Jump to content











Photo

Cloning and multibooting several Win7/8 partitions - also logical


  • Please log in to reply
79 replies to this topic

#51 cdob

cdob

    Gold Member

  • Expert
  • 1449 posts

Posted 18 February 2013 - 09:05 PM

Can we add (or clarify) that the patched bootsector file works with BOTH NTLDR and BOOTMGR?
Another logical partition added, XP installed
Ntldr, boot.ini and ntdetect.com copied to the new logical partition.
Grub4dos at MBR restored.

chainloader (hd0,5)/ntldr
Will booot NTLR from drive=0x80, partition=0x5(hiddens sectors=0x1f73800)

XP does boot.
However hidden sectors 2048 matches primary partition.
At primary partition exist boot.ini from XP installation still.

At primary partition boot.ini and ntdetect.com deleted.
'chainloader (hd0,5)/ntldr': BOOT.INI not found.



chainloader (hd0,4)/bootmgr:

There is a file (hd0,4)/boot.ini





[boot loader]
timeout=5
default=multi(0)disk(0)rdisk(0)partition(3)\WINDOWS
[operating systems]
multi(0)disk(0)rdisk(0)partition(0)\WINDOWS="0 Windows XP" /noexecute=optin /fastdetect
multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="1 Windows XP" /noexecute=optin /fastdetect
multi(0)disk(0)rdisk(0)partition(2)\WINDOWS="2 Windows XP" /noexecute=optin /fastdetect
multi(0)disk(0)rdisk(0)partition(3)\WINDOWS="3 Windows XP" /noexecute=optin /fastdetect
multi(0)disk(0)rdisk(0)partition(4)\WINDOWS="4 Windows XP" /noexecute=optin /fastdetect
multi(0)disk(0)rdisk(0)partition(5)\WINDOWS="5 Windows XP" /noexecute=optin /fastdetect
multi(0)disk(0)rdisk(0)partition(6)\WINDOWS="6 Windows XP" /noexecute=optin /fastdetect
multi(0)disk(0)rdisk(0)partition(7)\WINDOWS="7 Windows XP" /noexecute=optin /fastdetect
c:\logical_XP.bin="Windows XP - logical partition - bootsector file"
c:\grldr="Boot grub4dos"
c:\ntldr="NTLDR"
\dummy="Logical partition: Windows 7"
2048 hidden sectors (hd0,4) : (hd0,4)/bootmgr list some entries (hd0,4)/boot.ini
There are no multi() entries listed.
Following entries are listed.
\logical_XP.bin is a bootsector file from logical XP partition, hidden sectors 32978944 patched.


Hidden sectors (hd0,4)2048, (hd0,5)32978944
\logical_XP.bin does boot XP.

Hidden sectors (hd0,4)2048, (hd0,5)2048
\logical_XP.bin: back screen, blinking cursor at upper left corner

Ntldr requires a patched boot sector and boot sector file.
 
Basically it seems that BOOTMGR (when there is an entry in the \boot\BCD for the "legacy" OS) needs not NTLDR, accesses BOOT.INI directly and is (maybe) capable of loading the XP without NTLDR.
Do you have a BCD example?
I remember a relating hint at Longhorn times (boot.ini only), never tried myself.

Added:
From 2005:
http://chris123nt.co...eta1_5112/#boot
I copied the F:\Windows\System32\winload.exe (Vista Installation) file to C:\Windows\System32 (XP Installation)

wendy explained this
http://reboot.pro/to...t5x/#entry86739
It's possible at build 5112 and fails at later builds.

Is it required to test this at current files again?

Edited by cdob, 19 February 2013 - 04:23 PM.


#52 cdob

cdob

    Gold Member

  • Expert
  • 1449 posts

Posted 19 February 2013 - 03:56 PM

A solution for cross-linking problem will appear. Then I need only one base install. I'm sure this can be done somehow. I just don't understand Windows registers and startup procedure well enough.
As for a general rule: do NOT change %SystemRoot% drive letter.
I did this once at NT sucessfully, a lot of work, never again.

And Windows 7 uses letter C: by default.
It's a good idea to use default settings.

If there is no letter set so far, Windows 7 will set c: again for %SystemRoot%.
That's letter c: is free to be used. And %SystemRoot% is not defined already.
Delete MountedDevices C: and MountedDevices other letter, if clone target partition was mountetd already.
%SystemRoot% uses the same drive letter again.


There is Windows 7 at (hd0,4) using letter C: and MountedDevices C: delted.

