[hatari-devel] Improved timings for the FDC emulation

[Nicolas Pomarède <npomarede@xxxxxxxxxxxx>]

Hello

I committed a new version of fdc.c with many internal changes.

This should give very few compatibility's improvements compared to 
previous version, as I don't know of many programs which had problems 
with FDC's emulation, but it should be much closer to the WD1772 
datasheet and avoid future incompatibilities with usage not yet encountered.

The changes are :

 - much improved status register and commands' behaviour when the drive 
is off, or not selected or there's no floppy in it. Also, correct 
timeout (or no timeout at all) for those case (most commands should wait 
forever when a drive is off or empty).

 - more accurate index emulation ; delays are now counted using 
indexes, instead of using the corresponding number of expected cycles 
(which didn't take into account the case of disabled/empty drive)

 - much closer to the statemachine described in the WD1772 doc. Timings 
were OK, but not always applied at the exact place or in the correct 
order. This would certainly not have made a game/demo crash before, as 
those timings are hard to measure exactly on a real STF, but still, 
experiment prooved it's always better to stick to the HW as close as 
possible.

 - some code's reorganisation to try to have a general emulation's 
logic that follows all the steps described in the WD1772's doc.


I tested many demos/games I use as reference for FDC's emulation and 
didn't see any problem so far, don't hesisate do test some of your 
favorites ones (from my experience, games (even if cracked) are often 
more likely to show errors in FDC's emulation)

(one game not working so far is "Knightmare" on DBug 24, the way it 
worked in Hatari 1.7 was not correct anyway, so this needs more 
investigation).


Nicolas