Re: ubifs_check_node: bad CRC

[Mickael Chazaux <mickael.chazaux@xxxxxxxxxxxxxxx>]

On 13/08/2019 23:41, Richard Weinberger wrote:

> 
>> The problem occured in the field on several devices, and I cannot reproduce it
>> in the lab.
> 
> It sounds a bit like a memory corruption.
> 
>>> Please also disable compression in UBIFS, then the read back data maybe gives
>>> you a hint what went wrong.
>>
>> In the corrupted node I see data from syslog. This system is a bit different
>> than the others we usually make, as there are more power cuts and more log writes.
> 
> Hmm. Power cuts or many writes should not trigger such a corruption.
> I'd expect ECC errors or data structure errors in UBIFS. But not a good ECC
> with bad UBIFS data CRC.
> 
> But with only one sample it is almost impossible to figure...
> 

Memory corruption as in some kind of buffer overflow or use after free ? or physical 
bus errors? Anyway I think I will implement some mitigations at the application level, 
(eg. less log messages, recovery by creating a new volume, copy the now-readonly data, 
rename the volume) because the haystack just got too big for the needle.

Thank you for your help,

Mickael

-- 
Mickael Chazaux
Software Engineer
Tel : (33) 476 042 006
Fax : (33) 476 042 001

ETIC TELECOM
13, Chemin du Vieux Chêne
38240 MEYLAN
Tel: 33 4 76 04 20 00
fax : 33 4 76 04 20 01

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/