RHEL 8.8: migration_entry_wait_on_locked での Hang_task_timeout_secs

Issue

• RHEL 8.8 にアップグレードした後、いくつかのコマンドがハング状態になり、システム負荷の平均が非常に高くなります。
• ログには、バックトレースに migration_entry_wait_on_locked を伴ういくつかの Hang_task_timeout_secs が表示されます。

[150145.643958] INFO: task 2_scheduler:1618 blocked for more than 120 seconds.
[150145.644299]       Not tainted 4.18.0-477.10.1.el8_8.x86_64 #1
[150145.644510] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[150145.644723] task:2_scheduler     state:D stack:    0 pid: 1618 ppid:     1 flags:0x00000080
[150145.644728] Call Trace:
[150145.644733]  __schedule+0x2d1/0x870
[150145.644781]  schedule+0x55/0xf0
[150145.644784]  io_schedule+0x12/0x40
[150145.644787]  migration_entry_wait_on_locked+0x1ea/0x290
[150145.644794]  ? filemap_fdatawait_keep_errors+0x50/0x50
[150145.644797]  do_swap_page+0x5b0/0x710
[150145.644799]  ? pmd_devmap_trans_unstable+0x2e/0x40
[150145.644805]  ? handle_pte_fault+0x5d/0x880
[150145.644808]  __handle_mm_fault+0x453/0x6c0
[150145.644812]  handle_mm_fault+0xca/0x2a0
[150145.644814]  __do_page_fault+0x1f0/0x450
[150145.644827]  do_page_fault+0x37/0x130
[150145.644830]  ? page_fault+0x8/0x30
[150145.644837]  page_fault+0x1e/0x30