Fasttrack TX2000 RAID problem

[Greg Felix]

I think Sergey is onto something.  Looks like there is a host
protected area on that disk and it roughly matches the size of the
offset we're seeing after the metadata.  So what's the deal with HPA
and why does the kernel after 2.6.10 (I think I read) simply disable
it?  Anyway it's looking like that is the problem.  Here's the
relevant part of dmesg:

hde: max request size: 128KiB
hde: 78156288 sectors (40016 MB) w/2048KiB Cache, CHS=65535/16/63, UDMA(100)
hde: cache flushes not supported
 hde: unknown partition table
hdg: max request size: 128KiB
hdg: Host Protected Area detected.
	current capacity is 78125000 sectors (40000 MB)
	native  capacity is 78165360 sectors (40020 MB)
hdg: Host Protected Area disabled.
hdg: 78165360 sectors (40020 MB) w/2048KiB Cache, CHS=65535/16/63, UDMA(100)
hdg: cache flushes not supported
 hdg: unknown partition table

Possible solutions:
The kernel doesn't disable the HPA but instead ignores it. (In theory
then bringing our metadata back to totalSectors - 63)

The kernel continues to disable it and dmraid figures out about the
HPA then factors that in as an offset when looking for metadata


Anyone know more about the decision to disable HPA in the kernel? 
What do you think would be the proper fix?

Greg


2005/8/26, Sergey Vlasov <vsu at altlinux.ru>:
> On Fri, 26 Aug 2005 10:15:41 -0600 Greg Felix wrote:
> 
> > So this is proof to me that the BIOS is reading Disk 2 as 21MB (or
> > 41000 or so sectors) less than the kernel is.
> >
> > So where's the problem? The BIOS? The Kernel? The IDE disk? And why
> > does the RAID work in Windows?
> 
> Can you provide the full dmesg output?
> 
> Maybe Disk 2 has host protected area enabled; recent Linux kernels
> disable HPA unconditionally, and this could explain the difference in
> capacity.
> 
> 
>