Re: qtparted and kernel disagree about partitions
On Friday 05 March 2010, consul tores wrote:
> 2010/3/5 David Goodenough <david.goodenough@btconnect.com>:
> > On Friday 05 March 2010, consul tores wrote:
> >> 2010/3/5 David Goodenough
<david.goodenough@btconnect.com>:
> >> > On Friday 05 March 2010, consul tores wrote:
> >> >> 2010/3/4 Mike Dresser
<mdresser_l@router.windsormachine.com>:
> >> >> > On Thu, 4 Mar 2010, David Goodenough wrote:
> >> >> >> hda: Host Protected Area detected.
> >> >> >> ^Icurrent capacity is 268435455 sectors (137438 MB)
> >> >> >> ^Inative capacity is 312581808 sectors (160041 MB)
> >> >> >
> >> >> > Is this a PATA drive, and from there, has a jumper on the back
for
> >> >> > LBA48?
> >> >> >
> >> >> > Mike
> >> >>
> >> >> The problem does not look related with HD, it is more related to
> >
> > the
> >
> >> >> kernel. Anyway, You can use a specific tool which is provided by
the
> >> >> HDs' company.
> >> >
> >> > Well the linux tool to do the job is supposed to be hdparm -N, but
> >> > that does not work because the correct option is not selected in
the
> >> > kernel (CONFIG_IDE_TASK_IOCTL).
> >> >
> >> > David
> >>
> >> Yes, it is absolutely correct respect to kernel, but the second
> >> sentences is related to the HD (hardware). The first one reffers to
> >> the kernel.
> >>
> >> francisco
> >
> > Are you saying there is a hardware tool provided by Samsung?
> >
> > Anyway, if you look at bug 572618 you will see the solution, it involves
> > two parameters for modules.
> >
> > David
>
> Yes, it has been a norm for many brands, in fact, some tools work on
> many HDs, but i insist it does not look like a hardware problem, it
> looks clearly like kernel problem.
>
> If i were in this situation again, first thing to check, should be
> using the old kernel which worked well. after that, if i have the same
> problem, testing the HD with cfdisk, fdisk and sfdisk should be
> sufficient to discard a hardware problem; first, reading man pages,
> and DO NOT DOING any change, do not writing any thing. Also, there are
> Debian tools that work very efficiently as testdisk, smart-tools (i am
> not sure), and others to check HDs and recover data.
>
Actually the problem is (as I have pointed out in the bug report) worse
that this, because while the kernel has been updated to take note of
the HPA, the tools such as *parted have not. So if your disk has an HPA
and you try to partition it using *parted, presumably you will get what
looks like a hardware error, and it will make no sense as the sizes
reported by *parted are the physical ones, not the HPA reduced ones.
David
> francisco.
>
> "thanks for the information:
> bug # 572618.
> You should be able to make the kernel ignore the HPA thus:
>
> 1. Create a file under /etc/modprobe.d containing the lines:
> options ide_core nohpa=0.0
> options libata ignore_hpa=1
> 2. Run 'update-initramfs -u -k 2.6.32-2-686'
> 3. Reboot
>
> Ben."
>
Reply to: