umount xfs文件系统时hang住

1.使用附件的tmp.c编译，tmp.img挂载到tmp_dir
2.执行./tmp tmp_dir
3.umount tmp_dir时长时间卡住（超过24h），内核会有hung_task告警：
[ 1571.817116] umount D 0 2647 2237 0x00000208
[ 1571.817845] Call trace:
[ 1571.818245] __switch_to+0xe4/0x148
[ 1571.818742] __schedule+0x28c/0x938
[ 1571.819228] schedule+0x2c/0x88
[ 1571.819684] wb_wait_for_completion+0x94/0xc0
[ 1571.820283] __writeback_inodes_sb_nr+0xdc/0x100
[ 1571.820897] writeback_inodes_sb+0x34/0x40
[ 1571.821513] sync_filesystem+0x44/0xc0
[ 1571.822024] generic_shutdown_super+0x34/0x128
[ 1571.822673] kill_block_super+0x2c/0x70
[ 1571.823232] deactivate_locked_super+0x64/0x90
[ 1571.823932] deactivate_super+0x74/0x80
[ 1571.824563] cleanup_mnt+0x44/0xa0
[ 1571.825042] __cleanup_mnt+0x20/0x30
[ 1571.825528] task_work_run+0xbc/0xf0
[ 1571.825995] do_notify_resume+0x12c/0x138
[ 1571.826586] work_pending+0x8/0x10

Hey @ltx, Welcome to openEuler Community.
All of the projects in openEuler Community are maintained by @openeuler-ci-bot.
That means the developpers can comment below every pull request or issue to trigger Bot Commands.
Please follow instructions at https://gitee.com/openeuler/community/blob/master/en/command.md to find the details.

问题原因：

这个问题初步分析是由于xfs文件系统元数据区域中sb中空闲块个数和分配组ag内部free block个数不配置，导致在xfs进行writeback的过程中，长时间占有sb->umount信号量
在umount xfs文件系统的时候，由于需要等待sb->umount，进程切入到D状态时间超过hung_task_timeout，系统发生hung task报警

解决方案：

已经针对该问题和内核社区进行沟通，补充元数据校验。保证在元数据出现异常的情况下，不继续参与流程处理(具体信息参见邮件列表https://patchwork.kernel.org/patch/11395543/)

GVP openEuler / kernel

内容风险标识

评论 (2)

GVPopenEuler / kernel

内容风险标识