Eugene's observations

When I tell a client that I have fixed their PC and that damage was caused by ?bad blocks?, it is shorthand for the following.

The hard disk drive on a computer holds data even after power is turned off. This is done by magnetising the disk in a way that is reversible on demand. The part of the disk that holds the magnetised data is a platter, and if the platter is no longer able to hold the data reliably, that data is lost and that part of the platter is permanently unusable.

A disk platter is divided into sectors (which were originally known as blocks) of typically 512 bytes of data, where a byte is usually 8 bits and a bit is the smallest unit of data ? of either positive or negative polarity.

When data in a sector is detected as bad, the whole sector can be mapped (also known as resectored, reallocated, or revectored) to a hidden file that only the disk controller circuitry knows about. The operating system (e.g. Windows) does not know of the sector so marked and therefore it can continue using the rest of the disk without problem. The disk circuitry will never release the bad sector for use.

Marking sectors as ?bad? is an anticipated part of a disk?s lifecycle, and the manufacturer will (conceptually) allocate a file to point to such bad blocks when the disk is formatted. This is one of the reasons why Windows always sees fewer than the full number of gigabytes as available.

The data that was contained in the bad sector may cause a spreadsheet to become corrupted, a program to stop working or many other apparently spurious faults depending on what data happened to be on the block before it went bad. If the data was part of Windows, you may fail to be able to even start your PC.

So when I tell a client that I have repaired the fault, I mean that I have caused the controller circuitry in the disk drive to allocate the bad sector to a hidden file to prevent its reuse, and resolved the program or file that was held there so that it is no longer seen as corrupted. This may involve restoring from your backup (you do take backups don?t you ?) or even reinstalling Windows.

A single bad block event is an inconvenience, but if it becomes one of a series, you know that the disk is on the way out and it would be prudent to [have me] clone the disk to a new one while it is still possible.

The manufacturer allows for bad blocks in a table held in firmware on the disk, but there comes a point where that table is full and at that time the disk will fail catastrophically. Hence, it is wise to replace the disk when the table is almost full so that data can be saved.