Comment 9 for bug 1852122

Revision history for this message
Rafael David Tinoco (rafaeldtinoco) wrote :

The upstream fix is likely this:

From b73eba2a867e10b9b4477738677341f3307c07bb Mon Sep 17 00:00:00 2001
From: Gang He <email address hidden>
Date: Sat, 4 Jan 2020 13:00:22 -0800
Subject: [PATCH] ocfs2: fix the crash due to call ocfs2_get_dlm_debug once
 less

Because ocfs2_get_dlm_debug() function is called once less here, ocfs2
file system will trigger the system crash, usually after ocfs2 file
system is unmounted.

This system crash is caused by a generic memory corruption, these crash
backtraces are not always the same, for exapmle,

    ocfs2: Unmounting device (253,16) on (node 172167785)
    general protection fault: 0000 [#1] SMP PTI
    CPU: 3 PID: 14107 Comm: fence_legacy Kdump:
    Hardware name: QEMU Standard PC (i440FX + PIIX, 1996)
    RIP: 0010:__kmalloc+0xa5/0x2a0
    Code: 00 00 4d 8b 07 65 4d 8b
    RSP: 0018:ffffaa1fc094bbe8 EFLAGS: 00010286
    RAX: 0000000000000000 RBX: d310a8800d7a3faf RCX: 0000000000000000
    RDX: 0000000000000000 RSI: 0000000000000dc0 RDI: ffff96e68fc036c0
    RBP: d310a8800d7a3faf R08: ffff96e6ffdb10a0 R09: 00000000752e7079
    R10: 000000000001c513 R11: 0000000004091041 R12: 0000000000000dc0
    R13: 0000000000000039 R14: ffff96e68fc036c0 R15: ffff96e68fc036c0
    FS: 00007f699dfba540(0000) GS:ffff96e6ffd80000(0000) knlGS:00000
    CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
    CR2: 000055f3a9d9b768 CR3: 000000002cd1c000 CR4: 00000000000006e0
    Call Trace:
     ext4_htree_store_dirent+0x35/0x100 [ext4]
     htree_dirblock_to_tree+0xea/0x290 [ext4]
     ext4_htree_fill_tree+0x1c1/0x2d0 [ext4]
     ext4_readdir+0x67c/0x9d0 [ext4]
     iterate_dir+0x8d/0x1a0
     __x64_sys_getdents+0xab/0x130
     do_syscall_64+0x60/0x1f0
     entry_SYSCALL_64_after_hwframe+0x49/0xbe
    RIP: 0033:0x7f699d33a9fb

This regression problem was introduced by commit e581595ea29c ("ocfs: no
need to check return value of debugfs_create functions").

Link: http://<email address hidden>
Fixes: e581595ea29c ("ocfs: no need to check return value of debugfs_create functions")
Signed-off-by: Gang He <email address hidden>
Acked-by: Joseph Qi <email address hidden>
Cc: Mark Fasheh <email address hidden>
Cc: Joel Becker <email address hidden>
Cc: Junxiao Bi <email address hidden>
Cc: Changwei Ge <email address hidden>
Cc: Gang He <email address hidden>
Cc: Jun Piao <email address hidden>
Cc: <email address hidden> [5.3+]
Signed-off-by: Andrew Morton <email address hidden>
Signed-off-by: Linus Torvalds <email address hidden>

as reported in upstream bug. Giving it a try to finally suggest as a SRU to the kernel team.