New FS-UAE branches with updated emulation code from WinUAE

[jbl007]

Originally Posted by Foul

"WARNING: Z3 invalid map_banks(RTG RAM) start=10000000 size=00000000"

Looks like a gfx card with zero MB RAM.

[BastyCDGS]

Hi jbl007!

Originally Posted by jbl007

Looks like a gfx card with zero MB RAM.

Seems to be caused by not passing the amount of RAM properly to emulation. Have the same problem with the winuae4200 branch, too.
Also adding Zorro III RAM results in a very similar error (zero bytes added as well).

EDIT: Added log files for Frode.

[FrodeSolheim]

Zorro III memory seemed to work fine, but I can confirm that there is a problem with the initialization of uaegfx. Start address is not configured, which causes allocation to fail (you should see ""mapped_malloc(*) without start address!" in the log). Will look into the expansion initialization code...

[FrodeSolheim]

I've pushed a fix for the allocation problem (InitPicasso96(0) needed to be added to graphics_setup). I've not actually tested Picasso96 (usage) yet. EDIT: Just did, FS-UAE crashes - I didn't really expect it to work without modifications.

[BastyCDGS]

Thank you Frode!

Originally Posted by FrodeSolheim

I've pushed a fix for the allocation problem (InitPicasso96(0) needed to be added to graphics_setup). I've not actually tested Picasso96 (usage) yet. EDIT: Just did, FS-UAE crashes - I didn't really expect it to work without modifications.

Still crashes in ROM. New log attached after git pull.

[BastyCDGS]

FYI, Toni just released final WinUAE 4.2.1 version! Maybe time to merge it, as well.

Any progress regarding uaegfx crashing? Can I be of any help regarding this?

[FrodeSolheim]

Originally Posted by BastyCDGS

FYI, Toni just released final WinUAE 4.2.1 version! Maybe time to merge it, as well.

It already is (branch winuae4210)

Originally Posted by BastyCDGS

Any progress regarding uaegfx crashing? Can I be of any help regarding this?

The crash itself is already fixed (haven't pushed commit - EDIT: I have now!), but there's still work to do (currently only a black screen with RTG). Most likely, there's not much anyone can do to help without spending lots of time understanding how everything fits together. But I'll soon probably need help testing it at least

[BastyCDGS]

Hi again!

Originally Posted by FrodeSolheim

The crash itself is already fixed (haven't pushed commit - EDIT: I have now!), but there's still work to do (currently only a black screen with RTG). Most likely, there's not much anyone can do to help without spending lots of time understanding how everything fits together. But I'll soon probably need help testing it at least

Great! Just did git pull and tested. However, still crashes in boot phase. New logs attached!

The crash at boot seems to be related with clipboard sharing. See post below!

[BastyCDGS]

Some test results with disabling certain stuff:

Works:
JIT compiler with 68040-NOMMU (others not tested)!
Toccata (tested with OctaMED Soundstudio, plays fine)
Cyberstorm PPC (tested with ADoom PPC WOS, but PPC emulation is extremely slow, so init DOOM refresh daemon seems to hang)

Doesn't work:
uaegfx (gives complete white screen with its bottom and right it flickering between white/black)
bsdsocket.library (AWeb doesn't find TCP/IP stack)
Clipboard sharing (crashes at boot)

NOTE: Just realized I didn't configure build with --enable-clipboard. Will retry that and report ASAP.

Build fails with --enable-clipboard:

src/od-fs/clipboard.cpp: In function ‘uae_u32 to_amiga_start_cb(TrapContext*, void*)’:
src/od-fs/clipboard.cpp:58:69: warning: invalid conversion from ‘uaecptr {aka unsigned int}’ to ‘uae_u8* {aka unsigned char*}’ [-fpermissive]
   debugwrite(ctx, _T("clipboard_p2a"), clipboard_data, to_amiga_size);
                                                                     ^
src/od-fs/clipboard.cpp:48:13: note:   initializing argument 3 of ‘void debugwrite(TrapContext*, const TCHAR*, uae_u8*, int)’
 static void debugwrite(TrapContext *ctx, const TCHAR *name, uae_u8 *p, int size)
             ^~~~~~~~~~
src/od-fs/clipboard.cpp: In function ‘void from_iff(TrapContext*, uaecptr, uae_u32)’:
src/od-fs/clipboard.cpp:682:55: warning: invalid conversion from ‘uaecptr {aka unsigned int}’ to ‘uae_u8* {aka unsigned char*}’ [-fpermissive]
         debugwrite(ctx, _T("clipboard_a2p"), data, len);
                                                       ^
src/od-fs/clipboard.cpp:48:13: note:   initializing argument 3 of ‘void debugwrite(TrapContext*, const TCHAR*, uae_u8*, int)’
 static void debugwrite(TrapContext *ctx, const TCHAR *name, uae_u8 *p, int size)
             ^~~~~~~~~~
src/od-fs/clipboard.cpp:685:40: warning: invalid conversion from ‘TrapContext*’ to ‘uaecptr {aka unsigned int}’ [-fpermissive]
             from_iff_text(ctx, buf, len);
                                        ^
src/od-fs/clipboard.cpp:685:40: warning: invalid conversion from ‘uae_u8* {aka unsigned char*}’ to ‘uae_u32 {aka unsigned int}’ [-fpermissive]
src/od-fs/clipboard.cpp:685:40: error: too many arguments to function ‘void from_iff_text(uaecptr, uae_u32)’
src/od-fs/clipboard.cpp:185:13: note: declared here
 static void from_iff_text (uaecptr ftxt, uae_u32 len)
             ^~~~~~~~~~~~~
src/od-fs/clipboard.cpp: In function ‘void amiga_clipboard_init(TrapContext*)’:
src/od-fs/clipboard.cpp:836:23: error: too few arguments to function ‘void clipboard_read(TrapContext*, bool)’
     clipboard_read(ctx);
                       ^
src/od-fs/clipboard.cpp:699:13: note: declared here
 static void clipboard_read(TrapContext *ctx, bool keyboardinject)
             ^~~~~~~~~~~~~~
src/od-fs/clipboard.cpp: In function ‘void target_paste_to_keyboard()’:
src/od-fs/clipboard.cpp:962:6: warning: no previous declaration for ‘void target_paste_to_keyboard()’ [-Wmissing-declarations]
 void target_paste_to_keyboard(void)
      ^~~~~~~~~~~~~~~~~~~~~~~~
src/od-fs/clipboard.cpp: In function ‘void uae_clipboard_update()’:
src/od-fs/clipboard.cpp:969:20: error: too few arguments to function ‘void clipboard_read(TrapContext*, bool)’
     clipboard_read();
                    ^
src/od-fs/clipboard.cpp:699:13: note: declared here
 static void clipboard_read(TrapContext *ctx, bool keyboardinject)
             ^~~~~~~~~~~~~~
src/od-fs/clipboard.cpp: In function ‘void clipboard_unsafeperiod()’:
src/od-fs/clipboard.cpp:985:2: error: ‘vdelay2’ was not declared in this scope
  vdelay2 = 100;
  ^~~~~~~
src/od-fs/clipboard.cpp:985:2: note: suggested alternative: ‘vdelay’
  vdelay2 = 100;
  ^~~~~~~
  vdelay
Makefile:5616: recipe for target 'src/od-fs/libuae_a-clipboard.o' failed
make[2]: *** [src/od-fs/libuae_a-clipboard.o] Error 1
make[2]: *** Attente des tâches non terminées....
mv -f src/od-fs/.deps/libuae_a-clock.Tpo src/od-fs/.deps/libuae_a-clock.Po
mv -f src/od-fs/.deps/libuae_a-driveclick.Tpo src/od-fs/.deps/libuae_a-driveclick.Po
mv -f src/od-fs/.deps/libuae_a-filesys_host.Tpo src/od-fs/.deps/libuae_a-filesys_host.Po
mv -f src/od-fs/.deps/libuae_a-gui.Tpo src/od-fs/.deps/libuae_a-gui.Po
src/od-fs/hardfile_host.cpp: In function ‘TCHAR* hdf_getnameharddrive(int, int, int*, int*)’:
src/od-fs/hardfile_host.cpp:746:8: warning: no previous declaration for ‘TCHAR* hdf_getnameharddrive(int, int, int*, int*)’ [-Wmissing-declarations]
 TCHAR *hdf_getnameharddrive (int index, int flags, int *sectorsize, int *dangerousdrive)
        ^~~~~~~~~~~~~~~~~~~~
src/inputdevice.cpp: In function ‘bool inputdevice_devicechange(uae_prefs*)’:
src/inputdevice.cpp:7460:6: warning: ‘%d’ directive writing between 1 and 10 bytes into a region of size 7 [-Wformat-overflow=]
 bool inputdevice_devicechange (struct uae_prefs *prefs)
      ^~~~~~~~~~~~~~~~~~~~~~~~
src/inputdevice.cpp:7460:6: note: directive argument in the range [1, 2147483647]
In file included from /usr/include/stdio.h:862:0,
                 from ./src/od-fs/defines.h:44,
                 from ./src/od-fs/sysconfig.h:283,
                 from src/inputdevice.cpp:24:
/usr/include/x86_64-linux-gnu/bits/stdio2.h:34:43: note: ‘__builtin___sprintf_chk’ output between 5 and 14 bytes into a destination of size 10
       __bos (__s), __fmt, __va_arg_pack ());

[FrodeSolheim]

Btw, clipboard and bsdsocket features are disabled on purpose because I know they do not work (or compile for that matter) without additional work I'm working on the uaegfx issues.

[FrodeSolheim]

Pushed commit which makes uaegfx partially working (32-bit mode, <= 960x540 resolution).

[BastyCDGS]

Originally Posted by FrodeSolheim

Btw, clipboard and bsdsocket features are disabled on purpose because I know they do not work (or compile for that matter) without additional work I'm working on the uaegfx issues.

