Re: [gentoo-user] Heads-up: Several kernel versions have severe EXT4 data corruption bug

[Paul Hartman <paul.hartman+gentoo@gmail.com>]

On Fri, Nov 2, 2012 at 9:55 AM, Dale <rdalek1967@gmail.com> wrote:
> Kerin Millar wrote:
>> Helmut Jarausch wrote:
>>> On 10/24/2012 03:54:39 PM, Nikos Chantziaras wrote:
>>>> Kernels 3.4, 3.5, and 3.6 can result in severe data corruption if
>>>> you're using the EXT4 filesystem:
>>>
>>> It looks as if Eric Sandeen has found the culprit and Theodore Ts'o has
>>> suggested this patch yesterday
>>> https://lkml.org/lkml/2012/10/28/309
>>>
>>> Helmut.
>>>
>>
>> Here's the final patch:
>>
>> https://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=ffb5387
>>
>>
>> It turns out to have had nothing to do with nobarrier.
>>
>> --Kerin
>>
>>
>
>
> Could you explain a little on who it could have affected?  Is it more
> serious or less serious than originally thought?  This is for those of
> us who don't subscribe to the kernel mailing list, which I have read is
> hugely active.
>
> Thanks much.

AFAIK they originally thought it was related to normal reboots after
short uptime, which would potentially put all ext4 users at risk who
use newer kernels. That is when the news stories came out about the
disastrous ext4 bug. As it turns it, it seems to have been related to
enabling some combination of experimental, non-default mount options
with an unsafe shutdown cherry on top. So the number of people who are
using that would be far, far fewer than they originally thought. Still
a dangerous bug for those people, of course, but I would guess
probably 99+ percent of ext4 users weren't at risk from it.

That is my layman's understanding, I probably got it all wrong.
Experts, feel free to correct me. :)