Matches my AMD Stoney + S3 Trio64V2, and also Intel's 2023 GPU docs (which *still* tend to be more accurate than IBM's), and makes more sense than what we've (I've?) been doing.
1. As tiring as it seems, XOR properly the passthrough bits for mode changes. Fixes everything tested as of now (Mach8/32).
2. Implemented separate destination and source GE pitch and offsets. Fixes Windows 3.1 mach8/32 drivers using devicebitmap enabled (or dmp=1).
3. Properly fixed high and true color patterns so that the aforementioned patterns are drawn correctly in various stuff.
4. Implemented in the best possible way the CRT offset (currently in on conjunction with the GE offset).
5. On ScanToX with bpp set to 24 and dpconfig 0x6211, optmize the SRC/CUR X-Y coordinates at the end of the blit.
6. For mode changes, don't call the mode change recalctimings function if we aren't in the appropriate bits of the respective ports.
7. Separate 8514/A compatible and Extended foreground/background selection implemented, fixes Windows 3.x ATI Ultra drivers from 1992 without hacks.
8. Clear the busy flags when needed without stalling the entire guests (but not the emulator itself) (Mach8/32).
9. The MMIO regs should be enabled only when the VGA aperture has reached a graphics mode (0xA000). This fixes NT 3.x mach drivers in true color mode and making Win3.1 drivers working without a hitch at the same time.
10. Actually implement the hardware 4bpp cursor properly on Mach32.
11. Cosmetic cleanups.
1. Use proper identification to what chips use what.
2. Apply some mode switch fixes to the ATI 8514/A Ultra and make 1024x768 87Hz interlaced the default mode if htotal is 0 and on ati8514_init.
3. Add the undocumented ports to the ATI 8514/A Ultra add-on as well.
SVGA related:
If CRTC1/HDISP is odd (if bit 0 is set), then make it even accordingly, else leave it as is. Should fix some display skews.
S3 related:
Remove old code references (can still be accessed through history if one needs it).
8514/A compatible related:
Proper use of the htotal and hdisp timings.
XGA:
1. Added a proper poll so to have its timings independent of the SVGA core and changes its renderer accordingly, mainly a blank render one.
2. Workaround the Ctrl-Alt-Del reset of Windows 2.x so that it switches back to VGA text mode correctly when using the 0xA000 64K aperture (in vid_svga.c).
SVGA core:
Re-organized the way the different timers are organized so that the XGA and 8514/A pollers can coexist with the SVGA timer without conflicts.
ATI Mach32:
Reworked the true color renderer and made it a Mach32 exclusive due to its differences from the standard one. Fixes wrong rendered colors in NT using the 32bpp renderer in RGB mode.
Generic SVGA layer:
Added function pointers of the banked mapping for use with add-on cards with their own mapping when the VGA card banked mapping is not active or viceversa (e.g.: XGA).
XGA-1/2:
1. Reimplemented Area Fill and Boundary Mode as best as possible.
2. Fixed conflicts with banked mapping with VGA clones.
3. Fixed inverted colors (again) on accelerated 16bpp mode under OS/2.
Video7 with ATI 8514/A add-on.
Added a workaround (BIOS issue? I don't know) that disables 8514/A mode and reenables VGA mode when needed. Fixes screen freezes and polling issues with various drivers for Windows and others.
1. Finally add the ATI 8514/A Ultra add-on BIOS in both ISA and MCA variants (despite the ROM page failure on POST).
2. Fixed a mode switch bug in the Mach8 (both add-on and vga combo) when exiting the Win3.x GUI using the native drivers.
3. Corrected (best possible) the eeprom load of the Mach8 products.
4. Partial implementation of the FIFO on the 8514/A compatible products including Mach8/32 and plain IBM 8514/A.
1. vga_on global variable removed, as it didn't play well with 2 subsystems at once (8514/A and XGA both enabled).
2. Emulate the Foreground/Background Color aliases of PIX_TRANS properly when not executing a command.
3. Voodoo 3D override now works properly (again) with Mach32 PCI cards and others by turning the 8514/A timer off and on accordingly.
8514/A/Mach8/32:
Reworked the acceleration a bit as well as the mode switches from VGA to ATI/8514/A mode and viceversa based on the documentation. Fixes the Linux GUI fonts using the Mach32 driver (possibly Mach8 too) and other anomalies (still needs more work on the acceleration though).
PVGA:
Minor fixes to the banking.
S3-based (Pre-ViRGE):
1. Made the chip class use its own banking so that the Enhanced Mode mapping (equivalent to 64K-based A0000) is taken into account (per bit 3 of CRTC31, which forces the mapping to be 64K A0000, regardless of the GDCREG6 bits). Fixes NT 3.1 347.1's S3 driver.
2. Initial rework on 15bpp/16bpp accelerated mode of the 911/924 chips (currently not 100% bug free, I need help with this chips).
XGA-1/2:
Get rid of the linear endian reverse variable hack used by OS/2 and, instead, use the already declared ones more wisely, fixes OS/2 2.1 reversed fonts and keeps everything else working as it should.
