On Thu, Mar 8, 2012 at 12:57 PM, Dave Dunfield <dave12 at dunfield.com> wrote:
2) By changing the track sector-size field to a
formerly invalid
value, the existing tools will report "Bad sector size" instead
of "getting lost" with the unexpected extra data in the track
and generating random errors.
Ah yes, that does seem a better idea :)
I haven't actually tried, but my read on the NEC
765 controller
manual suggests that the PC controller can't format mixed-
sector sized disks.
Not directly, but the size code in the format command determines the
physical size written on the track for each sector, independent of the size
values in the ID data. So you can create a format template from smaller
sector sizes to position the sector headers, then write to the data fields
of the real sectors to complete the track. In cases where the header size
is larger than the template sector size, the extra headers will be
overwritten by the data fields.
The only control you have is the format size and gap3 value, but it's
enough to achieve the mixed sector size combinations I've seen so far. One
popular format is 5x1024 + 1x512, to squeeze a bit more on to a 250Kbps MFM
track. For that you could use a template of 17x256 with gap3 around 48.
Each 1K sector spans 3x256 sectors, plus the accumulated gaps, with the
final sector spanning two. The ID header values for the dummy sectors
aren't important, as long as they don't match one of the real sectors.
Calculating the unit and gap sizes to use for a general case is a bit more
work than normal, but I'd imagine you could have a fixed list for known
formats (as is done for some gap values).
In other words, this is an optional extension to the .IMD file
format specification - My tools probably won't
implement it any
time soon. I added it specifically so that others who needed this
capability could support it in a consistant way.
That's good enough for me to include it -- thanks!