linux 内核 xfs ail,linux 3.10内核 xfs的一次io异常导致的hung crash

最新推荐文章于 2024-08-16 09:53:20 发布

weixin_39931362

最新推荐文章于 2024-08-16 09:53:20 发布

阅读量776

点赞数

文章标签： linux 内核 xfs ail

这篇博客详细记录了在Linux 3.10内核中，XFS文件系统遇到一次IO异常，具体表现为xfs_log_force返回错误-5，进而触发了系统的hung crash。作者分析了umount任务被阻塞超过1200秒，发现问题是由于xfs_ail_push_all_sync函数中的调度导致的。在函数执行过程中，由于IO异常，xfsaild内核线程无法完成其任务，从而导致死锁。日志显示存在针对设备sdi的I/O错误，提示设备离线。博主分享了排查过程及相关的内核调用栈，提到了可能的解决方案链接。

摘要由CSDN通过智能技术生成

8556395.925003] XFS (sdi): xfs_log_force: error -5returned.

[8556407.425047] INFO: task umount:19286 blocked for more than 1200seconds.----这个默认是120，该环境上是被人为设置1200

[8556407.425653] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables thismessage.

[8556407.426050] umount D ffff881fffdd47c0 0 19286 19283 0x00000080------------------------19286是pid，19283是ppid

[8556407.426056] ffff883e39587d80 0000000000000082ffff883e39587fd8 ffff883e39587fd8

[8556407.426066] ffff883e39587fd8 00000000000147c0 ffff881fd3260000 ffff881f860fec80

[8556407.426072] ffff883ed79e1a20 ffff881f860fecc0 ffff881f860fece8 ffff881f860fec90------------这12个64位的值为栈里的值

[8556407.426078] Call Trace:

[8556407.426092] [] schedule+0x29/0x70[8556407.426137] [] xfs_ail_push_all_sync+0xc1/0x110[xfs]

[8556407.426143] [] ? wake_up_atomic_t+0x30/0x30[8556407.426165] [] xfs_unmountfs+0x68/0x190[xfs]

[8556407.426186] [] ? xfs_mru_cache_destroy+0x6b/0x90[xfs]

[8556407.426204] [] xfs_fs_put_super+0x32/0x90[xfs]

[8556407.426211] [] generic_shutdown_super+0x56/0xe0-------已经执行完fsnotify_umount_inodes[8556407.426215] [] kill_block_super+0x27/0x70[8556407.426218] [] deactivate_locked_super+0x49/0x60[8556407.426222] [] deactivate_super+0x46/0x60[8556407.426229] [] mntput_no_expire+0xc5/0x120[8556407.426234] [] SyS_umount+0x15b/0x5b0[8556407.426239] [] system_call_fastpath&