Misc:
Added the BT481 RAMDAC for future card use.
8514/A changes:
1. Correct interlaced display resolution.
2. Added a limit to cursor coordinates.
3. Test/WIP features of the add-on Mach8 side (ATI 8514/A Ultra) such as configurable BIOS.
4. Made the CMD 5 of the acceleration (Polygon Boundary) more accurate per manual (as much as I could regarding the clipping).
Cirrus related:
1. Added SUBSYS PCI vendor/device ID of the 5480 (per manual).
IBM VGA:
1. Built-in/option rom-less VGA don't need the "available" flag.
ATI Mach8/32:
1. As with 8514/A, corrected interlaced display.
XGA-1/-2:
1. Moved the XGA R/W memory size tests out of the SVGA R/W routines to reflect the per card basis, although anything that uses its own SVGA mapping would call the tests there (such as Cirrus, Headland and ATI) when not accessing the LFB. This finally puts an end to the XGA MCA mapping enabling bugs.
2. Re-organized the ISA standalone and non-standalone (INMOS) sides of the chips so that they work properly and remove the FILE rom loading hack from init.
3. The Memory Mapped R/W sides now account for instance in their address range.
4. INMOS only: prevent any ROM address access to anything lower than 0xc8000 to not conflict with the main BIOS rom loading.
5. Fixed native pitch by using the correct register, this fixes non 1024x768 resolutions under NT.
6. More logs when enabled to see any future bugs.
SVGA: Added an indicator of the internal line to make matters easier for some vertical resolution problems.
TGUI 9400CXi: Fixed the basic acceleration as much as I could.
TGUI 9440AGi+: As with the line indicator, make matters easier for some resolution problems plus some refactoring of the bpp. Added write protection to some regs based on some documentation on the vt8601.
TGUI cursor: fixed the left edge of the cursor X offset.
S3: Apply the vblankstart = dispend statement only on enhanced modes, fixes vertical display on The Incredible Machine for DOS.
Commented out some excess logs.
IBM 8514/A set:
Fixed a one off video bug that was causing the VGA video to not be restored after quitting a 8514/A GUI session (namely on Win2.x' 8514/A's driver).
ATI Mach32:
Fixed another vertical display issue when activated (in 1280x1024 mode, when 1024 was 1022).
All involved:
Apply RG's changedvram fix to 8514/A, Mach8/32 and XGA's changedvram pointers.
1. Corrected the ATI 1881x clocks for use with the ATI Mach8/32 and VGA Wonder chips. The CPU slowdowns should now be gone.
2. Merged I/O ports common to both the ATI and IBM 8514/A compatible chips where they have identical code (extended behavior is still separate). Code duplication is now less than before.
3. Fixed a general polygon pattern issue in the Mach8/32 affecting calc.exe in Win3.x and other stuff.
4. Mode changes are, once again, changed (ATI and IBM), as close as possible to the real thing without destroying existing resolutions.
5. The 8514/A Vertical Counter has been extended to 0xfff so that it can take 1280x1024 resolutions well offered by the Mach32 as well as a better way to change the IBM/ATI modes through a callback swap where approprietate.
6. in 8514/A mode, reads from the 0x3c6-0x3c9 ramdac range is redirected to the 8514/A RAMDAC (0x2ea-0x2ed).
7. LFB access in the Mach32 now no longer takes account of the SVGA derived rops. Fixes Mach32 display on NeXTSTEP/OPENSTEP 3.x/4.x
8. Reworked the Display Sense Status and Subsystem Status ports so that they're not copycats from MAME and instead follow the datasheet.
Now takes constants with human-readable names instead of
Win32 string table IDs, language.h is no longer needed
ui_msgbox*() no longer accepts string IDs as arguments directly,
plat_get_string() must be explicitly called to retrieve the string
IBM/ATI 8514/A side:
Made mode switches more immediate when actually switching from VGA to 8514/A and viceversa.
Tseng ET4000AX/W32 series side:
Actually use bit 2 of index 0x3f for the horizontal blank start bit 8 instead of bit 4 (horizontal retrace start), fixes skew issues with the et4000w32i rev B. card with resolutions like 1024x768 at 15 or 16bpp and others.
S3 side:
Temporary changes to match the release (due to tomorrow) of 86box. Said changes are about the horizontal display of the tvp3026-based S3 chips.
IBM/ATI 8514-based:
Temporarily commented out the hblank side of it due to htotal bugs.
TVP3026 side:
When the upper clock selection bits are 0 and when extended VGA modes are set, double the hdisp.
* Fixes cursor disappearing completely in Matrox cards in some cases.
* Allows emulated video adapters allowing negative Y values for hardware cursor to render those properly
1. Second attempt to fix the banking in the Cirrus (sigh, why doesn't banking get nulled automatically...)
2. Introduce a new timer to the 8514/A side so it won't slow the VGA clock down it was shared before.
