[AMIGA] Winuae 5.3.0 beta V 23/04/24

Winuae 5.3.0 Beta V 23/04/2024

• Fixed pre-KS 1.2 HDF/directory filesystem boot. Two bugs: crash due to config handling changes (3.6) and KS 1.0/1.1 seems to have a bug where level 2 CIA keyboard interrupt handler returns "I handled this interrupt" status even if it was caused by external device, skipping all following handers, including UAE filesystem interrupt handler, causing a hang. UAE FS interrupt level was changed to -1 if KS 1.3 or lower (4.9) because there was some buggy (KS1.x compatible only) program that poked interrupt list and broke if CIA handler wasn't first in list. Now UAE FS handler priority is lower than CIA handler only if KS 1.2 or 1.3.
• Multi-monitor active state was not fully reset when system was hard reset. (Secondary window was open -> hard reset -> window didn't open anymore again)
• If x86 bridgeboard SVGA and also other hardware emulated RTG board was active and END+F9 was used to switch monitors: both monitors' output was drawn in same window, causing corrupted graphics and possibly also crash if monitor sizes were too different.
• Adjusted 520b5 audio period hack, do not change period immediately after first loop but only after it has looped multiple times. Does not anymore mess up weird audio test programs.
• Permedia 2 Video Control register video enabled/disabled (disabled = blanked) and Line doubling bits emulated.
• Show error message without resetting config back to defaults when GUI Load button could not open the file.
• GD-CL54xx LFB VRAM reads fixed. (For example NetBSD console scrolling was broken)
• Fixed S3 Virge MMIO address calculation (fix from 86box update). NetBSD "No mapping" log message fix.
• A2410 NetBSD text console works again.
• Autoswitch to hardware emulated RTG board also when text mode gets enabled (*nix boot consoles etc)

Winuae 5.3.0 Beta IV 14/04/2024

• ET4000W32/Merlin fixes, ProBench (Early RTG software for Merlin cards) now works without glitches. Merlin serial eeprom(?) emulation added: ProBench requires matching Merlin serial number and valid checksum. Serial number is configurable in Expansions -> Graphics boards -> Merlin. Note: amiga.resource.cx ProBench install disks have different serial numbers (1.10=937875, 2.1=23531, 2.2=10010). Note2: there is "cracked" ProBench 3.0 floating around but it is worse than useless, it has modified owner/user string but critical encrypted code is not decrypted. Decryption requires original user string and serial..
• ET4000AX/W32 lores mode not getting horizontal doubled in >8bit modes fixed. Picasso96 RGBA mode (only available in low resolutions due to chip limits) colors fixed. Interlaced modes fixed.
• OCS Denise left overscan special case emulation updated and fixed. Previously BPL1DAT write before cycle $2D was ignored but that is not correct. BPL1DAT writes work normally but only BPL1DAT writes >=$2D disable border. This is not same enable flag as DIWSTRT horizontal match. Bitplanes or sprites are not visible until both DIWSTRT and BPL1DAT border disable are active but bitplane shifters work normally. Only difference compared to previous version is that first few bitplane/sprite pixels in max overscan will not be visible (replaced by color 0) depending on DDFSTRT position if OCS Denise. ECS Denise/AGA does not have this limitation.
• DMA debugger only showed Denise horizontal counter value if also CPU interrupt level was non-zero.
• Only "insert" CV/BVPPC in to CSMK3/CSPPC internal PCI slot. Insert any other PCI card in to G-REX PCI slots. OS4 does not detect boards in first "internal" PCI slot. (b3)
• Fixed ECS Denise superhires mode hires sprite pointer strange odd/even horizontal movement.
• Check all possible Munt ROM directories, including root UAE ROM path when MT32 ROM is needed. Check also other directories if previously detected path existed but didn't contain requested ROM(s).