As for cloning: boot a PE
Default Windows 7 startup repair command prompt is sufficient.
http://windows.micro.../startup-repair
create new logical partition at diskpart.
imagex.exe /export install.wim 3 win7.wim
imagex.exe /compress maximum /check /norpfix /flags "Professional" /append E:\ .\win7.wim "Windows 7"
imagex.exe /apply win7.wim 2 h:\

bcdedit /store h:\boot\bcd /set {bootmgr} device   partition=h:
bcdedit /store h:\boot\bcd /set {default} device   partition=h:
bcdedit /store h:\boot\bcd /set {default} osdevice   partition=h:
Adjust settings to your configuration
3 insta.wim part relating your Windows installation
E: offline windows 7: at offline C: is used for %SystemRoot% and MountedDevices C: is deleted.
H: current used letter at PE for the new logical partiton

grub4dos: chainloader (hd0,6)/bootmgr does load Windows 7.
Windows 7 set letter C: as for (hd0,6).
There are two independend Dektops.

#53 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 15103 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 19 February 2013 - 06:27 PM

 Do you have a BCD example?
I remember a relating hint at Longhorn times (boot.ini only), never tried myself.

Added:
From 2005:
http://chris123nt.co...eta1_5112/#boot
wendy explained this
http://reboot.pro/to...t5x/#entry86739
It's possible at build 5112 and fails at later builds.

Is it required to test this at current files again?
 

But the links are for "the other way round", that is a way to use Winload.exe from (the Beta/Longhorn) using the /USENEWLOADER switch in NTLDR (which AFAIK was added ONLY to the Longhorn NTLDR).

Previous versions of NTLDR did not have that added switch /USENEWLOADER, so, unless there is some "strange" hotfix that contains a NTLDR with that switch added (and provided that the "current" Winload.exe 

 

 

At least the report in the given link is about "current" (Windows 7) BOOTMGR used for "legacy" XP and not needing NTLDR:
http://www.msfn.org/...er/page__st__20
 
Ilko is not "a newcomer" and I tend to believe his reports at face value :thumbup:, but that one is incomplete (not fully tested up to the actual booting of a XP without NTLDR and not reproducible "directly" unless we find a way to replicate the "which already has an entry for NTLDR in BCD" )
I presume that the reference is the entry made in \boot\BCD when you install 7 on a system that already has XP installed.
AFAIK it should be reproducible as follows (from elevated command prompt), from here:

http://www.multiboot...dows_seven.html
 

  • bcdedit /create {ntldr} /d “Windows XP”
  • bcdedit /set {ntldr} device partition=X:
  • bcdedit /set {ntldr} path \ntldr
  • bcdedit /displayorder {ntldr} /addlast
 

 


Better than #2 above might be:

 

bcdedit /set {ntldr} device partition=\Device\HarddiskVolume1

which is independent from the drive letter assigned to the volume in 7 (this is the volume where the actual XP %systemroot% is).

 

The NTLDR, NTDETECT.COM and BOOT.INI are to be put in the same partition (let's say the "active" one) where BOOTMGR and \boot\BCD are.

 

The above should allow to boot a XP from BOOTMGR (and till now it is a "normal" dual boot 7/XP with BOOTMGR as "main" boot manager).

Then, what happens when you rename NTDLR to NTLDR.bak (or delete it)?

According to Ilko :

http://www.msfn.org/...post__p__901814

a grldr listed in BOOT.INI will be chainloadable from BOOTMGR (whilst normally you need to chainload grldr.mbr from \boot\BCD) but he didn't experiment "all the way" to see if also a "normal" BOOT.INI entry that would load a XP will work once NTLDR is not anymore there.

This is what I cannot find a confirmation (or denial)

 

If you prefer, the behaviour could mean two things (depending on results of the test):

  • that we can use BOOT.INI (i.e. a plain text file) to work as "middle-man" between BOOTMGR and grldr
  • that *somehow* BOOTMGR includes the *whatever* is needed to boot a XP even without using NTLDR

and a further step (IF the latter is proved true) would be to try renaming/deleting also NTDETECT.COM (as it is possible that in this "strange" configuration it is not needed at all :unsure:).

 

Additionally, a test is needed to understand if a "legacy-legacy" or "legacy2" entry in BOOT.INI is chainloadable.

In other words the BOOT.INI to be tested should be something *like:

 

 

[boot loader]
Timeout=30
default=multi(0)disk(0)rdisk(0)partition(1)\WINDOWS
[operating systems]

multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="Microsoft Windows XP Professional" /noexecute=optin /fastdetect
C:\CMDCONS\BOOTSECT.DAT="Microsoft Windows Recovery Console" /cmdcons
C:\grldr="grub4dos"

verifying that all three options are accessible from BOOTMGR when NTLDR is there (and they all three boot OK) and then what happens to them after having renamed/deleted the NTLDR, and what happens when also the NTDETECT:COM is renamed/deleted (the last one won't need NTDETECT.COM anyway).

