Post new topic   Reply to topic
View previous topic Printable version Log in to check your private messages View next topic
Author Message
snvvOffline
Post subject: EXT4 Data Corruption Bug Hits Stable Linux Kernels  PostPosted: 25.10.2012, 08:19



Joined: 2010-09-13
Posts: 287

Status: Offline
Hello
I found a scaring article in phoronix

http://www.phoronix.com/scan.php?page=n ... px=MTIxNDQ

Do we need to downgrade?

Thank you
snvv

UPDATE
https://plus.google.com/117091380454742 ... cc5tMiCgq7
 
 View user's profile Send private message  
Reply with quote Back to top
DonKultOffline
Post subject: RE: EXT4 Data Corruption Bug Hits Stable Linux Kernels  PostPosted: 25.10.2012, 10:03
Team Member


Joined: 2010-09-02
Posts: 481

Status: Offline
Quoting myself from another thread.
      DonKult wrote:
      manul wrote:
slh,
Please advise is this potentially ext4 corrupting bug going to be applied to next aptosid kernels? https://lkml.org/lkml/2012/10/23/690


If you happen to have a current system you will already have the patch:
      Code:
linux-aptosid (3.6-12) unstable; urgency=emergency

  * set ABI to 3.slh.2
  * jbd2: fix a potential fs corrupting bug in jbd2_mark_journal_empty
  * update ar5523.

 -- Stefan Lippers-Hollmann <s>  Wed, 24 Oct 2012 02:07:55 +0200


Note though that as far as I have read you need to have explicitly set your (first) superblock to be at 0 to be effected by this, so it isn't as serious as it might sound (and a "normal" user isn't going to to do enough 'frequent' mounts/umounts to trigger this I guess).


Ask your self: If it would be such a serious issue as the article suggests, how could it went undetected for 3 kernel releases (which means ~6 months). ext4 is not some small driver nobody uses but the default filesystem for nearly all distros nowadays… (not to say you shouldn't upgrade, but just because Halloween is near we don't need to be scared of everything)


As an off-topic remark, I wonder what Jorge Castro (the canonical employee with the gun in the image) has to do with ext4 bughunting, but i shouldn't question the high-quality articles of phoronix I guess.

_________________
MfG. DonKult
"I never make stupid mistakes. Only very, very clever ones." ~ The Doctor
 
 View user's profile Send private message Visit poster's website  
Reply with quote Back to top
snvvOffline
Post subject: Re: RE: EXT4 Data Corruption Bug Hits Stable Linux Kernels  PostPosted: 25.10.2012, 10:45



Joined: 2010-09-13
Posts: 287

Status: Offline
Hello,
I agree with what you just said.
My updated link shows that it is not a real problem.

However, I you check various sites (i.e. gentoo, arch etc) there are so many posts for the "problem" and patches to an non-existen problem Smile

Regards
snvv
 
 View user's profile Send private message  
Reply with quote Back to top
LatOffline
Post subject: RE: Re: RE: EXT4 Data Corruption Bug Hits Stable Linux Kerne  PostPosted: 27.10.2012, 02:56



Joined: 2010-09-19
Posts: 159

Status: Offline
Probably I've gotten sucked in into the panic wave, but if the urgency is emergency, why there's no advisory or headlight in the news section of this website?

Also, besides of individuals using ext4, what other criteria are affected? individuals with kernels newer than 3.2.x?

Thank you developers for your time
 
 View user's profile Send private message  
Reply with quote Back to top
DonKultOffline
Post subject: RE: Re: RE: EXT4 Data Corruption Bug Hits Stable Linux Kerne  PostPosted: 27.10.2012, 21:34
Team Member


Joined: 2010-09-02
Posts: 481

Status: Offline
This level is indeed high and not that often used, but slh has some trackrecord on using it even for issues which only apply to a fraction of the userbase. Last time e.g. here:
      Quote:
* import 2.6.27.1, to prevent EEPROM corruption due to dynamic ftrace
running amok and using cmpxchg on ioremapped memory.
All of our (inofficially) released kernels had FTRACE disabled alltogether
and have therefore been safe against this kind of hardware corruption.

So even if it is high, it doesn't need to effect you, or many, or anyone at all. Kind of like a "FTBFS on sparc" RC-bug in Debian: Bad and should be fixed, but not more than a notice if you are a amd64 user …

We had frontpage news for serious issues effecting our users, but this is not a serious bug and I wonder if we can find a single aptosid user effected by this given that for now there is only one linux-kernel user known to be effected.


Anyway, the issue itself was in the meantime downgraded from "unlikely" to "nearly impossible" by Ts'o (ext4 maintainer). Many news sites got that right on first try (e.g. heise "jumped" on the train relatively late with mentioning explicit that only users with exotic mountoptions could be effected), many others unfortunately not. Lets see if they will correct it.
Ts'o posted an update in the meantime (he consistently posted comments/wrote mails regarding the issue): https://plus.google.com/u/0/11709138045 ... 5a1eHxUzSh

_________________
MfG. DonKult
"I never make stupid mistakes. Only very, very clever ones." ~ The Doctor
 
 View user's profile Send private message Visit poster's website  
Reply with quote Back to top
LatOffline
Post subject: RE: Re: RE: EXT4 Data Corruption Bug Hits Stable Linux Kerne  PostPosted: 27.10.2012, 22:10



Joined: 2010-09-19
Posts: 159

Status: Offline
Thanks for the update, DonKult. Very informative as usual!
 
 View user's profile Send private message  
Reply with quote Back to top
Display posts from previous:     
Jump to:  
All times are GMT - 12 Hours
Post new topic   Reply to topic
View previous topic Printable version Log in to check your private messages View next topic
Powered by Zafenio