I remember someone tried to boot windows 3.1 from an EXT2 partition and, as one might guess, it didn't boot.
It doesn't matter that we have an EXT2 driver for Windows XP if we don't have a bootloader that is capable to boot windows from an EXT2 partition.
The bootloader HAS to be capable to boot WINDOWS from an EXT2 partition, it has to be specifically developed to boot windows from such a partition.
Taking NTLDR and trying to boot windows from an EXT2 partition won't work, taking a GRUB or LILO and trying to boot windows from an EXT2 partitin won't work either.
Anyone is free to try, but some have already tried and Windows didn't boot.
At least the driver for the desired partition type and a bootloader capable to boot a windows installed on such a partition type (written specifically for this purpose or a modified version of a bootloader which is licensed under GPL or BSD like GRUB or LILO) is required for this thing to work.
Just putting a driver for a new filesystem in Windows won't make NTLDR boot Windows from such a partition.
The fact that EXT2,EXT3, REISERFS, REISER4 or any other FS is the fastest on Linux, it doesn't mean that's certain to work under windows as fast as it does under Linux.
There's no XFS driver for Windows for the time being.
I will make further investigations. Information about what happens during the windows boot process would be appreciated.
Grub and lilo give back the task of booting windows to NTLDR.
Grub does that by chainloading it (chainload, boots windows by first booting NTLDR which actually boots Windows) and lilo does it by setting an "other" tag on the boot entry (another term for saying that it chainloads Windows).
This post has been edited by razvannh: 31 August 2005 - 03:12 PM