 

Another way (maybe :dubbio:) to understand if when you choose the "legacy" entry in BOOTMGR and \boot\BCD and you see the choices in BOOT.INI you are actually seeing BOOT.INI "through" NTLDR or "directly" through BOOTMGR could be to try the UNdocumented (but actually documented ;)) NTLDR cheatcode of "UNSUPPORTEDDEBUG":

http://sanbarrow.com...opic.php?t=1696

 

 

 

:cheers:

Wonko



#54 cdob

cdob

    Gold Member

  • Expert
  • 1449 posts

Posted 19 February 2013 - 10:30 PM

But the links are for "the other way round", that is a way to use Winload.exe from (the Beta/Longhorn) using the /USENEWLOADER switch in NTLDR (which AFAIK was added ONLY to the Longhorn NTLDR).

What's the major question?
 

Basically it seems that BOOTMGR (when there is an entry in the \boot\BCD for the "legacy" OS) needs not NTLDR, accesses BOOT.INI directly and is (maybe) capable of loading the XP without NTLDRIs the the major question to avoid ntldr?

Or bootmgr list entries from \boot\BCD and BOOT.INI without ntldr, but fails at  loading the XP without ntldr.

 

 

There is another hint:
http://www.betaarchi...opic.php?t=9602
futchi: The only thing you should do is put the winload.exe to system32 of winxp and create a vista boot item in bcd to load the winload.exe in xp's system32.

 

 

At least the report in the given link is about "current" (Windows 7) BOOTMGR used for "legacy" XP and not needing NTLDR:
http://www.msfn.org/...er/page__st__20

I understand this different: there are reports about displayed items, not about real booting.
 

Ilko is not "a newcomer" and I tend to believe his reports at face value :thumbup:, but that one is incomplete (not fully tested up to the actual booting of a XP without NTLDR and not reproducible "directly" unless we find a way to replicate the "which already has an entry for NTLDR in BCD" )
I presume that the reference is the entry made in \boot\BCD when you install 7 on a system that already has XP installed.

Yes, that's it.
The string 'Earlier versions of Windows' is defined in \boot\bcd.
The section {ntldr} and the file \ntldr is referenced.
That's the documented feature.  http://technet.micro...e829686(v=ws.10).aspx

Yes, bootmgr reads boot.ini contents, no ntldr is required for that part.
And given a current grub4dos: 'c:\grldr="Boot grub4dos"' does work. This fails at a historic grub4dos.

 

a grldr listed in BOOT.INI will be chainloadable from BOOTMGR (whilst normally you need to chainload grldr.mbr from \boot\BCD)

As far as I remember, tinybit described this behaviour here lately.

I'll boot up a testing machine in the next days.



#55 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 15103 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 20 February 2013 - 08:23 AM

The post by futchi (GOOD find :thumbsup: ) opens yet another approach, i.e. seemingly adding:

  1. BOOTMGR
  2. \boot\BCD (how exactly configured is to be understood/tested, most probably a "normal" entry similar to the one for Vista :ph34r: or 7)
  3. %systemroot%\System32\winload.exe

To a "normal" XP, it can be booted.

I.e. without going through neither of NTLDR/BOOT.INI/NTDETECT.COM.

 

And yes, the report by Ilko relates ONLY to showing items in BOOT.INI, and that grldr can be chainloaded this way (and he did not report about attempting booting a XP without NTLDR, this latter is the test that is missing, as well as the ability to chainload a bootsector listed in BOOT.INI).

At the light of the report by futchi, it is worth a try to add (if the test booting XP without NTLDR does not work) a winload.exe to the XP's %systemroot%\System32\.

 

:cheers:

Wonko



#56 sir_bootalot

sir_bootalot

    Member

  • Members
  • 41 posts
  •  
    Finland

Posted 22 February 2013 - 06:12 PM

The cross-linking problem got solved by emptying the whole 'MountedDevices', and clones at both primary and logical partitions boot now, and seem to be ok. I havent installed Grub4dos yet, but boot menu is Windows Bootmanager.

 

When I test if the other machine with Windows 8 works ok, I'll delve into Grub4dos. Windowses after Vista don't seem to need hiding of other Windows boot partitions, but what do you guys think, if hiding them (other Win boot partitions) in a boot would add some security from viruses or trojans? I suppose hiding them could be done with Grub4dos.

 

And thanks for advice.


Edited by sir_bootalot, 22 February 2013 - 06:17 PM.


#57 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 15103 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 22 February 2013 - 10:34 PM

Besides Viruses, you should also take into account the way you assign (or let the OS automatically assign) drive letters to volumes.