Winuae 5.3.0 Beta III 05/04/2024

• Merlin RTG board blitter fixed. 86box ET4000 blitter/MMU was incomplete (No linear blitter/MMU apertures, missing blit start mode check, pattern/source alignment restrictions are ET4000w32p only). Added vblank interrupt and monitor switcher support. Fixed horizontal panning in 24-bit modes. Overlay/PIP not yet implemented. It is quite useless because overlay and main bitmap depth must match.
• Added oMniBus RTG board (Zorro II to ISA adapter). ET4000AX (no blitter, no HW cursor) or ET4000W32 (blitter and HW cursor. Note: Current Picasso96 drivers have a bug that makes ET4000W32 blitter unusable, bug that apparently has always been in this driver. Fix will be in next Picasso96 release.)
• Added Graffity Z2/Z3 RTG board. Cirrus Logic GD5428 based. Has monitor switcher.
• Added Rainbow II "RTG" card. Not really proper RTG card, has only hardwired PAL and NTSC resolutions (768x576 and 768x476). 24-bit depth only. Note that emulation always refreshes whole screen every frame (not just modified parts but whole screen) which is not very optimal but it really isn't worth the trouble because I am sure now one cares about this old board.
• Some hardware emulated RTG boards had flickering colors during horizontal panning (>8 bit modes only)
   
• Small PCI emulation changes (For example G-REX + Permedia 2 + some other PCI card: same card appeared 8 times)
• Added Apollo 630 emulation and boot ROM 5.61. Nothing special, ROM is only used to detect the RAM: "32Bit Power-Burstmode-FastRAM", apparently cool sounding name can make RAM even faster! (oddly ROM contains at least parts of IDE and SCSI drivers from other Apollo accelerators).
   
• In a rare situation it was possible to get false positive "CPU Trace" error when loading statefile.
• Debugger breakpoint "f 123 <extra parameters>" didn't work, "<extra parameters>" was incorrectly detected as PC end address even when it wasn't a number.
• Programmed mode with only VARVBEN enabled (Programmed vblank but hardwired syncs) didn't check if programmed VB ended earlier than hardwired PAL/NTSC vblank, resulting in vblank ending too late, hiding part of display at the top. (ross test(tm))
• If strobe conflicts start after STREQU strobes but before first STRHOR strobes: Denise vblank state stays stuck active until next valid STRHOR strobe is seen by Denise. (SOS / Triangle + ECS Agnus). Interesting detail: If OCS Denise, STRVBL strobe DOES NOT enable vblank (fixed in ECS Denise). Only STREQU strobes enable vblank. STRHOR disables vblank.
• OCS Denise "buggy" vblank start/end (visible only in Overscan+ and higher) start/end position was 2.5 CCKs too early.
• Left and right A1000 Denise/OCS Denise/ECS Denise hardwired hblank position slight adjustments (1 hires pixel) to match new test results.
• Both "SaveImage" (automatically created ext adf when write-enabling non-writable image like ipf) file extension variants are now detected automatically when checking if file exists. Paths panel option now only affects name when new saveimage is created.
• All Picasso96 supported RTG boards are now emulated. All CGX4 supported boards except Inferno. Inferno can't work, only known Wildfire boot ROM fails to configure it properly. PCI config logic seems to get confused if PCI BAR size is too large (GD5446 is 16M or 32M depending on revision). It seems Inferno support needs newer boot ROM which may not have been officially released. Apparently Inferno was also never officially released. EGS not tested, reports accepted if someone really wants to check EGS drivers.

Winuae 5.3.0 Beta II 27/03/2024

• Release raw input devices when GUI is open (except during input Test/Remap) or when debugger window is active. Allows Windows shortcut keys like Win+E now work without need to unfocus the window.
• Added Misc panel option to disable default on screen keyboard gamepad button 4 mapping.
• Added Domino RTG board emulation. (ET4000, basic SVGA chip, no HW cursor, no blitter. Has linear frame buffer support. Emulator code from 86box.)
• Added Merlin Z2 and Z3 RTG board emulation. (ET4000/W32, ET4000+ HW cursor and blitter. Emulator code from 86box.). Blitter does is yet hooked up properly. (If you try it, make sure NOBLITTER=YES)
• uae-configuration AKS_ events (on/off/toggle type only) now work again as designed. ("AKS_xxx -1" = toggle, "AKS_xxx 0" always off, "AKS_xxx 1" always on)
• Some uae-configuration (and other similar) debugging logging messages automatically opened the console window. Now only entering debugger will open it.

