NTSC/PAL border setup at frame access, cleaner debug output.

[hernandp]

Changes:

• per-frame video mode setup, the sane way to do it.
• legacy registers interpret routine did the vertical border setup before, this was bad as HOTREGS is not required to change when NTSC/PAL is switched, but borders must be setup.
  Now I partitioned the old routine into vertical border, horizontal border and legacy registers.
  Moved them at top to mantain static declarations.
  And as a side effect, you will see cleaner debugging information.

[lgblgblgb]

Thanks, I'll have a look once I have some time to do so (hopefully tonight, ehm, I meant in European time or such ... as usual).

[lgblgblgb]

Have you put this part:

if (!in_hypervisor) {
    vic4_update_sideborder_dimensions();
    vic4_update_vertical_borders();
}

into vic4_open_frame_access? I have a worry about this ... Maybe the PAL/VIC change by register write happens some time, but for the time, when a new frame is started, it's already not true that we're still in hypervisor mode, or vice-versa. That was the problem of mine, when I mentioned that I'm really out of ideas how to manage things in a "split situation" like this (split in time, I mean, the effect is not happening when the write itself).

[hernandp]

Have you put this part:

if (!in_hypervisor) {
    vic4_update_sideborder_dimensions();
    vic4_update_vertical_borders();
}

into vic4_open_frame_access? I have a worry about this ... Maybe the PAL/VIC change by register write happens some time, but for the time, when a new frame is started, it's already not true that we're still in hypervisor mode, or vice-versa. That was the problem of mine, when I mentioned that I'm really out of ideas how to manage things in a "split situation" like this (split in time, I mean, the effect is not happening when the write itself).

We should not check of hypervisor presence as of yet, and simply let the video mode setup go through at vic4_open_frame_access. I think we can remove that check and annotate a reference to the topic for later.

[lgblgblgb]

Probably. I think we can't figure out a 100% correct solution, works also after a million years from now ;) But, maybe this is not even very frequent that somebody changes video mode. At least I wouldn't find that useful to try to set NTSC/PAL at every second frame or so (a modern LCD monitor may even blank on video mode change for seconds anyway ... at least for more than zero frame). One thing that can be interesting that it seems freezer may use another video standard than the "user program" so there can be a true "danger" of video mode transition on hypervisor enter/leave boundaries. But again, I "blog" these here that we can refer it later maybe. On shorter term, the goal now to have something which in usual cases works (and anyway Xemu cannot use freezer currently, that's another point ...).

[lgblgblgb]

I've merged now, honestly I don't even know why I waited with this, since we have no "danger" here to merge anything, it's a kind of "wild" branch ;)

[hernandp]

No problem at all with commenting, waiting, reasoning about this etc. I appreciate your opinions very much.