 

There are two main "schools of thought":

  1. first one (among which you can find me) assigns a given set of letters to the volumes in such a way that a given volume (if visible in the specific booting choice) gets the SAME drive letter NO MATTER WHICH OS is booted.
  2. second one leaves the OS always assign to the booted system volume drive letter C: and has all the other  (if visible in the specific booting choice) volumes an "autoassigned" drive letter

People supporting first one will argue (correctly BTW) that if you do not assign the SAME drive letter to a given volume before or later you will delete a file on the "wrong" volume (or format the "wrong" volume, etc.).

People supporting the second will argue (correctly BTW) that a given (badly coded, and often perfectly replaceable by a "good" one) program that only installs to C:\ will create troubles.

 

A virus may decide to target C:\ volume (no matter which is currently "system") making the first theory - probabilistically - "safer".

Another virus may decide to target "boot" or "system" volume and thus make both approaches equally safe or UNsafe.

 

By cloning, you are however "confined" (unless you go through a whole lot of trouble - changing fixed paths in the Registry, .ini's and what not - and not necessarily ending with success :ph34r:) to second theory (the "wrong" one :whistling:)

 

Of course grub4dos can hide and unhide partitions at boot time.

 

:cheers:

Wonko



#58 sir_bootalot

sir_bootalot

    Member

  • Members
  • 41 posts
  •  
    Finland

Posted 22 February 2013 - 11:36 PM

By cloning, you are however "confined" (unless you go through a whole lot of trouble - changing fixed paths in the Registry, .ini's and what not - and not necessarily ending with success :ph34r:) to second theory (the "wrong" one :whistling:)

 

I don't understand why. Does it matter if I change the partition letters similar afterwards from Disk Manager (in every Windows)? Traces from original drive letters would remain (from D: upwards), but because C: is correct from the beginning in every one of Windowses, no cross-linking problem. And all the other partitions from D: upwards are for data or whatever (like linux partitions).

 

If it does matter for whatever reason, I could also hide the other boot partitions before booting the Windows (any one of them) the first time. That way the non-hidden partitions should be labeled with the same drive letters. Hidden ones don't receive the drive letter, or do they?



#59 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 15103 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 23 February 2013 - 01:08 AM

I don't understand why. Does it matter if I change the partition letters similar afterwards from Disk Manager (in every Windows)? Traces from original drive letters would remain (from D: upwards), but because C: is correct from the beginning in every one of Windowses, no cross-linking problem. And all the other partitions from D: upwards are for data or whatever (like linux partitions).

 

If it does matter for whatever reason, I could also hide the other boot partitions before booting the Windows (any one of them) the first time. That way the non-hidden partitions should be labeled with the same drive letters. Hidden ones don't receive the drive letter, or do they?

 

Let's assume - to simplify - two partitions, both primary, respectively first partition and second partition, each with an install of Windows.

 

I like to install the system(s) in such a way that first partition gets C: and second partition gets D: BOTH when the OS residing on first partition is booted AND when the OS residing on second partition is booted. 

 

This means that when I boot the system on first partition %systemroot% will be C:\Windows, whilst when I boot the system on second partition the %systemroot% will be D:\Windows.

 

Any program (malware) that is hardcoded to damage C:\Windows will have 50% of probabilities of working on the booted system (provided tha tI boot 50% of the time from C: and 50% of the time from D:).

 

I can move files from C: to D: and viceversa.

I can boot from D: and perform on C: operations that I cannot when booted from C: (like deleting files otherwise in use, defragmenting without reboot, deleting the pagefile, etc.)

I can boot from C: and perform on D: operations that I cannot when booted from D: (like deleting files otherwise in use, defragmenting without reboot, deleting the pagefile, etc.)

All this without risking to be confused by a different drive lettering, C: is always first partition and D: is always second one.

I can hide C: and boot from D:, or viceversa I can hide D: and boot from C:, and I can see from which partition I booted at a glance (as I won't see in Explorer the drive letter of the partition I hid).

 

Of course if you plan to have all the time only one install of windows "visible" (by hiding the volumes containing the other instance) having the System marked as C:, no matter if it is the one on first or second partition, may be OK.

 

All other partitions (data) can of course (in both approaches) have the same drive letterings.

 

And yes, hiding a partition/volume, makes it have no drive letter assigned.

 

:cheers:

Wonko 

 

.



#60 sir_bootalot

sir_bootalot

    Member

  • Members
  • 41 posts
  •  
    Finland

Posted 23 February 2013 - 04:40 PM

I like the method of every Windows seeing itself as C:, and every data partition labeled with same letters from whatever Windows I boot from.

 

So hiding the other Windowses from each other, at least when booting the first time, would be the way to go.

 

Because I have to use Bootmanager to choose between Windowses (too bad Grub4dos can't load straight the Winload.exe), and obviously Bootmanager can't do the hiding job, it must be done by Grub4dos. But because I only have one entry in Grub4dos menu for all the Windowses (single BOOTMGR/BCD on P1 'System' partition), I can only come up with these choises:

 

Let's say four Windows 'Boot' partitions: P2, P3, L1, and L2.