• 68000 branch instruction bus error fixes (CPUtester + external bus error hardware tests was not properly done for these instructions for some reason, only following instruction non-branch prefetch bus error condition was tested few years ago). Bcc, BSR, DBcc, JMP, JSR, RTE, RTR, RTS confirmed having accurate bus error stack frame now. Hatari reported bug. Does not affect Amiga emulation.

(These all only because I got report that SuperPlus on ECS has first line blanked. So me and Ross did some weird tests again and things started to snowball. Again. Almost no normal program cares, only exception is ECS blanking fix)

• Programmed display mode state log lines (VS=xx, HS=xx etc) had illogical mix of hexadecimal and decimal values.
• Programmed vertical blanking start/end corrections. ECS Denise EXTBLANKEN=1 + BEAMCON0 BLANKEN=1 blanking starts 1 line earlier and ends 1 line earlier than AGA programmed blanking or ECS EXTBLANKEN=0 (Use hardwired horizontal, programmed vertical blanking). Agnus tells Denise blanking via CSYNC line but AGA Lisa handles it fully internally. Some ECS programmed mode screen modes had missing line at the top.
• Writing to BEAMCON0 didn't check if write would cause immediate vertical DIW close (or open) if HARDDIS/VARVBEN/VARBEAMEN changed. (No normal programs care as usual)
• Fixed corruption in some "weird" programmed modes, introduced in 5.0 betas.
• Sprite/bitplane DMA conflict when bitplane ends very early with active sprite slot in same position as last BPL1DAT: in some situations sprite DMA "won" and conflict didn't happen.
• BPLCON0 ERSY bit getting set without genlock now stops the copper after 2 lines. Previously copper worked normally and overwrote BPLCON0 with valid value if ERSY bit getting set was caused by something else, like a DMA conflict.
• Vertical frequency calculation always expected odd/even line length toggle in NTSC modes, even if BEAMCON0 LOLDIS was set.
• If programmed VBSTRT == VBSTOP: vblank interrupts are not generated.

Winuae 5.3.0 Beta I 20/03/2024

Earlier than expected new beta series with some fixes and some new PCI and RTG boards. Original plan was to start yet another custom chipset emulation rewrite but so far I haven't found any good enough ideas that would solve all remaining known limitations without new side-effects.

