Hardware RTG board emulation

[mark_k]

Not sure whether it's of any interest, but I uploaded some CD/DVD drive manuals to The Zone just now.

As for chip datasheets, various but probably not too many specifically Amiga-related. Some old SCSI controller chips, graphics chips etc. etc. Some day I'll try to get around to making a list.

[Toni Wilen]

I managed to take qemu's cirrus SVGA chip emulation and make it work with UAE using some software duct tape.

Currently attempts to emulates Picasso II board (easiest to debug, driver source is available)
Switching to 640x480x256 mode seems to work, cirrus chip emulation returns correct mode information.
Very important missing piece is actually showing the emulated graphics data... (And when/if it works, it probably shows that graphics are totally garbled..)

[Toni Wilen]

It seems to work now, mostly...

PicassoII 640x480 and 800x600 modes at least work, all bit depths.

But for some unknown reason PicassoII driver (that comes with Picasso96) seems to do some weird blits that cross end of VRAM which will be rejected by QEMU Cirrus Logic blitter emulation. This causes display glitches, also sometimes screen does not refresh automatically.

Should be good enough for non-OS tests (Amix, NetBSD, Linux) if someone is interested enough..

Expansion panel Zorro II/III option now have 3rd option, PicassoII.

Usual url: http://www.winuae.net/files/b/winuae.zip

[Leandro Jardim]

Hi Toni, I installed Picasso96, but the PicassoII card isnt shown and I cant attach the card to the modes in Picasso96Settings. The combo box in WinUAE reads: "Picasso II (Do not use)". I did something wrong?

[Toni Wilen]

Originally Posted by Leandro Jardim

Hi Toni, I installed Picasso96, but the PicassoII card isnt shown and I cant attach the card to the modes in Picasso96Settings. The combo box in WinUAE reads: "Picasso II (Do not use)". I did something wrong?

Do you have devs:monitors/PicassoII and PicassoII.info?

All boards have same monitor driver, only name is important, for example rename devs:monitors/Picasso96 -> PicassoII (and .info too)

[Leandro Jardim]

Originally Posted by Toni Wilen

Do you have devs:monitors/PicassoII and PicassoII.info?

All boards have same monitor driver, only name is important, for example rename devs:monitors/Picasso96 -> PicassoII (and .info too)

Toni, I made a new WinUAE config and now it works as expected. I changed the screenmode but the WinUAE display flashes constantly, I cant see anything, it is right?

Anyway its fantastic to see it working so early.

[Toni Wilen]

Originally Posted by Leandro Jardim

Toni, I made a new WinUAE config and now it works as expected. I changed the screenmode but the WinUAE display flashes constantly, I cant see anything, it is right?

Anyway its fantastic to see it working so early.

Works as expected but can't see anything. Does not compute

The usual and more is required: config file, log, picassoii.card and cirrusgd524x.chip versions in libs:, picasso96settings configuration.

[Leandro Jardim]

Originally Posted by Toni Wilen

Works as expected but can't see anything. Does not compute

The usual and more is required: config file, log, picassoii.card and cirrusgd524x.chip versions in libs:, picasso96settings configuration.

I have all these. I attached the card to the settings like the P96 manual says to do and then it disabled some modes automatically, and now, when I test the mode 640x480 16 bit in fullscreen, the screen gets blank without VGA/video output at all.

[Toni Wilen]

Never use fullscreen (or any other extra settings like vsync etc) with pre-beta versions until it is said to be safe.

[Toni Wilen]

"Match host and RTG color depth if possible" setting should be unticked.

Updates:

Added more totally untested Cirrus Logic based boards (Board info from NetBSD sources). PicassoII is still the only supported board.

At least in theory VGA text modes may also work, does something use it? (netbsd or linux console?)

Graphics glitches are actually caused by bug in cirrus logic blitter emulation, it has incorrect "blit is outside of vram" tests, it assumes all blits are normal copy blits, for example it can incorrectly reject color expansion blits (used by RectFill).

Because I am too lazy to create logins to qemu bugtracker or mailing list, at least not yet, here are bugs I have found:

cirrus_vga.c: BLTUNSAFE() macro is broken. (Graphics glitches are gone if I uncomment them all, of course it can introduce crashes if there are actual invalid blits..)

cirrus_vga.c: cirrus_do_copy() "extra x, y", s->cirrus_blt_srcpitch or s->cirrus_blt_dstpitch can be zeros: division by zero.

CIRRUS_BLT_STATUS bit was missing. PicassoII driver polls this to check if blit has finished.

And one thing that may not be a bug:

vga.c: vga_draw_graphic() seems to assume shift_control = (s->gr[VGA_GFX_MODE] >> 5) & 3; must be >=2 (bit 6 of Mode Register is "256 color mode") or it uses standard VGA graphics mode, even if cirrus extended registers are used to enabled extended (SVGA) modes. My workaround was to set bit 6 when any extended mode is enabled.

