GEOM_PART: Integrity check failed (ada2, MBR)

[Ronald F. Guilmette]

In message <alpine.BSF.2.00.1405160453130.34394 at wonkity.com>, 
Warren Block <wblock at wonkity.com> wrote:

>On Thu, 15 May 2014, Ronald F. Guilmette wrote:
>>...
>> May 15 21:53:33 segfault kernel: GEOM_PART: integrity check failed (ada2, MBR)
>
>Some Linuxes (Linii?) might be creating "hybrid" GPTs, with a PMBR that 
>is non-standard.  I can't speak to what Gparted does.

I would like to get to the bottom of this, and find out what is really
happening, and which piece of software is actually at fault here.

If I were to post the first 512 bytes of the drive in question (run
through "od -c") then could you, Warren, tell me what the real problem
is here and which piece of software is actually at fault?

To me it is more than a little inexplicable that two such closely
related operating systems (linux, FreeBSD) cannot even seem to agree
on what the age-old and time-tested MBR format should look like.

>There is a sysctl to relax the strict checking in FreeBSD, but I would 
>suggest using gpart(8) instead.

If something is checking for correctness of the MBR then I most certainly
*do not* have any desire to disable that.  Certainly not now, when I don't
even have the foggiest idea what the problem is.

Nor do I have an immediate desire to simply use FreeBSD gpart to re-
partition the drive and do the equivalent of newfs (for an ext3 fs)
all over again.

In short, I want to actually find out what the bleep is going wrong, and
why, and then learn how to fix it.

I greatly appreciate that you are willing to give me this particular
fish Warren, but in this instance I'm going to (mix my metaphores and)
look this gift fish in the mouth and ask you to _teach_ me how to fish
instead.

If Linux and FreeBSD can't even agree on what constitutes a valid MBR,
then I think that we owe it to the next poor fool who tries to get these
two OSes to ``interoperate'' in this specific manner to find out why not.



Regards,
rfg