Changelog

• CyberVision PPC/BlizzardVision PPC RTG board (Permedia 2 chip) partial emulation. G-REX needs to be enabled, C/BVPPC is basically Permedia 2 PCI card in CyberStorm MK3/PPC/Blizzard PPC internal PCI bus but I didn't want to to emulate "fake" G-REX with only single virtual slot. Yes. I wrote yet another SVGA chip emulator.
• Visiona RTG board emulation (INMOS G300 chip. And yet another video chip emulated.)
• Rainbow III RTG board emulation (INMOS G360 chip)
• EGS 110/24 RTG board emulation (INMOS G364 chip. Same as G360 functionally). Goes to GVP accelerator local slot, it is not autoconfig device. Can be configured without GVP accelerator card but at least Picasso96 driver needs to find GVP autoconfig device before detecting EGS.
• If PCI device was non-UAE built-in (It was PCem SVGA card) and PCI was configured by single write to PCI BAR (Base Address Register), without first doing PCI bar size/alignment detection, device ignored all BAR writes. Fixes pre-44.71 CSMK3/PPC/BPPC boot ROM CyberVision/BlizzardVision PPC PCI configuration (Pre-44.71 support CV/BV PPC but not G-REX PCI). This was not a problem with PCI bridgeboards because bridgeboard drivers always first autodetect each PCI device's BAR size and required alignment.
• Added Prometheus FireStorm PCI bridge emulation (Prometheus with different autoconfig ID, different PCI space mapping, few config bits and some fixes that don't affect emulation). Prmscan works and detects and configures boards, no other tests done. Please report if it works or not.
• When PCI bridge interrupts were enabled, possible alreadying pending interrupts were not checked immediately. (Probably makes no difference)
• Delay ALG LDP seek complete status reply slightly. Fixes Platoon (Nova) few second hangs.
• Added partial Sony LDP-1450 OSD font emulation. Platoon (Nova) "YOU'RE HIT!!" messages are now visible.. (Yes, game uses LDP-1450 internal font generator to show this message. Amiga does not generate it.). If you want some other font than selected OSD font: config file genlock_font=<font name> can be used to override it.
• Lightpen/gun cursor in any edge of screen: lightpen/gun not pointing at the screen (position can't be read by the program)
• Fixed XT based Bridgeboards non-working floppy drive emulation, broke during 5.1 DraCo updates (wrong density detection)
• Statefile short cut keys now select current config file name as base statefile name (for example, if "test1.uae" config file is loaded, SHIFT+END+0 will save statefile as "test1.uss". Quickstart mode config uses previously always used "default.uss")
• CD32 FMV video image size and positioning was weird in most scaled modes.
• OSD font/statusline size was very large in full screen modes if monitor had >100% scaling.
• Statusline font is now slightly taller (few pixels), same as height of OSD boxes. In recent versions statusline font was changed to same as OSD font, earlier versions' statusline text had different font and was also slightly taller than default OSD font.
• Allow 68020+ CPUs to fetch instruction words from custom chipset space. Only if more compatible is set. If some program really wants to do something weird like execute code located in AGA palette registers
• It was not possible to write and read ECS/AGA genlock transparency bit in color registers. It must have been broken long time ago... (I think some program used this to detect AGA)
• Added "Always on (Fixed only)" sound filter option. This enables A500 fixed filter but never enables "led filter". (A1200 does not have fixed filter, only "led" filter)
• Fixed serial port missed received data if received data arrived from "outside world" (master/slave serial port link or real serial port or direct checbox checked). Now waits for both SERDATR read and INTREQ bit clear before new serial word is accepted. Loopback mode was not affected. Previous serial loopback emulation updates made it too "accurate" in normal modes...
• -serlog serial port logging mode always enabled "direct" mode internally which affected serial timing. This does not happen anymore.
• Exclusive fullscreen and RTG resolution switch (RTG to RTG direct switch, not RTG->quick tempory blanked native->RTG) didn't change native resolution.
• Windows mouse capture/hide logic slightly changed, previously it always did temporary mouse unhide/hide which caused mouse cursor to very quickly appear and disappear during screen mode switches, even if mode switch didn't resize or reopen the window. (This also fixes some odd bug in recent Windows 11 insider preview builds where mouse is captured but does not always re-hide if done "too quickly")
• Replaced old Windows 2000 compatible Windows clipboard handling with newer, simpler to use method introduced in Vista.

• All usual Graphics Processor mode screen modes (8/15/16/24/32 bits). Other weird modes are not supported.
• VRAM aperture byte swapping and RAMDAC red/blue swapping modes. (Used at least by Picasso96 driver)
• Hardware cursor (64*64*2).
• Blitter. Most 2D parts emulated. Even the 2D blitter is quite complex, it for example supports point, line, trapezoid and rectangle drawing primities, bitmask (if bit is set: use color x, if bit is not set, skip destination write or use color y, other chips usually call this "color expansion"), "area stippling" (8x8 mask aligned to X and Y coordinates, can do same operations as bitmask), 16 different raster operations (ROP) and more. 3D features are even more complex.

• SVGA core. Amiga programs/drivers only use Graphics Processor mode.
• 3D. 3D is Someone Else's Problem.
• Other Permedia 2 special features (front/back buffer swapping, stereo support, local buffer access etc)

• INMOS G300/G360 ("Colour video controller") chip. It is not VGA or SVGA chip like others but sort of "simple" programmable video generator + RAMDAC.
• G300 does not do much. Supports 8-bit 256 color palette mode and 24-bit truecolor. Linear frame buffer. Nothing else. No hardware cursor, no blitter.
• G364 is updated and modified G300 + 64*64*2 hardware cursor + 15/16 bit modes that have non-standard color component order.
• Some glitches remain. (Some might be emulation bugs or driver doing something wrong that accidentally works)