664] RBP: 00007eff07806f10 R08: 00007eff07807700 R09: 00007eff07807700
[  332.057665] R10: 00007eff078079d0 R11: 0000000000000246 R12: 0000000000000000
[  332.057665] R13: 0000000000801000 R14: 0000000000000000 R15: 00007eff07807700

问题原因

Alibaba Cloud Linux系统默认开启了THP（透明大页功能），GC内存回收时，会调用MADV_NOHUGEPAGE关闭大页，然后调用MADV_FREE释放部分4K页，并在操作系统中会切割THP大页。当其它进程内核Page Fault异常占用CPU资源时，导致切割THP大页的进程没有调度完成，而切割THP大页的进程无法完成，会导致Page Fault的进程一直无法结束，因此它们会一直等待对方结束进程，最终会导致SOFT LOCKUP。若Alibaba Cloud Linux实例中配置了/proc/sys/kernel/softlockup_panic，SOFT LOCKUP的产生会触发内核宕机。

解决方案

阿里云提醒您：

如果您对实例或数据有修改、变更等风险操作，务必注意实例的容灾、容错能力，确保数据安全。

如果您对实例（包括但不限于ECS、RDS）等进行配置与数据修改，建议提前创建快照或开启RDS日志备份等功能。

如果您在阿里云平台授权或者提交过登录账号、密码等安全信息，建议您及时修改。

当遇到该问题时，您可以参考以下方案处理：

登录ECS实例，详情请参见连接方式概述。
执行以下命令，确认系统内核版本适用此方案。
```
uname -r
```
系统显示类似如下。
```
4.19.91-19.1.al7.x86_64
```
根据系统内核版本选择对应的解决方法：
- 对于4.19.91-19.1.al7.x86_64（不含）之前的版本：
  1. 执行以下命令，将操作系统版本更新至最新的内核版本。
```
yum update kernel
```
  2. 更新内核版本之后，需重启生效，请执行以下命令，重启服务器。
```
reboot
```
  3. 若最新内核版本的操作系统同样存在该问题，请执行以下步骤，更新内核热补丁。
- 对于4.19.91-19.1.al7.x86_64（包含）到4.19.91-23.al7.x86_64（包含）之间的版本，可通过安装内核热补丁解决，安装命令如下。
```
yum install -y kernel-hotfix-5902278-`uname -r | awk -F"-" '{print $NF}'`
```

适用于

云服务器ECS

如果您的问题仍未解决，您可以在阿里云社区免费咨询，或提交工单联系阿里云技术支持。