[Leandro Jardim]

Its working now. Thanks!

[Toni Wilen]

Great

PicassoII+ is now working, vblank interrupt emulated (implemented yesterday) but vblank is currently hardwired to chipset vblank timing.

PicassoII+ = PicassoII + vblank hardware interrupt, newer Cirrus Logic chip, higher memory bandwith. Both supported by PicassoII driver.

[Toni Wilen]

Could someone also test CyberGraphX + emulated PicassoII/PicassoII+ combination?

[Mad-Matt]

Originally Posted by Toni Wilen

Could someone also test CyberGraphX + emulated PicassoII/PicassoII+ combination?

Ill give it a shot. procedding to install CGX v4 v43_rc6 on an updated os39 setting winuae gfx card to PicassoII+ in expansion options. (Winuae gfx driver removed)

Edit: Im not getting very far. The hardware isn't being detected so nothing in screenmode prefs.

Edit2: Although not supported in emu yet, the os driver does allow screenmodes for PicolloDS64

[Toni Wilen]

Originally Posted by Mad-Matt

Ill give it a shot. procedding to install CGX v4 v43_rc6 on an updated os39 setting winuae gfx card to PicassoII+ in expansion options. (Winuae gfx driver removed)

Edit: Im not getting very far. The hardware isn't being detected so nothing in screenmode prefs.

Edit2: Although not supported in emu yet, the os driver does allow screenmodes for PicolloDS64

I adjusted Picasso II autoconfig data to 100% match my real board (Only CL based board I have). Perhaps it helps. I guess I'll also need install CGX myself next

Piccolo and others probably expect working vblank interrupt (if it simply hangs after selecting). Need some reverse engineering to find the used enable/disable addresses.

[Mad-Matt]

The cgx4 manual states PicassoII(+) should be in Linear mode (Not segmented) if that's relevant to the issue at all.

EDIT: I downloaded winuae.zip and now Picasso modes are in screenmode prefs. However the emulated Amiga is crashing before the new mode is opened.

[Toni Wilen]

Originally Posted by Mad-Matt

The cgx4 manual states PicassoII(+) should be in Linear mode (Not segmented) if that's relevant to the issue at all.

It is in linear mode. Segmented emulation would be useless.

EDIT: I downloaded winuae.zip and now Picasso modes are in screenmode prefs. However the emulated Amiga is crashing before the new mode is opened.

Both Picasso II and II+ ?

[Mad-Matt]

Both Picasso II and II+ modes tested. Both use the same cgx driver. The crash seemed to be a one off relating to screenmodes im trying to open. I'm using the 64kHz default profile. So far the lower modes are just opening a grey screen which falls back to aga screen after the screenmode Test timesout.

[Toni Wilen]

Originally Posted by Mad-Matt

Both Picasso II and II+ modes tested. Both use the same cgx driver. The crash seemed to be a one off relating to screenmodes im trying to open. I'm using the 64kHz default profile. So far the lower modes are just opening a grey screen which falls back to aga screen after the screenmode Test timesout.

Grey screen is probably passthrough switch not working correctly. Redownload, enable logging and you should see "GFX BYTE PUT IO X = Y" in the log when chipset->RTG switch and vice versa happens. Include both lines, thanks.

[Mad-Matt]

Ok, I just booted os39, I ran screenmode prefs and pressed test on PicassoIIP 8bit 640x480 which resulted in a 10 second grey screen before returning to aga screen.. and quit uae. Passthrough is enabled in the cgx driver.

The requested lines appear a few times so instead of pasting what I think looks right, i've attached the full log.

Edit2: Just to note, I enabled bootpic in the driver (which loads when drivers are activated in the startup) and this is shown correctly in full colour. In 8bit 640x480 I think but not 100% sure. I guess the driver does this directly and not through the os. When testing with screenmode prefs the bootpic can be seen as a corrupted mess in some modes instead of just a grey screen.

When testing some higher screenmodes like 8bit 1600x1200 (60hz laced) which I believe is highest mode supported on 2mb Picasso2, screenmode prefs reports that it can be opened due to lack of free video ram. The odd effect here is that 1.4mb of my chipmem is eaten up and not returned. This results in the same error if I then try low modes.

EDIT:- Extra Info:
Board (Village Tronic): Prod=2167/11($877/$B) (@$200000 2meg)
cgxsystem.library 42.7 (68K mode)

1 graphics device(s) installed

DEVICES:
PicassoII: Driver Version 4.4 (68K mode)
Prod=2167/11($877/$b)
BaseName=Picasso
CardID=3($3) Number=1($1)
Developed by VillageTronic