 

A)

Four Windows entries in Grub4dos menu. The first one hides P3, L1, and L2 and unhides P2.

The second one hides P2, L1, and L2 and unhides P3.

And so on...

All the four entries load the same BOOTMGR/BCD, though, at the P1. So it's up to me to select correct Windows in Bootmanager, so that the right partitions are hidden and the one I'm booting into is not.

Pros: Fast to implement

Cons: Mistakes can happen

 

B)

Hiding the way like in A), but the menu.lst entry also selects one of four BOOTMGR/BCD combos.

I need to implement the floppy image approach, with every one of them different BOOTMGR/BCD

Pros: Error free bootup

Cons: Yet more hassle to set up

 

C)

Similar hiding with Grub4dos, and menu.lst entry also selects one of four BOOTMGR/BCD combos.

Here I'd go with my original idea of separate four BOOTMGR/BCD combos, this time located on the Windows partition I'm booting, so no common 'System' partition (P1), but every Windows partition is both 'System' and 'Boot' for itself.

Pros: Error free bootup

Cons: Even more hassle to set up (but should be possible)

 

Opinnions?



#61 cdob

cdob

    Gold Member

  • Expert
  • 1449 posts

Posted 24 February 2013 - 09:57 AM

how exactly configured is to be understood/tested, most probably a "normal" entry


XP at primary active partition

boot.ini
multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="XP: ACPI Multiprocessor PC" /noexecute=optin /fastdetect /kernel=ntkrmp.exe /hal=halmacpi.dll /NOEXECUTE=OPTIN /FASTDETECT /bootlog
NTLDR does read boot.ini and XP does boot.
BOOTMGR longhorn 5112 does read boot.ini and XP does boot.
Windows Boot Loader
-------------------
identifier {143b337c-7ad0-11e2-99d2-ab34df12bb4d}
device boot
path \windows\system32\winload.exe
description XP winload.exe testing
loadoptions DDISABLE_INTEGRITY_CHECKS
testsigning Yes
osdevice boot
systemroot \windows
kernel ntkrmp.exe
hal halmacpi.dll
nx AlwaysOff
pae ForceDisable
sos Yes
Vista gold bootmgr and winload.exe:
F8: Disable Driver Signagture Enforcement
textmode drivers are loaded
last driver loaded: mup.sys
Autochecks partition D: and C: without a errror.
No further reaction, system freeze.

Windows 7 gold bootmgr and winload.exe:
F8: Disable Driver Signagture Enforcement
textmode drivers are loaded
last driver loaded: mup.sys
Message: 'Please wait...'
No further reaction, system freeze.

Conclusion: anything is possible still.
Works in general, but not at tested hardware / software.
Not possible at all.
Possible with further adjustments.
I take a break at current state.

BOOTMGR longhorn 5112 at logical parition is the next question.
 
So hiding the other Windowses from each other, at least when booting the first time, would be the way to go.
Yes, hiding has some benefits. A not mounted partition is protected from yourself, accidential writing.
Contrary don't rely on virus protection, a virus may mount the partition..

Go for c)
bootmgr and boot\bcd at each partition. 
Adjust boot\bcd: set device and osdevice to boot.

#62 cdob

cdob

    Gold Member

  • Expert
  • 1449 posts

Posted 25 February 2013 - 08:46 PM

BOOTMGR longhorn 5112 at logical parition is the next question.
 
Hidden sectors 2048 at logical parition:
 
Info: An error occured while attemping to read the boot configuration data.
File:\BOOT.INI.
 
No, longhorn 5112 bootmgr dosn't read a default logical partiton. Hidden sectors has to be adjusted first.

#63 sir_bootalot

sir_bootalot

    Member

  • Members
  • 41 posts
  •  
    Finland

Posted 27 February 2013 - 10:37 PM

I got the things rolling, thanks to you guys in this thread (and some others, like diddy and the guy behind multibooters.co.uk).

I made a test setup with Win8 and some linux distro, multibooting them with Grub4dos. BOOTMGR and BCD copied from the small primary (P1) to the second primary (P2), that became combined SYSTEM/BOOT partition. I made a script to generalize BCD and also emptied MountedDevices before cloning. So the scheme C) from above was used.

Drive letter problems were solved the way I liked, so each Windows, cloned or original, sees itself as C:, and all the other partitions that Windows recognized have the same letters whether you watch them from the original Windows or from some cloned one.

I hide some partitions (that were made beforehand with Partition Wizard, using 63 sector alignment), that didn't need to have a drive letter before installing the Win8 from USB-stick, but IF I hadn't done that before the original Win8 install, that could be done afterwards by hiding them with Grub4dos, and then emptying the MountedDevices before cloning. (When the unwanted drives are hidden, and Windows boots the first time after emptying MountedDevices - clone or original - the drives that are not hidden get their letters in order.)
 

