Re: [SLE] Partition Magic vs. Linux partitioner
Thank you for your reply. I think there must be a story
about all these different disk geometry reports. Knowing
will make it a bit easier to make any necessary
partitioning decisions. So, I will have to take some time
to know more about this issue. I'll report anything I
find.
On Tue, 17 Aug 2004 17:55:36 -0400
Felix Miata
johnswolter@provide.net wrote:
I just decided to give a try to GNU parted. Yep, error message, so I stopped.
Here's parted's startup message.
"Error: The partition table on /dev/hda is inconsistent. There are many reasons why this might be the case. However, the most likely reason is that Linux detected the BIOS geometry for /dev/hda incorrectly." [Bad parted]. "GNU Parted suspects the real geometry should be 4866/255/63 (not 77557/16/53)." [cyls/heads/sectors?]. "You should check with your BIOS first, as this may not be correct. You can inform LINUX by adding the documetation for mor information. If you think Parted's suggested geometry is correct, you may select Ignore to continue (and fix LINUX later). Otherwise, select Cancel (and fix LINUX and/or the BIOS now)."
Selecting Cancel...
"Information: The operating system thinks the geometry on /dev/hda is 77557/16/63. Therefore, cylinder 1024 ends at 503.999M."
I also used Partition Commander to make drive copies and got the following as geometry from its information window.
Cylinders: 4866 ....Heads: 255 ..Sectors: 63
The BIOS has multiple reports.
When set to Access Mode = Auto Cylinders: 19161 ....Heads: 16 ..Sectors: 255
Highly unlikely.
When set to Access Mode = LBA Cylinders: 4866 ....Heads: 255 ..Sectors: 63
Almost certainly.
When set to Access Mode = Large Cylinders: 1277 ....Heads: 240 ..Sectors: 255
Definitely not if there has ever been an M$ OS on it.
Interesting results. Which information is right? The real way to find out is to get out your favorite low level disk editor and see what is where. Since I have a blank target drive and I'm short of time, it's time to throw caution to the wind a see what happens.
DFSee and its author can help you get it figured out, but not for free: http://www.dfsee.com/dfsee.htm http://groups.yahoo.com/group/dfsee-support
It's worth the money, the only thing I ever partition with. -- "Never tire of doing what is right." 2 Thessalonians 3:13 NIV
Team OS/2 ** Reg. Linux User #211409
Felix Miata *** http://members.ij.net/mrmazda/partitioningindex.html
[ prior discussions omitted ] Be sure also to read this page: http://portal.suse.com/sdb/en/2004/05/fhassel_windows_not_booting91.html Basically, there is a bug in the GNU parted that comes with SuSE 9.1 that may cause problems if you are doing multi-boot on the same disk with SuSE and Windows. -Ti
I've read the bug page which is interesting but while the drive is multiboot it only boots it is a LINUX system only and was created as a LINUX system. The standard choices of LINUX, floppy, and Failsafe are the only boot choices. The bug page says: "The problem occurs if 1.the BIOS and Linux "see" different disk geometries AND 2.the Windows partition is larger than about 8 GB (more precisely: if the first hard disk partition ends on cylinder 1024 or beyond this point). [The first partition is the /boot filesystem and is 250 MB. When the system is booted, Windows may use the values in the partition table, which causes a failure. Currently, this problem also occurs on other Linux distributions using kernel 2.6." Still my system meets the first criteria, seeing different geometries. I did set the disk BIOS to LBA and still received the error. This disk is a near to production setup and I'm rightly concerned not to trash it by making partition table changes. Further, I have been using Partition Commander(PC) to copy as checkpoints during this process. Each drive image takes 5 hours, slowing the integration process. PC appears to me to be a mostly Windows oriented product. That may imply when partition copies are made they follow some yet undiscovered aspect of Windows partitioning practices. Wait. Since my last check point has the same problem and the current drive is working just fine, I could use that last check point to test this parted fix. Bad parted, bad!
[ prior discussions omitted ]
Be sure also to read this page:
http://portal.suse.com/sdb/en/2004/05/fhassel_windows_not_booting91.html
Basically, there is a bug in the GNU parted that comes with SuSE 9.1 that may cause problems if you are doing multi-boot on the same disk with SuSE and Windows.
-Ti
-- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
[prior comments deleted] I'm responding to my own last comment. I ran the parted update that corrects the partition table such that parted does not report a partition table error. See... http://portal.suse.com/sdb/en/2004/05/fhassel_windows_not_booting91.html Parted does report the following message... "Using /dev/hda Information: The operating system thinks the geometry on /dev/hda is 77557/16/63. Therefore, cylinder 1024 ends at 503.999M." Inside parted, the print command reports partition table with the extended partition as "lba, type=0f". I had set the drive in the BIOS to LBA rather then AUTO as was previously. Parted seems to know the raw drive geometry and appears to report the extended drive correctly. It is a bit confusing to see the raw drive parameters reported but not the LBA parameters. Is all ok? Does parted have an option for more detail about partitions? I'll try to look that up at gnu.org. Some notes about the patch and fix: 1. It has a perl script that runs the update. I don't know Perl as of now but someone on the list may want to look it over. 2. A replacement library, libparted-1.6.so.0.0.6, is loaded. I'm wondering now about other sources of parted like knoppix, the SUSE install disks are obviously problematic. I'll have to check parted's development page to find if this is a general fix or just SUSE specific. My utility disks will have to be redone. Does this complete the story?
[prior comments deleted]
I've found a procedure that works for a couple of my systems to overcome the Windows will not boot after 9.1's install. These systems are using GRUB. I performed this procdure from the rescue disk, I used the LBA drive parameters, discussed below, as boot parameters. Set the BIOS to LBA mode when you BIOS allows that to be done. Write down the LBA numbers for cylinders, heads, and sectors. Boot into the GRUB menu and type the LBA numbers into the boot parameter line. Explicity, hda=<cylinders>, <heads>, <sectors> and boot LINUX. Now, LINUX should know the correct drive parameters. Check this by using grub's interactive prompt.
grub grub>geometry (hd0).....grub numbers drives from 0
...check the numbers and be sure they are the same as the disk you took them from. Grub now knows the correct LBA parameters. grub>find /boot/grub/stage1 ... will tell you where /boot is if you don't know. grub>root (hd0,0)...which is my case grub>setup (hd0)...writes the MBR with boot code and data ...the description of the disk LBA parameters are included. grub>quit When I did this procedure I could boot Windows. My setup is simple in that I have a single Windows partition but a rather complicated LINUX partition setup. If you need to hide partitions or other special boot considerations, you will have to modify the /boot/grub/menu.lst. YaST can do that although I found that to be tedious but necessary. Thank you SUSE for makeing YaST Open Source recently, now we can understand and fix it. If you do not have LINUX installed you can put LBA parameters in at the start of install but I have not done this. If you have trouble with that procedure you can proceed with the install and fix it afterwords. Takes guts. Note:Grub does not inquire to the BIOS for drive parameters it just uses whatever LINUX hands to it.
participants (2)
-
johnswolter@provide.net
-
ti@amb.org