gpart

[Marcel Moolenaar]

On Dec 19, 2008, at 10:47 PM, Takahashi Yoshihiro wrote:

> In article <354036B7-A4AA-4D66-A2AC-9CB66D292FAA at mac.com>
> Marcel Moolenaar <xcllnt at mac.com> writes:
>
>>> Please see:
>>> http://lists.freebsd.org/pipermail/freebsd-arch/2008-September/008627.html
>>>
>>> The da0 has a BSD slice.  GEOM_BSD/GEOM_PC98 recognize it correctly.
>>> But the gpart does not.  From debug printf (dmesg.txt), it seems  
>>> that
>>> the gpart read incorrect sector.
>>
>> This seems to indicate that the pc98 slice is wrong, or
>> you don't have the BSD disklabel in the right sector.
>>
>> Can you give me a dump of the first 32 sectors on the
>> disk and the first 32 sectors of slice 1.
>
> I put to the following.
> http://home.jp.freebsd.org/~nyan/geom/da0
> http://home.jp.freebsd.org/~nyan/geom/da0s1
>
> BTW, the result of using the gpart is:
> http://home.jp.freebsd.org/~nyan/geom/gpart.da0
> http://home.jp.freebsd.org/~nyan/geom/gpart.da0s1

I think the problem is with the perceived geometry.
If you boot verbose with GEOM_PC98, do you see that
the fwsectors/fwheads are being guessed? If not, can
you tell me what the fwsectors and fwheads values
are?

With GEOM_PART_PC98, can you run:
	gpart list da0

You should get something like:

	Geom name: da0
	fwheads: 16
	fwsectors: 63
	last: 80293247
	first: 63
	entries: 4
	scheme: MBR

Of course scheme should be PC98 in your case.

I suspect that the number of heads and/or sectors
per track is different.

Thanks,

-- 
Marcel Moolenaar
xcllnt at mac.com