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

[Kerin Millar <kerframil@fastmail.co.uk>]

Grant Edwards wrote:
> On 2012-10-25, Kerin Millar<kerframil@fastmail.co.uk>  wrote:
>
>> The comment you linked to was fairly bereft of technical content,
>
> That "comment" was from _Ted_Ts'o_ for pete's sake.

I don't care it was from the heavens upon high. The only remark that was 
meaningful in a technical sense was the thoroughly inconclusive 
"Update", suggesting (but not establishing) that it might be related to 
certain combinations of mount options. The rest was just hand-waving 
about how it couldn't be a big deal because, if it was, "lots of people 
would have been complaining", which is denying the antecedent.

Let's separate two fundamental issues here. One issue is that of data 
corruption, which is a big deal. I'm sure most of us would agree on that 
point. That such an issue is going to generate news - and no small 
degree of discussion - is a given. People will want to know what the 
problem is and what they can do about it in order to be safe. What did 
Ted expect?

The second issue is that of the scope of the bug. This is where I took 
issue with the comment. The overall meaning of his comment could be 
interpreted as "Your collective concerns are overblown because the scope 
of this bug is minimal. Oh, and I think it may have something to do with 
these mount options which, being esoteric, nobody in their right mind 
would be using anyway so, hey, big deal."

The fact of the matter is that the investigation, even as I write this, 
is ongoing and no patch has been produced. Consider that for a moment. 
It doesn't matter how brilliant Ted is, or that you have seen fit to 
sample his mucus. Telling us all that we should be unconcerned because 
the scope is minimal *before* he and his peers have completed their 
investigation and a line been drawn under the affair was simply premature.

That the bug reporter has since demonstrated that the corruption can 
occur in kernel versions that don't include commit eeecef0af5 - kernels 
which we were previously told were not affected - only serves to 
demonstrate this point.

--Kerin