Re: SiI3112 Serial ATA Maxtor 6Y120M0 incorrect geometry detected

[micah milano <micaho@gmail.com>]

I agree that parted reports an error, however I think there is
something else going on. When I run fdisk there is a different CHS
reported for each disk, even though they are the same geometry. I
can't partition the second disk to have the same partition sizes as
the other disk because of this. I only mentioned parted because I was
trying every partitioner there is.

I am glad to hear I am helping tighten up some code to make things
clearer, but I was hoping someone might have a solution for how I can
partition this second identical disk so it has identical partition
sizing as the first so I can put software RAID mirroring on them. From
what it sounds like you guys are saying (IANALKH) dmesg is reporting
the wrong CHS, and parted has a fix for its error, but is there
something I can do to get the partition sizing right?

thanks for the quick replies!
micah


On Sun, 26 Sep 2004 12:59:48 +0200, Bartlomiej Zolnierkiewicz
<bzolnier@gmail.com> wrote:
> On Sun, 26 Sep 2004 12:29:37 +0200, Andries Brouwer
> <andries.brouwer@cwi.nl > wrote:
> > On Sun, Sep 26, 2004 at 12:15:33AM +0200, Bartlomiej Zolnierkiewicz wrote:
> >
> > > I'm tired of this issue and this is what I'm going to do:
> > > - remove CHS info from IDE printks and /proc/ide/
> > > - add BLKGETSTART ioctl for getting partition's start sector
> > >   (this is the only legitimate use of HDIO_GETGEO currently)
> > > - at least obsolete HDIO_GETGEO in IDE or even remove it (failing is
> > >   better than returning unexpected results)
> > > - silence complainers :)
> > >
> > > Bartlomiej
> >
> > Yes. Consider returning 0 for C/H/S in HDIO_GETGEO.
> >
> > We need the BLKGETSTART ioctl for another reason:
> > the field start of a struct hd_geometry has 32 bits
> > and fails for partitions starting past the 2TB mark.
> > So, just like BLKGETSIZE64, this BLKGETSTART must
> > return a u64 and return an offset in bytes.
> 
> Yes, I though so. Also u64 is 64-bit safe.
> 
> > As we all know, geometry is meaningless for IDE.
> > But it is not for MFM/RLL and similar ancient stuff.
> > If support for old disks is not broken yet, try not to break it.
> 
> OK.
>