Changes
/* Alignment of HSYNC Interrupt */
In r005.8.7, arnoldemu testbench "cpctest" is OK
==== TODO : Vadjust lines ====
[http://www.6502.org/users/andre/hwinfo/crtc/crtc.html CRTC operation]
Only R5 still needs to be explained. To allow a finer adjustment of the screen length than by the number of character lines (R4), R5 adds a number of blank scanlines at the end of the screen timing.
Vadjust is adding lines at end of v_counter, so after V-SYNC. So it is adding a black line. Question : are all Vadjust line black ones ?
In r005.8.12, -LittleOne demo raster arrives one line too late, and prehistoric II seems hiding line one line too late resulting a pixel-garbage line. This same pixel-garbage line are visible in somes demo, like Blocus, Batman Forever (several times one pixel-garbage line) and battro (10 following pixel-garbage lines in this last one)
Some hypothesis :
* VRAM is not writing pixels at this moment, resulting on a "not newly written line" (badly out of DISP)
* palette offset is badly shifted
* bad HSYNC interrupt
* bad v_counter range
* instruction timing problem
=== TODO : rupture ===
