UBIFS - ubifs_get_pnode.part.4: error -22 reading pnode

[Erdrich, Frank]

Hello,

we are encountering an error on UBIFS that prevents mounting of a partition. Maybe one of you can tell me the directly the reason for that or can help me hunting this error down.

Some data of the system:

- Linux Kernel version 4.9 (unmodified)
- SAMA5D36
- Flash: Micron 29F8G0ABABA (PEB size of 512k)
- 256 MB of RAM

The error seems to occur after some power-cycles and is therefore hard to reproduce. We are seeing it also on only some of the systems.
The error looks like the following:

UBIFS (ubi2:0): background thread "ubifs_bgt2_0" started, PID 867 UBIFS (ubi2:0): recovery needed UBIFS error (ubi2:0 pid 865): ubifs_get_pnode.part.4: error -22 reading pnode at 6:188460 (pid 865) dumping pnode:
	address cd9bb7c0 parent cd9bb700 cnext 0
	flags 0 iip 0 level 0 num 0
	0: free 0 dirty 501576 flags 1 lnum 0
	1: free 0 dirty 254864 flags 1 lnum 0
	2: free 0 dirty 515672 flags 1 lnum 0
	3: free 507904 dirty 524128 flags 34 lnum 0
CPU: 0 PID: 865 Comm: mount Not tainted 4.9.51 #124 Hardware name: Atmel SAMA5 [<c0110570>] (unwind_backtrace) from [<c010d488>] (show_stack+0x20/0x24) [<c010d488>] (show_stack) from [<c0390c7c>] (dump_stack+0x24/0x28) [<c0390c7c>] (dump_stack) from [<c0319e54>] (ubifs_get_pnode.part.4+0x250/0x2b8)
[<c0319e54>] (ubifs_get_pnode.part.4) from [<c031bf08>] (ubifs_lpt_lookup_dirty+0x2f8/0x340)
[<c031bf08>] (ubifs_lpt_lookup_dirty) from [<c031f910>] (ubifs_update_one_lp+0x44/0x150) [<c031f910>] (ubifs_update_one_lp) from [<c0309540>] (ubifs_tnc_add+0x170/0x1ec) [<c0309540>] (ubifs_tnc_add) from [<c030d638>] (ubifs_replay_journal+0xfcc/0x1670)
[<c030d638>] (ubifs_replay_journal) from [<c0300380>] (ubifs_mount+0x1194/0x1e8c) [<c0300380>] (ubifs_mount) from [<c0209ea0>] (mount_fs+0x54/0x170) [<c0209ea0>] (mount_fs) from [<c0228dec>] (vfs_kern_mount+0x58/0x124) [<c0228dec>] (vfs_kern_mount) from [<c022c6e0>] (do_mount+0x1bc/0xcc0) [<c022c6e0>] (do_mount) from [<c022d548>] (SyS_mount+0x84/0xac) [<c022d548>] (SyS_mount) from [<c0109380>] (ret_fast_syscall+0x0/0x3c) UBIFS error (ubi2:0 pid 865): ubifs_get_pnode.part.4: calc num: 4 UBIFS error (ubi2:0 pid 865): ubifs_update_one_lp: cannot update properties of LEB 29, error -22 UBIFS (ubi2:0): background thread "ubifs_bgt2_0" stops

Output of ubinfo:

Volumes count:                           1
Logical eraseblock size:                 516096 bytes, 504.0 KiB
Total amount of logical eraseblocks:     140 (72253440 bytes, 68.9 MiB)
Amount of available logical eraseblocks: 0 (0 bytes)
Maximum count of volumes                 128
Count of bad physical eraseblocks:       0
Count of reserved physical eraseblocks:  40
Current maximum erase counter value:     11
Minimum input/output unit size:          4096 bytes
Character device major/minor:            247:0
Present volumes:                         0


I'm not deep enough in the ubifs system to completely understand what is happening here but for me it seems that there are more dirty data to write than the size of the LEB.
-> 3: free 507904 dirty 524128 flags 34 lnum 0
I've seen that the values are checked in validate_pnode() where the -EINVAL (-22) comes from. The question is, how can dirty data bigger that the LEB size?

Does someone have a suggestion what is happening here? Or do you need more information to rate that error?


Best Regards,
Frank Erdrich