I made the (hopefully) final setup then after the test, again with a small 100 MB primary partition, but this time it was hidden during the original Win8 installation. It was type 1c aka hidden LBA FAT32 and it was made active (P1, that is). Now Win8 installed boot files BOOTMGR and BCD straight to the second primary (P2), that was now combined SYSTEM/BOOT. No need to copy the boot files and fix the BCD by booting the install-DVD (-USB actually) and edit the BCD. I again ran the generalizing script, tho, and emptied MountedDevices before cloning (with fsarchiver).

 

One odd thing I noticed. Grub4dos menu sometimes starts scrolling around, and doesn't respond to keyboard. I had only hit arrow down key a few times before it happened. (grub4dos-0.4.5c-2013-02-02-2)

 

Another odd thing: Disk Manager in Windows 8 shows extended partition starting from the first partition that is the type Windows understands and is not hidden. So it shows there are something like 7 primary partitions and a few logicals. I hope it doesn't mess them up.

 

 - sir_bootalot



#64 sir_bootalot

sir_bootalot

    Member

  • Members
  • 41 posts
  •  
    Finland

Posted 28 February 2013 - 09:17 PM

I can't make a full shutdown from Windows 8 that is on logical partition!

 

Restart works, but shutdown goes to that 'metrosexual' screen or whatever it is.

 

I cloned Win8 (with fsarchiver) according to the scheme C) described above, to another primary and to couple of logicals. (BOOTMGR/BCD are on the same partition, that is both SYSTEM/BOOT).

 

Full shutdown works with the clone that is on the primary partition. But not on those that are on logicals.

 

Any guesses what would be cause and fix to that?

 

The small primary P1 has to be active one when booting from logicals (with this scheme, BOOTMGR/BCD on logical too). I wonder if it has something to do with this odd phenomenom. And what would be the fix...

 

(And no, I haven't tried sleep, hibernation or hybrid sleep. Gotta go now to the place where the Win7 machine waits, so the Win8 machine would be on hold a few days. Let's see if the same happens with the Win7.)


Edited by sir_bootalot, 28 February 2013 - 09:19 PM.


#65 sir_bootalot

sir_bootalot

    Member

  • Members
  • 41 posts
  •  
    Finland

Posted 02 March 2013 - 05:24 PM

Tested now with Win7 machine. Similar setup, aka BOOTMGR/BCD on every combined System/Boot partition - original instal partition or cloned.

 

Shutdown works on logical partition clone Windows 7, at least in that laptop.

 

Sleep works. But hibernation or hybrid sleep dont. They just cause logoff. (Like shutdown in Win8 machine.)



#66 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 15103 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 02 March 2013 - 06:21 PM

Are "sectors before" corrected in those setups? :unsure:

 

:cheers.

Wonko



#67 sir_bootalot

sir_bootalot

    Member

  • Members
  • 41 posts
  •  
    Finland

Posted 02 March 2013 - 06:38 PM

No... I'll try it later (with Win7 machine - that didn't have the shutdown from logical problem, only hibernation issue from logicals) and report if it helps.

 

There was some stuff on hibernation probs at multibooters.co.uk but I think it didn't have it solved through. I'll check that later too.

 

(Testing Win8 machine will have to wait a few days untill I'm back there from here...)

 

WAIT...

 

I can't get hold of ptedit32.zip/ptedit.zip. Symantec's FTP site is down (for good, temporarily?). Anyone knows another site where to get it?


Edited by sir_bootalot, 02 March 2013 - 07:15 PM.


#68 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 15103 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 02 March 2013 - 07:41 PM

I can't get hold of ptedit32.zip/ptedit.zip. Symantec's FTP site is down (for good, temporarily?). Anyone knows another site where to get it?

Try using Beeblebrox instead.

http://web.archive.o...broxntsetup.zip

(cannot rememeber right now if it will do as well :unsure:)

Or use grub4dos as hinted before.

However the Symantec FTP is up and running, at least from here:

ftp://ftp.symantec.com/public/english_us_canada/tools/pq/utilities/

 

:cheers:

Wonko



#69 sir_bootalot

sir_bootalot

    Member

  • Members
  • 41 posts
  •  
    Finland

Posted 02 March 2013 - 09:06 PM

No, doesn't work their ftp server from here.

 

But, I do have Grub4dos in use.

 

Something like this when booting from logical:

 

root (hd0,4)

chainloader /bootmgr

makeactive (hd0,0)

 

(And those hide/unhide commands).

 

So correcting the 'sectors before' number wouldn't solve these issues, right? Or still worth trying?