Well, if it clipboard sharing is disabled altogether, fs-uae shouldn't crash when starting with clipboard_sharing = 1 config entry, but issue a warning to the user, instead.

I now have tested Cyberstorm PPC OS4.1FE mode. Booting only works with no graphics card enabled.
PPC speed is normal here instead of the previous test with OS3.9.

However, the new boot ROM mode making UAE expansions compatible with OS4.x also doesn't work, i.e. directory HD's aren't working in OS4.x mode.
Do I have to enable some uae_ setting in config file to enable this?

EDIT: Had missing uae_uaeboard = full+indirect config entry. Directory harddrives work now!

[BastyCDGS]

Originally Posted by FrodeSolheim

Pushed commit which makes uaegfx partially working (32-bit mode, <= 960x540 resolution).

Just tested the commit and it works for me even with 1080p resolution on Workbench! Great!

Note that I have a graphics_card_memory = 16 in the config file. Might be the issue for you!

EDIT: Tested uaegfx now with OS4.1FE, too. Works also in 1080p now! Excellent work!

[jbl007]

Originally Posted by BastyCDGS

Just tested the commit and it works for me even with 1080p resolution on Workbench! Great!

32bit works for me up to 1344x1080. 4k res crashes (not useful anyway). But I miss 1080p resolution. Where does uaegfx get the available resolutions from?

EDIT: Tested uaegfx now with OS4.1FE, too. Works also in 1080p now! Excellent work!

Oh, that's interesting. Do you need to install a special driver or does it work out of the box?

[FrodeSolheim]

Originally Posted by BastyCDGS

Well, if it clipboard sharing is disabled altogether, fs-uae shouldn't crash when starting with clipboard_sharing = 1 config entry, but issue a warning to the user, instead.

Of course, I just haven't gotten to it yet. This is all work-in-progress.

Originally Posted by BastyCDGS

I now have tested Cyberstorm PPC OS4.1FE mode. Booting only works with no graphics card enabled. PPC speed is normal here instead of the previous test with OS3.9.

Excellent

Originally Posted by BastyCDGS

However, the new boot ROM mode making UAE expansions compatible with OS4.x also doesn't work, i.e. directory HD's aren't working in OS4.x mode.
Do I have to enable some uae_ setting in config file to enable this? EDIT: Had missing uae_uaeboard = full+indirect config entry. Directory harddrives work now!

I'm almost a but surprised that it work without modifications (also uaegfx in OS4.1FE), great work @toni -And thanks for testing.

Originally Posted by BastyCDGS

Just tested the commit and it works for me even with 1080p resolution on Workbench! Great! Note that I have a graphics_card_memory = 16 in the config file. Might be the issue for you! EDIT: Tested uaegfx now with OS4.1FE, too. Works also in 1080p now! Excellent work!

Hmm, strange, it works for me up to about 1920x1020x32, and I supposedly have 16 MB graphics memory. Strange, but will investigate further. Are you sure you get full HD @32-bit color?

Originally Posted by jbl007

32bit works for me up to 1344x1080. 4k res crashes (not useful anyway). But I miss 1080p resolution. Where does uaegfx get the available resolutions from?

Mostly hard-coded in FS-UAE. But I always add the native fullscreen resolution (and the window resolution you started FS-UAE with). I'll add 1920x1080 to the list of built-in resolutions so it's always available (when I get it to work again).

[FrodeSolheim]

Hmm, it looks like 1920x0180x32 works if I also enable some Zorro III memory. (RTG RAM is then moved to memory address 0x50000000 - may or may not be relevant, or maybe the graphics driver simply needs sufficient amount of fast RAM?)

[jbl007]

Is 16MB really enough for full hd res? Who can do the math?
BTW: 3840x2160 is fully working with

graphics_card_memory=128

!

Edit: Ah, yeah, Z III mem was enabled.

Originally Posted by FrodeSolheim

I'll add 1920x1080 to the list of built-in resolutions

Thanks!

[FrodeSolheim]

Originally Posted by jbl007

Is 16MB really enough for full hd res? Who can do the math?
BTW: 3840x2160 is fully working with

You need slightly more than 8 MB for full HD (1980 * 1080 * 4 / (1024 * 1024.0)) and slightly less than 32 MB for 3840x160 (3840 * 2160 * 4 / (1024 * 1024.0)). This is for the framebuffer itself. The picasso driver might need additional memory.

EDIT: Pushed a commit fixing 8-bit and 16-bit modes.

[jbl007]

OK, so there's a bug somewhere. For me 4k works only with 128MB, but not with 64MB. (with or without Z III mem). Does it make sense to find the exact amount where it stops working?

Edit: That was a stupid question -> can't set something in between.

[jbl007]

Originally Posted by FrodeSolheim

Pushed a commit fixing 8-bit and 16-bit modes.

Nice, seems to work fine now.

Full hd still doesn't work for me, now I have the following available:
1384x1080
1900x1030
1919x1026

Edit: Just noticed WinUAE sets

z3mem_start=0x40000000

and FS-UAE defaults to

z3mem_start=0x10000000