UBI/UBIFS corruptions during random power-cuts

[Richard Weinberger]

Am 15.02.2017 um 16:06 schrieb Bhuvanchandra DV:
> On 02/15/2017 02:42 PM, Richard Weinberger wrote:
> 
>> Am 15.02.2017 um 08:22 schrieb Bhuvanchandra DV:
>>> I was wrong, systemd version which we use lately is enabled with access control lists(ACL).
>>> ...
>>> [    3.324846] systemd[1]: systemd 230 running in system mode. (+PAM -AUDIT -SELINUX +IMA -APPARMOR +SMACK +SYSVINIT +UTMP -LIBCRYPTSETUP -GCRYPT -GNUTLS +ACL +XZ -LZ4 -SECCOMP
>>> +BLKI)
>>> ...
>>>
>>> Will do the test with a systemd without ACL support.
>> Well, do files have xattrs?
>> See getfattr.
>> I'm trying to understand why your system makes so different.
> 
> systemd's journald uses ACL, the journald's logs files seems to have extended attributes.
> 
> root at colibri-vf:~# getfattr -R /var/log/journal/3e283b492ea046a29683c0801ceaba91
> getfattr: Removing leading '/' from absolute path names
> # file: var/log/journal/3e283b492ea046a29683c0801ceaba91/system at 000547ec00621f43-a0f5cd8e864efc36.journal~
> user.crtime_usec
> 
> # file: var/log/journal/3e283b492ea046a29683c0801ceaba91/system at 000547ec25f38147-3ede62d61071df3c.journal~
> user.crtime_usec
> 
> # file: var/log/journal/3e283b492ea046a29683c0801ceaba91/system at 0820e4037f504446ad35bb362be7b809-0000000000000001-l
> user.crtime_usec
> 
> # file: var/log/journal/3e283b492ea046a29683c0801ceaba91/system.journal
> user.crtime_usec
> 
> # file: var/log/journal/3e283b492ea046a29683c0801ceaba91/system at 37549ee306da42479500fb717cb2dada-0000000000000001-l
> user.crtime_usec
> 
> # file: var/log/journal/3e283b492ea046a29683c0801ceaba91/system at 000547ecafe0f654-e0c2574255c05aed.journal~
> user.crtime_usec
> 
> # file: var/log/journal/3e283b492ea046a29683c0801ceaba91/system at 000547ec78bce8a1-02b48f08c5d6a410.journal~
> user.crtime_usec
> 
> # file: var/log/journal/3e283b492ea046a29683c0801ceaba91/system at 0005488c6421da38-c4690dc0bb71068f.journal~
> user.crtime_usec

Hmm, not sure whether systemd journal triggers such a bug in UBIFS.
Anyway, can you provide me a UBIFS image after a failure?
A dd from the UBI volume (i.e. /dev/ubiX_Y) will do it.

Maybe it can give me some hint.

Thanks,
//richard