#70 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 15103 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 03 March 2013 - 11:34 AM

No, doesn't work their ftp server from here.

 

But, I do have Grub4dos in use.

 

Something like this when booting from logical:

 

root (hd0,4)

chainloader /bootmgr

makeactive (hd0,0)

 

(And those hide/unhide commands).

 

So correcting the 'sectors before' number wouldn't solve these issues, right? Or still worth trying?

No, you need to go back a few posts, read them, then read the grub4dos readme.txt with reference to the command partnew.

 

You don't need to change the Active status of a partition if you directly chainload the BOOTMGR or NTLDR.

The Active status is ONLY useful for a "standard" MBR CODE (which chore is to ONLY chainload the PBR of the partition that is marked as active in the partition table, and thus *needs* to have a partition marked as active).

 

:cheers:

Wonko



#71 sir_bootalot

sir_bootalot

    Member

  • Members
  • 41 posts
  •  
    Finland

Posted 03 March 2013 - 01:16 PM

No, 

 

 

Do you mean with "No", that fixing the 'sectors before' wouldn't solve those issues?

 

I found the Partition Table Editor from Hiren's. (There's also some similar proggy for linux, some short tittle, can't remember it right now)

Haven't tried the fix yet, thought

 

If fixing the 'sectors before' might help, it would certainly be preferred to the trick with 'partnew' you obviously are referencing to.

 

About 'partnew' you wrote:

 

grub4dos can fix the Sectors before without need of PTEDIT or a similar tool, by using the partnew command (a free entry in the PT of MBR is needed, or you need a more complex workaround) or use the map --in-situ command to TEMPORARILY fix the problem.

 

 

I have only quickly glanced that trick yet (I'd rather solve these issues another way, if possible), but if I understood it right, it's a way to trick the Windows to think its a primary partition, although its on logical partition really - with aid of grub4dos partnew command.

(Or to put it another way: temporarily alter the partition table so, that logical partition becomes primary while we are in Windows that was cloned to logical partition). I'm i right or went to south?

 

OR... Did you mean that I'd _only_ use 'partnew' just for correcting the 'sectors before', like in this from another thread:

You can do the second thing (which is what you NEED to do) manually, if you want to save the hassle of doing it manually :

  • backup the MBR

  • run the partnew command

  • restore the MBR

the "Sectors Before" will be corrected by the partnew command and they will remain corrected


Edited by sir_bootalot, 03 March 2013 - 01:38 PM.


#72 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 15103 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 03 March 2013 - 02:02 PM

Look, what has been suggested since post #12 of this thread at least, is to fix the sectors before or "hidden sectors" in the BPB of Logical volumes.

Extensive resources on how exactly to do that were given.

 

The idea is that once you have those data correct (i.e. the offset in the BPB of the PBR of the Extended volume set to it's ABSOLUTE value) *anything* that will use that data will be able to find the PBR, i.e. the starting location of the volume on hard disk properly.

 

Whether this will affect the issues you found I cannot say, but it should have been the first thing you should have tried, before even posting about the issue.

 

The idea is that the *only* difference between a volume in a primary partition and a volume in an extended partition are:

  1. the addresses of the volume (partition) are stored in the MBR vs. stored in one of the EPBR's in the "extended chain"
  2. the address of the volume (partition) first sector (aka PBR) are stored as ABSOLUTE (with reference to the MBR or sector 0 of the device) vs. stored as RELATIVE (withe reference to the EPBR)

You cannot do anything about the first (actually you can, but that will bring us OT) but you can correct the second allright.

Just do it. ;).

 

About the possible OT deviation, if you are interested (mind you complex workaround :ph34r:) read here (BOTH the lines and "between" them :w00t:):

http://reboot.pro/to...the-same-drive/

 

 

Yes, partnew can be used as in the quoted text to just correct the address in the BPB, that's what I meant when suggesting it as an alternative to manually editing the PBR with PTEDIT or similar as grub4dos will calculate and fix the DATA automatically, of course the "plain" use of partnew implies having a partition entry in the MBR empty (and after using partnew re-zeroing it), otherwise you need to backup the MBR before and restoring it afterwards.

 

 

 

So correcting the 'sectors before' number wouldn't solve these issues, right?

NO.

This is the reply to "right?" -> NO, wrong.

and also the reply to "So correcting the 'sectors before' number wouldn't solve these issues?" ->NO, correcting the sectors before will most likely DO solve those issues.

If you prefer:

 

 

Would correcting the 'sectors before' number solve these issues?

YES, most likely. 

