experimental parted in sarge ...
On Tue, Dec 21, 2004 at 10:56:19PM -0500, martin yazdzik wrote:
> Dear S,
> Many, many thanks for your kind e-mail copy of the newsgroup post
> about the dual boot disk geometry issue. It is indeed hard to track
> down, as there are three factors which must be in place to have it go
> awry.
Thanks martin for your report, and sorry for not having responded earlier, i
think your email did get lost in my mailbox, even if i salvaged it from the
spam box where is almost went.
I am forwarding this mail to debian-boot and debian-release, in hope that it
is still time to do something about this.
Anyway, to debian-boot, and joeyh in particular, and to the debian release
managers, this is the exact example of why we should include the experimental
version of parted, and not the one currently in sarge/sid. The number of
dependent package is minimal, and easy to fix (Colin Watson can attest to
this), and the fix is solving a real problem, where martin did all the
checking and reproduction work that have been called for.
If this is not reason enough to add this info, i wonder what would be, or if
it is definitively too late for a fixed parted in sarge.
Ok, here follows the rest of martin's analysis, and again thanks to you martin
for following this up.
Friendly,
Sven Luther
> 1. The windows install must be on a BIOS set to auto or c,h,s, rather
> than the more modern LBA.
> 2. 2.6 kernel must be used to install
> 3. parted must write the partition table
>
> The problem is that the current installer does not allow one simply to
> install on a clean, formatted partition, or if it does, I am too stupid
> to figure out how. At the moment, with rc2, one cannot create such a
> partition and install, the installer sends one back to the partitioner,
> which then insists on writing the partition table to disk, in so doing,
> creating the havoc.
>
> You suggested in your e mail and reply to the group "Can you do an
> install without touching the partition table, and then try the
> experimental version of parted, and see if it fixes your issue ? "
>
> I cannot see how to do it. If you can tell me how, the problem
> vanishes, since one need no longer use parted at all in the install
> process.
>
> However, the really good news is as follows:
>
>
> I installed the latest parted with the following results:
>
> Using /dev/hda
> (parted) print
> Disk geometry for /dev/hda: 0.000-57231.562 megabytes
> Disk label type: msdos
> Minor Start End Type Filesystem Flags
> 1 0.031 29000.148 primary ntfs boot
> 2 29000.149 56353.007 primary reiserfs
> 3 56353.008 57231.562 primary linux-swap
>
> This signifies that it no longer uses the kernel read, as in the earlier
> version(courtesy of ubuntu live cd! One must love ubuntu for the sheer
> brilliance of the idea. If they would just admit we all need commercial
> drivers, java, etc, it would rule the earth.....) In other words,
> writing a table with this would not rewrite the existing artificial
> reading.
>
>
> Information: The operating system thinks the geometry on /dev/hda is
> 116280/16/63. Therefore, cylinder 1024 ends at 503.999M.
> (parted) print
> Warning: Unable to align partition properly. This probably means that
> another
> partitioning tool generated an incorrect partition table, because it
> didn't havethe correct BIOS geometry. It is safe to ignore,but ignoring
> may cause
> (fixable) problems with some boot loaders.
> Ignore/Cancel? ignore
> Disk geometry for /dev/hda: 0.000-57231.562 megabytes
> Disk label type: msdos
> Minor Start End Type Filesystem Flags
> 1 0.031 29000.148 primary ntfs boot
> 2 29000.149 56353.007 primary reiserfs
> 3 56353.008 57231.562 primary linux-swap
>
> Of course, fdisk, in any situation gets it right, as it reads
> independently of the kernel:
>
> Disk /dev/hda: 60.0 GB, 60011642880 bytes
> 255 heads, 63 sectors/track, 7296 cylinders
> Units = cylinders of 16065 * 512 = 8225280 bytes
>
> Device Boot Start End Blocks Id System
> /dev/hda1 * 1 3697 29696121 7 HPFS/NTFS
> /dev/hda2 3698 7184 28009327+ 83 Linux
> /dev/hda3 7185 7296 899640 82 Linux swap
>
>
> Obviously, if the release used the newest parted, it would solve the
> issue, or if one could simply do as you ask, and install to a partition
> formatted with something else, which I cannot see how to accomplish, the
> issue would cease to exist.
>
> The fedora solution was to allow the passing of the c,h,s parameter to
> the kernel, thus upon install, one could do in d-i terms:
>
> expert26 c,h,s=xxxx,255,63
>
> This clearly does not work, although, in theory, one should be able to
> adjust the heads for the ibm laptop as it says it does. So, one should
> be able to install with any other parameter affecting geometry. I have
> not found this to work.
>
> The solutions seem to me to be three:
>
> 1. allow the parameter, supra, to be passed(simplest)
> 2. use parted experimental,
> 3. make sure people know that if they are dual booting to a non-LBA bios
> upon which windows was installed first, they must either use the 2.4
> kernel, or at least install on it in expert mode, selecting the 2.6
> kernel when the prompt arises.
>
>
> The danger here is that the people who are most likely to hit the
> problem unawares are the people like myself, who are ignorant newbies.
> They will do, as I did, panic, and reinstall. I fortunately, read the
> newsgroups, and research well, so the next few tries, I got the
> workaround I use with my current working install from a few months
> ago.
>
> In the end, that it is idiotic that one cannot adjust laptop bios is
> beside the point, it is so, and must be dealt with. I noticed on the
> last few new dell desktops we have bought, the bios is set to LBA, so
> there is no issue.
>
> I deeply appreciate the time you took to read this, and hope that my
> rather direct style of writing offends none, as it is intended, perhaps
> overly colourfully, to point out was many others already do, that we
> need to become more user oriented if we are to create a better world for
> all of us. I am not at all angry, just passionate that those who use
> monopolies to create power be reined in, and for this to happen to
> microsoft, we need a much larger base, which perforce must include the
> regular home user. If my kids hate linux because their web content
> viewing is limited, we have no future. Thus, we need to make an
> installer available to pass to not very skillful teens who want to
> experiment, but cannot afford to wreck their machines. The real geeks
> will always be there; we need to help the "möchtegern" geeks to use
> linux to force the hardware manufacturers to open their
> specifications.
>
> Anyway, we know, at least for me, the experimental parted works without
> destroying the necessary table for the necessary evil.
>
>
> Very best wishes,
>
> Martin
>
>
Reply to: