Everyone should have an expecation of zero user-visible faults. The diagnostic equipment at the factory will map out any defective sectors. Thus, any bad sectors that the user sees are “grown” faults. Either there’s some physical damage to the media (in which case the number of faults will likely increase over time) or there was a glitch during the writing of a sector (such as a power failure) which corrupted the on-disk format.
If a drive detects a bad sector on a write, it will re-allocate the sector to one kept in reserve. If the defect is detected on a read, the drive will use a number of internal methods to try to correct the data. If the error is uncorrectable, it will be reported to the operating system. A correctable error will also be reported via one of a number of different mechanisms, but a user-space program won’t see those reports unless it goes looking for them.
Any software tool that claims to “repair” bad sectors is (at best) trying to force the drive to reallocate the sector. The drive’s internal hardware / software has far more sophisticated error recovery available to it than anything available at the operating system or application level.
In the old days, the operating system needed to map out bad sectors. DEC used standard #144 (BAD144), for example. Disks would come with labels on the case giving cylinder / head / bytes from index (or sector #) and users or service technicians could manually enter that data in an on-disk table.
However, once the processing of sector data moved onto the drive (as opposed to the drive just sending streams of data to the controller and letting the controller sort it out), drives became logically perfect (no bad sectors reported to the OS) even if the underlying physical media had defects. This happened nearly 30 years ago, in SASI. For some years after that, it was still possible to do a “real” field format of a disk drive, which would rewrite the tracks and possibly correct marginal sectors. However, physical formatting became unavailable due to a number of advances in hardware - initially, the wedge servo and later, zone bit recording - meant that a true format became more and more complex, until it was only possible to do it at the factory when the drive was manufactured.
I disagree. I don’t know of any drive manufacturer who will refuse a warranty RMA for even a single grown bad sector.
Background - I’ve worked with quite a few of the original ST-506 drive designers, written firmware for some of the first SASI controllers, and worked on the design of several disk drives. Here is a 20+ year old thread talking about (among a vast number of other things) old-time bad sector handling.