 

:cheers:

Wonko



#73 sir_bootalot

sir_bootalot

    Member

  • Members
  • 41 posts
  •  
    Finland

Posted 04 March 2013 - 12:16 AM

Well, it wasn't the fix. The 'Hidden sectors' field was already fixed on those logical partitions, where I hade restored the clone with fsarchiver. (Meaning fsarchiver does the job)

 

Anyway, because those Vista+ Windowses are planned to be installed on logical partitions, the cause of those issues (no hibernation, and no shutdown for Win8) is probably something else.

 

So because clone restored to another primary partition works, but clone taken from primary and restored to logical partition doesn't, the cause must be something like:

 

BOOTMGR, BCD and Winresume.exe are involved in hibernation process. When BOOTMGR and BCD are transfered to logical partition, some of those components misbehave. Why?

 

With Win8 shutdown process has something to do with hibernation also. I'd guess. Restart works, tho.

 

IF the solution would be just to use another boot manager (perhaps Bootit Bare Metal would do), then there must be some fix to be found, because they have found it.

 

http://multibooters.co.uk/cloning.html

It is this auto specializing of the filedevice Element just before hibernation shuts the system down that is part of the cause of hibernation or hybrid sleep problems in independent Vistas on second or higher hard drives or logical partitions. If your bootmanager is not fully Vista compatible then the wrong BCD store can be set for resume, see BCD is Always Open. If you do want hibernation to function in independent Vista installs on anything other than boot drive primary partitions, then I have found no solution other than making sure your bootmanager is fully Vista compatible.

Edited by sir_bootalot, 04 March 2013 - 12:28 AM.


#74 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 15103 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 04 March 2013 - 09:01 AM

No.

Meaning that BOOTMGR is at the same time a boot manager and a boot loader (or OS loader).

*any* third party boot manager will behave (for Windows Vista :ph34r: and later) ONLY as bootmanager and will need before or later to chainload the "original MS" BOOTMGR (using it as a pure OS loader).

 

But I am missing if you already attempted the re-specializing of the \boot\BCD on those problematic partitions as suggested on the multibooters.co.uk page. :unsure:

As said, it is possible that there is something "connected" to the "cloning procedure". :dubbio:

You could try (after having backed it up) to reinstall on one of those volumes and see if the hybernation with a "fresh install" behaves the same.

 

:cheers:

Wonko



#75 sir_bootalot

sir_bootalot

    Member

  • Members
  • 41 posts
  •  
    Finland

Posted 04 March 2013 - 05:29 PM

(I made one error before, but the situation is still quite the same. It still does not hibernate / resume from hibernation correctly. The error was: when doing the testing, I made multibooting with MS Bootmanager before installing Grub4dos. And then I modified the setup to use Grub4dos/menu.lst and messed up something so that clone used wrong BCD. I noticed that when comparing results of:

bcdedit

bcdedit /store c:\boot\bcd

And the plain 'bcdedit' command listed the old, multiboot version of output

I fixed the situation by making fresh new installation according to the scheme C) described earlier )

 

So after the fresh setup, now the situation is: it does hibernate (or seem to), but doesn't resume correctly. When resuming, it displays that Error during recovery or something screen, and you can select Safe mode or Startup Windows normally.

 

So either hibernation and/or resume from hibernation does not happen correctly.

 

I tried respecializing BCD (using C:, not HarddiskVolume4 way). But it didn't help.

 

So I'm gonna try what you suggested, another install from DVD/USB, this time straight to logical partition (second logical I think, to keep the clone on the first logical for possible future testing). Going to keep the P1 hidden as type 1c aka hidden FAT32 LBA, that should put the BOOTMGR/BCD straight to logical partition. If it doesn't work, then unhiding the P1 before anoher fresh install to L2, so that BOOTMGR/BCD go to P1 that becomes 'System' and moving BOOTMGR/BCD this time afterwards to logical partition L2.

 

Let's see.

 

I'd like to see hibernation to work correctly also on cloned logical System/Boot Windows for two reasons:

1) If hibernation does not work correctly on logical, perhaps there is also something else that does not work correcly on cloned Windows on logical

2) On laptop, like my Win7 setup, hibernation is nice, if you use it without power. Even with my huge size laptop that I use only places that has power outlet (size and 1.5 h max working time when unplugged), I have sometimes forgotten to plug the power cable

 

One question:

Let's assume, I would have a multiboot Win7/8 setup where the hibernation works on all the instances of Windowses.

If I hibernate WindowsA, having open file handles on some common data partition, and while WinA is still in hibernation,

I boot WindowsB from Grub4dos (forgotting which one I used when it hibernated, or forgot I had used the hibernation the last time

I used the computer), and then open the same file (from WindowsB) that was  in use from WindowsA when it hibernated,

this would mess up the file / data partition, right?

 

If so, hibernation in multiboot environement should be avoided, anyway...

 

Hi wimb! Any ideas?


Edited by sir_bootalot, 04 March 2013 - 06:25 PM.





0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users