Dual Boot, Grub, FC3-WinXpSP2, 2 drives. No go.

[Nat Gross]

Robert Locke wrote:

>On Mon, 2005-02-28 at 15:46 -0500, Nat Gross wrote:
>  
>
>>Robert Locke wrote:
>>
>>    
>>
>>>On Mon, 2005-02-28 at 15:18 -0500, Nat Gross wrote:
>>> 
>>>
>>>      
>>>
>>>>Robert Locke wrote:
>>>>
>>>>   
>>>>
>>>>        
>>>>
>>>>>On Mon, 2005-02-28 at 14:06 -0500, Nat Gross wrote:
>>>>>
>>>>>
>>>>>     
>>>>>
>>>>>          
>>>>>
>>>>>>Robert Locke wrote:
>>>>>>
>>>>>>  
>>>>>>
>>>>>>       
>>>>>>
>>>>>>            
>>>>>>
>>>>>>>On Mon, 2005-02-28 at 13:18 -0500, Nat Gross wrote:
>>>>>>>         
>>>>>>>
>>>>>>>              
>>>>>>>
>>><snip>
>>>
>>> 
>>>
>>>      
>>>
>>>>>I guess you need to define which "partition" contains the "WINDOWS"
>>>>>directory.  That is the one that you would be "booting" from.  So is
>>>>>that on hdb1 or hdb5.....
>>>>>
>>>>>But I must admit that my Windows boot process knowledge is getting
>>>>>mighty rusty, now that I use VMWare to run it.
>>>>>
>>>>>As I recall, the Windows bootloader is mighty cheesy....  It simply
>>>>>pointed to the first sector of the "active" partition....  I wonder if
>>>>>playing with hide and unhide in grub might help.  Can you hide a whole
>>>>>drive or just a partition, haven't had to do one myself?  But that way
>>>>>you might be able to allow Windows to think it is the only drive again
>>>>>which is perhaps what it is expecting?  Windows may be trying to
>>>>>interpret the first drive's partition table and getting itself confused
>>>>>as it tried to boot....
>>>>>
>>>>>
>>>>>     
>>>>>
>>>>>          
>>>>>
>>>>Grub intercepts the boot.
>>>>
>>>>   
>>>>
>>>>        
>>>>
>>>Well, that's open for a little interpretation....
>>>
>>>chainloader is essentially just passing control back to Windows to boot
>>>itself.  
>>>
>>>      
>>>
>>Oh.
>>
>>    
>>
>>>The Windows MBR component is what is replaced/intercepted by
>>>GRUB.  But the chainloader line is execute sector +1 on the rootnoverify
>>>line.  Generally Windows has historically held the NTLDR executable
>>>there that then reads the boot.ini file and performs it's boot.....
>>>
>>> 
>>>
>>>      
>>>
>>One minute. So, the boot.ini might have been on C:?<gasp>
>>On the other hand, it should not be too hard to write a boot.ini, if I 
>>recall correctly from my last win crash. But, where should I stick that 
>>file? Or, maybe I should try to mount the the ntfs partition (via the 
>>3rd party driver) just to look for boot.ini...
>><sigh>
>>Mr. Gates, I ain't going back to you!!!! (after I'm thru with this.)
>>-nat
>>
>>    
>>
>
>I would think that it would be on the partition that you are booting
>Windows from, which I presume is hdb5, which houses the boot.ini....
>
>Presuming that hasn't been changed...
>  
>
I thought so, too. But I just mounted both (vfat and ntfs) partitions 
under FC3, and there is no boot.ini file. At this point I am certain 
that for some reason (maybe due to an earlier install of Windows) 
Windows kept the boot.ini on C, hda, and used that to boot from E.

>So, perhaps the bigger issue is that did we (Linux/anaconda) re-write
>your partition table to confuse Windows. 
>
Since I can mount all the partitions under fc3, and the files look a-ok, 
I don't think fc3 caused any problem.
After all, I told the installer to format hda (since I assumed that all 
win boot stuff is on the second drive.

> This has happened in the past,
>but was with FC2.  But have you changed any BIOS parameters related to
>LBA (Logical Block Addressing) since you last booted Windows
>successfully?  
>
No.

>Perhaps Windows interpretation of the drive size and what
>is written in the MBR of the second drive are confused, and that might
>be affected by the Fedora/anaconda installer or by changes in the BIOS
>related to how to interpret CHS (Cylinders, Heads, Sectors)....
>
>One other thought... have you moved the drive?  Was it originally the
>primary drive but now you have moved it to be secondary?  That might
>account for a confused boot.ini file.....
>  
>
Nope. Hardware, bios, etc., untouched.

>I have heard of a recovery console available from Windows XP but do not
>have any direct experience with it
>
I do. It's not fun, and would only use it on a dual boot system if in 
real emergency.

> (I always equated it with "linux
>rescue").  You could also try to boot one of the Live CDs that includes
>the NTFS driver to at least view the boot.ini, but I wouldn't try to
>launch it in writable mode.  For that you want to figure out the Windows
>Recovery Console.
>
>  
>
One key question is, does grub require that I make hdb5 bootable?

>I'm off for the next several hours, but good luck...
>  
>
<sigh> and I'm off *after* that. But, hey, I have the data, so I guess I 
can wait.

>--Rob
>
>  
>
Thank you!
-nat