China DOS Union

Sender: dj (Doctor Jia), Board: DOS
Title: More on Low-Level Formatting of IDE Hard Disks
Posting station: BBS Shuimu Tsinghua Station (Wed Aug 27 17:17:47 1997)

I don't have time to get online, and I'm bringing up an old topic that has already been discussed. Sorry about that.

In an earlier POST I roughly mentioned that for IDE hard disks, low-level formatting is meaningless. Since at the time I could not
write in Chinese, the reasons were not stated fully enough. Below I will discuss it in detail.

1.What is disk low-level formatting: In order for a disk to achieve random access, sectors need to be laid out
on the disk's tracks. Each sector uses a lead-in mark and a sector mark as the start of the sector, and only then comes the contents
of the sector. Low-level formatting is merely putting these marks on the tracks. So the low-level formatting operation is actually only
a simple write process; what is written is not data but marks. Low-level formatting is neither equivalent to finishing the platter,
much less regenerating the platter. Both floppy disks and hard disks have a low-level formatting process, and this process is needed, and needed only once.
This is why after DOS 5.0, when FORMATting a floppy disk, if /U is not specified, it only performs verification. For IDE
hard disks, this process has already been done after HDA manufacturing is completed; at that time the hard disk has not even taken shape yet!
2.Why old-style hard disks need low-level formatting: Old-style hard disks refer to those using the ST506/412 dual-cable interface.
This kind of hard disk has a simple structure; at best it can only be considered an HDA with a simple signal regeneration circuit added. The main disk data
generation and recovery work all has to be handed over to the controller card on the host. Controller cards had many variations; in terms of encoding format
alone they were divided into MFM and RLL. MFM is simple and cheap, but the density is low, with only 17 sectors laid out on one track.
RLL is complex and expensive, but the density is high, with 26 sectors laid out on one track. A 40M hard disk using an MFM controller card, after switching
to an RLL card, could reach a capacity of 60M. This made hard disks not interchangeable between different kinds of controller cards; a disk originally
on an MFM card would be completely unreadable and unwritable when moved to an RLL card, because the disk formats were different. Hard
disk manufacturers simply left the task of low-level formatting to the controller card manufacturers. According to their own
needs, controller cards could format hard disks into all kinds of structures. Many large hard disk manufacturers were also large
controller card manufacturers, such as West Digit. Users could buy one company's hard disk and another company's controller card and combine
them for use, as long as they were both ST506/412 interface. WD's controller card could format a SEAGATE hard disk
into a disk structure with WD characteristics.
3.Why IDE hard disks do not need low-level formatting: In view of the poor compatibility between disks with the ST506/412 interface, a disk from this machine
might very likely be unreadable on another machine. The ST506/412 port also had limited ability to support disks above 100M, so large manufacturers such as COMPAQ
agreed on an IDE interface standard, requiring hard disk manufacturers to put the controller card and the hard disk
together, using an IDE port, an interface similar to a computer bus, to connect to the control part on the hard disk. According to this
standard, the simplest IDE port merely uses buffers to send the signals on the computer bus directly to the IDE port, while the hard disk
integrates all components including the ST506/412 controller card and the hard disk. This method greatly improved compatibility
between IDE hard disks. This was probably also one of COMPAQ's original intentions in initiating this interface standard. However, the consequence of doing
this was that the electronic control part on the hard disk became particularly complex, and some companies that only made hard disk controller cards but did not produce HDA
went out of work. Because the control part controls only a specific HDA, its functions were also appropriately simplified; support for low-level
formatting is one example. As described in 2, support for low-level formatting was originally provided by the controller card in order to adapt to HDAs produced by different
manufacturers, while the control part of an IDE disk controls only a specific HDA. A certain HDA is used only in products from its own manufacturer,
and there is no need to consider compatibility with other manufacturers' control parts. Therefore, when the HDA has just been produced and has not yet
been connected to the control part, the manufacturer's special machine is used to perform low-level formatting, and unqualified ones detected are
eliminated. After a qualified HDA is connected to the controller board, they will never be separated, so the controller board no longer needs to have the
function of low-level formatting.
4.What low-level formatting of IDE hard disks does: The current AWARD BIOS and the ADM 8.0 mentioned by Yong both claim
to be able to do low-level formatting on IDE hard disks. This doctor was not willing to take a good hard disk to try it, but the following guesses can be made:
First, the designers of the controller on an IDE hard disk would not add redundant
hardware to the controller board for low-level formatting that is no longer necessary, thereby increasing cost. Second, in order to simulate the process of low-level formatting,
an IDE hard disk can make outsiders think it has done low-level formatting by filling sectors with the same value.
This may even be the trick played by AWARD BIOS and ADM 8.0.
Experimental verification methods are as follows:
(1) Whether the low-level format software provides an item for changing the interleave factor in the formatting settings. If it does not, then it is definitely not
true low-level formatting.
(2) After selecting different interleave factors, whether the hard disk transfer rate changes. If it does not change, then it is definitely not true low-level formatting.
(3) When formatting the disk, select one kind of disk parameters (C)ylinder,(H)ead,secto(R). But when using it, select another
C,H,R. If no error occurs in disk access, then it is definitely not true low-level formatting.

If any expert is interested, you might as well give it a try. POST the results and we will know the answer.
--
Doctor Jia With a wine flask I help the world, a true miracle doctor

※ Source: ·BBS Shuimu Tsinghua Station bbs.net.tsinghua.edu.cn·

(This article was copied using the S-Term article copy script)
==================================================