Welcome, Guest. Please login or register.

Author Topic: Drives in HDToolBox not detected full size (CHS settings?)  (Read 11254 times)

Description:

0 Members and 1 Guest are viewing this topic.

Offline olsen

Re: Drives in HDToolBox not detected full size (CHS settings?)
« on: December 29, 2016, 01:15:50 PM »
Quote from: LoadWB;818320
I am attempting to install a couple of different SSDs into my 4000D.  These are KingSpec IDE SSDs, one 64GB and one 128GB, which work just fine in other machines, as well as with USB-to-IDE adapters.

When installing in HDToolBox they do not show full capacity.  They are initially seen as their proper capacity in the selection window (minus over-head,) but upon installation the 64GB shows up as 18GB, the 128GB shows as 22GB.


Which HDToolBox version are you using? Would that be the Workbench 3.5/3.9 version or the original Workbench 2.x/3.x version?

I am asking because the Workbench 2.x/3.x version (that would be V39; the partition size calculation code changed very little since the Workbench 2.x release of HDToolBox) may not correctly display partition sizes greater than 4 Gigabytes. The calculations which yield the partition layout generally work correctly, it's just that displaying the sizes of the partitions is flawed. These problems cover both displaying the size as numbers ("2000 Meg") and showing the sizes and positions of partitions as clickable/movable blocks. Don't click on and resize these partitions: select the partitions using the left/right cursor keys, then change their positions/sizes by editing the number of cylinders (lower left corner of the window) instead.

The layout calculations are based upon the number of cylinders involved, which is mostly safe. For display the number of cylinders need to be multiplied by the number of sectors per cylinder and the number of bytes per sector (the HDToolBox V39 user interface refers to sectors as "blocks"). The result of that calculation easily overflows with storage devices in the >4 Gigabyte range, resulting in wildly implausible partition sizes.

I wrote "mostly safe" and "generally work correctly" because HDToolBox V39 may stumble when choosing the "default disk layout", which involves creating two partitions of equal size for the medium. You might get better results by manually deleting one partition, tweaking the size of the remaining partition by editing the number of cylinders allocated for it, then repeating the process after adding another partition.

Quote
Any advice on adjusting the CHS values to match the proper capacity?  Would bad values cause problems in use after partitioning, formatting, and OS installation?


As far as I can tell you should not deviate from the number of cylinders, heads and blocks per track which HDToolBox "divines" by probing the drive.

It pulls together the drive's properties from three different sources of information, any of which may be somewhat "off". Historically, this was done because for some 10-15 years hard disk manufacturers were prone to create wonky firmware which reported incomplete, inconsistent or plainly wrong information about the drive (the HDToolBox code is chock full of profanity, cursing various hard disk vendors; most of these went out of business or were gobbled up in the decade that followed, so maybe those curses did have an effect after all). I guess that pattern now repeats with SSDs, as it did with each innovation that came to the mass storage market in the 1980'ies and 1990'ies (fast SCSI, notebook drives, IDE, wide SCSI, ATA, etc.).

So anyway: HDToolBox V39 will err towards making reasonably consistent size figures when setting up the disk's size. These calculations take into account that the drive may reserve part of the medium for its own use (e.g. for bad sector remapping), or may report the number of sectors available as a figure which doesn't match the number calculated from the cylinders/heads/sectors.

This code tends to work out OK, as far as I can tell. It may not give you everything which the medium is supposed to provide, but it's certainly safer than trusting a drive which claims in one set of data that there is storage space available which cannot be accessed, according to a different set of data produced when probed.
« Last Edit: December 29, 2016, 01:25:36 PM by olsen »
 

Offline olsen

Re: Drives in HDToolBox not detected full size (CHS settings?)
« Reply #1 on: December 29, 2016, 01:38:16 PM »
Quote from: LoadWB;818361
Ah, that is a good question which I punch myself for leaving out.  I am using HDToolBox provided by a 3.9 Emergency Boot disk set.  I am pretty certain this is a non-BB CD but I will get the version output of HDToolBox when I get back to the 4000 later this morning.  I had originally tried using the 3.1 installation disks but v39 refused to acknowledge a device was present at all.

I actually have not proceeded with the default detection, yet.  My concern was the ridiculous difference between the size of the device and what HDToolBox said was its capacity.  I certainly do not want to only be able to use 20GB of a 128GB drive, even if I will never have that much stuff on this computer :)

I just recalled that there may be another punch line here: the drive may report its size differently with respect to how it is asked.

The ATA scsi.device may receive an answer fit for a disk controller as present in an early 2000's PC. Back then the size of a sector was 512 bytes, and this is no longer the case with today's SSDs. Also, the number of sectors per track, etc. was much lower than what we use today.

Hence, the SSD may report its properties skewed towards "compatibility".
 

Offline olsen

Re: Drives in HDToolBox not detected full size (CHS settings?)
« Reply #2 on: December 29, 2016, 02:08:37 PM »
Quote from: kolla;818364
The whole disk device driver + filesystem of OS3.x needs an overhaul... again, lol :p


It's not that terrible, or at least it didn't get much more complex and confusing since 2001.

The disk driver API still supports disks of up to about 1 Exabyte. Partitions larger than 4 GBytes are still unwieldy and the FFS still scales poorly in terms of performance when used on partitions of that size.

Same old, same old...