A. If Windows NT is installed on a logical drive in an extended partition (the 4th partition is usually the extended start), after you select the OS choice and NTDETECT runs, this error message will appear:
Windows NT could not start because of a computer disk hardware configuration problem.
Could not read from the selected boot disk. Check boot path and disk hardware.
Please check the Windows NT Documentation about hardware disk Configuration and your
hardware reference manuals for additional information. "
The Master Boot Record consists of boot code that is used by the system BIOS to read the partition table. From data contained in the partition table, the MBR can determine which partition is set to be bootable (active) and also the starting sector of that partition. Once that location is determined , the BIOS jumps to that sector and begins the next phase of the boot process by executing additional code that is operating-system specific.
If you have files required for boot located above 1024 cyl, it will fail. If you're running scsi, there's a chance you can get around it by using the scsi driver as ntbootdd.sys. If you're on IDE you're out of luck.
Windows NT 5.0 gets round the boot failure if any files needed for boot are above cylinder 1024 with an updated NTLDR. This file can be copied to a Windows NT 4.0 installation on the active partition without any ill effects, just make sure you have Service Pack 4 applied to the system before copying the NT 5.0 NTLDR.
If the only thing wrong with sector zero was that the last two bytes are not 55AA, this can be fixed with a disk editor such as Norton Diskedit. However, this message is usually indicative of something overwriting or destroying the entire boot sector (sector zero) including the partition table entries.
When you install Windows NT on a logical drive in an extended partition - OSLOADER needs to "walk the extended partition table" through BIOS calls in order to get to the partition you have Windows NT installed in. Each of these logical drives are addressed in a "daisy chain" of partition tables. Each sector that contains a partition table entry MUST end with a 55AA as the last 2 bytes in the sector.
The best way to determine how to recover is to use a disk editor to see if the partition table entries are still intact. Each sector occupies 512 bytes. The first 446 bytes of sector zero contain the MBR boot code followed by the partition table entries, and ends with 55AA. If the partition table entries are still intact at offsets 1BE through 1FD, manually record their values, then write 55AA starting at offset 1FE. Once the signature 55AA is written the MBR boot code can be regenerated by using the Fdisk.exe program from MS-DOS version 5.0 or later.
WARNING: This process will repair the bootstrap code and the 55AA signature by rewriting sector zero but will also overwrite the partition table entries with all zeros, rendering your logical drives useless (unless, that is, the 55AA signature is manually entered using a disk editor prior to your performing the FDISK /MBR).
If the partition table entries are not intact or were overwritten with unreadable characters, the problem is more involved and entails locating the master boot sector (MBS) for each partition and manually rebuilding the partition table entries. This process is beyond the scope of this article.
To speed recover from future MBR corruption, use the Windows NT 4.0 Resource Kit utility Disksave.exe to save a copy of the MBR to a floppy disk. This can be used if needed at some future date to restore the MBR using Disksave.exe.
In the case where Windows NT is installed on a logical drive in an extended partition, you will need a disk editing utility like Norton Diskedit to examine each sector containing an extended partition logical drive entry to make sure it ends with a 55AA. This process is beyond the scope of this article.
A virus in your boot sector may also cause this problem so run an anti-virus program on your boot sector if in doubt.
Most of this information is from Knowledge base article 149877.