-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Package: e2fsprogs
Version: 1.44.4-2

>From time to time, some processes get stuck and I cannot kill them. Even 
>SIGKILL does not work. The only way is to reboot (which I usually cannot 
>afford because I have ZPAQ compression processes that last days).
When I view them in "htop", they are in "D" state, with the wait channel being 
"call_rwsem_down_read_failed".
Like this: https://i.ibb.co/Ry86CJ2/2018-11-28-110558-1600x1200-scrot.png
It happens randomly, and "dmesg" does not show anything unusual, no hardware 
problems, no NFS involved.
I tried to trace the problem with SysRq+W:
===BEGIN OUTPUT===
[40732.001687] sysrq: SysRq : Show Blocked State
[40732.001691]   task                        PC stack   pid father
[40732.001751] dpkg-preconfigu D    0 19836      1 0x00000004
[40732.001753] Call Trace:
[40732.001760]  ? __schedule+0x2a2/0x870
[40732.001761]  schedule+0x28/0x80
[40732.001763]  rwsem_down_read_failed+0x10f/0x180
[40732.001765]  call_rwsem_down_read_failed+0x14/0x30
[40732.001767]  down_read+0x1c/0x30
[40732.001783]  ext4_find_inline_entry+0x4c/0x160 [ext4]
[40732.001794]  ext4_find_entry+0x47f/0x540 [ext4]
[40732.001798]  ? d_alloc_parallel+0x99/0x4a0
[40732.001806]  ext4_lookup+0x6f/0x200 [ext4]
[40732.001809]  __lookup_slow+0x97/0x150
[40732.001810]  lookup_slow+0x35/0x50
[40732.001811]  walk_component+0x1bf/0x4a0
[40732.001813]  path_lookupat.isra.48+0x6d/0x220
[40732.001815]  filename_lookup.part.62+0xa0/0x170
[40732.001817]  ? __check_object_size+0xa3/0x181
[40732.001819]  ? strncpy_from_user+0x4a/0x170
[40732.001821]  vfs_statx+0x73/0xe0
[40732.001822]  __do_sys_newstat+0x39/0x70
[40732.001824]  ? handle_mm_fault+0xda/0x200
[40732.001826]  ? __do_page_fault+0x26c/0x4f0
[40732.001828]  do_syscall_64+0x53/0x100
[40732.001830]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[40732.001831] RIP: 0033:0x7fa7bb720dd5
[40732.001834] Code: Bad RIP value.
[40732.001834] RSP: 002b:00007ffd327671a8 EFLAGS: 00000246 ORIG_RAX: 
0000000000000004
[40732.001836] RAX: ffffffffffffffda RBX: 0000558e2745e260 RCX: 00007fa7bb720dd5
[40732.001836] RDX: 0000558e2745e478 RSI: 0000558e2745e478 RDI: 0000558e277a3430
[40732.001836] RBP: 0000558e278a4ae0 R08: 0000000000000000 R09: 0000000000000033
[40732.001837] R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
[40732.001837] R13: 0000558e266f1198 R14: 0000558e277a3430 R15: 0000558e276ae340
===END OUTPUT===
I am no programmer, but since it shows "ext4" and "inline", I think there is a 
problem with EXT4, especially with its "inline_data" feature.
This is my current configuration using "tune2fs -l":
===BEGIN OUTPUT===
tune2fs 1.44.4 (18-Aug-2018)
Filesystem volume name:   DEBIAN
Last mounted on:          /root
Filesystem UUID:          c5971091-77d2-40fe-9095-54bb0ff96a2c
Filesystem magic number:  0xEF53
Filesystem revision #:    1 (dynamic)
Filesystem features:      has_journal ext_attr dir_index sparse_super2 
needs_recovery extent flex_bg inline_data large_file uninit_bg
Filesystem flags:         signed_directory_hash
Default mount options:    discard
Filesystem state:         clean
Errors behavior:          Remount read-only
Filesystem OS type:       Linux
Inode count:              228960
Block count:              117220824
Reserved block count:     0
Free blocks:              102491793
Free inodes:              107444
First block:              1
Block size:               1024
Fragment size:            1024
Blocks per group:         8192
Fragments per group:      8192
Inodes per group:         16
Inode blocks per group:   4
Flex block group size:    16
Filesystem created:       Wed Oct 17 19:58:28 2018
Last mount time:          Fri Nov 30 08:32:36 2018
Last write time:          Fri Nov 30 08:32:36 2018
Mount count:              9
Maximum mount count:      -1
Last checked:             Wed Nov 28 17:01:44 2018
Check interval:           0 (<none>)
Lifetime writes:          437 GB
Reserved blocks uid:      0 (user root)
Reserved blocks gid:      0 (group root)
First inode:              11
Inode size:               256
Required extra isize:     32
Desired extra isize:      32
Journal inode:            8
First orphan inode:       168961
Default directory hash:   tea
Directory Hash Seed:      a4bd0a95-469a-48ff-8d9e-c1a3e4674ec8
Journal backup:           inode blocks
===END OUTPUT===
--
"And in the naked light, I saw ten thousand people, maybe m\
ore. People talking without speaking, people hearing withou\
t listening. People writing songs that voices never shared,\
no one dared disturb the sound of silence."

-----BEGIN PGP SIGNATURE-----
Version: ProtonMail
Comment: https://protonmail.com

wsBcBAEBCAAQBQJcASmeCRDYtWA5RV10HwAAte4H/ixkZtYgCuWCfyCg9Mdf
VDKnvNSxCjsSZ23fA1BxEuf2DkBHCyvHMw9qIAugO0jlkePeWLwJdR15jF93
Tbf9o5fT8+BHtnyEfXi2uK1BnOAOlIsR21SHc/r4qgXQLBd0AftCWbfxHNIP
2t+2mRzhj2TjUjT9vjg9NlnOVU2HkdmaCOrSYwfbb1WmP/oeLybFd/pstU46
CuWS8twRwIBxZUw/UzhXHP5YT+F4gCO+rIzhvzziiJwhpI+G8Y18/bcWBw6l
6drE3l3+GSXrsdXeoaf9yPEEGrwlzvh2urTkz7GE0+lu0hBT5fi3qjZgeIF7
Aci4M8YemcoSbvGcBd6/6tI=
=x95p
-----END PGP SIGNATURE-----

Attachment: publickey - pthfdr@protonmail.ch - 0xAB77ABA4.asc
Description: application/pgp-keys

Attachment: publickey - pthfdr@protonmail.ch - 0xAB77ABA4.asc.sig
Description: PGP signature

Reply via email to