WinUAE 4.9.0 beta series (Was 4.5.0)

[Toni Wilen]

Another fix attempt, winuae.7z updated.

[vulture]

Hi Toni,

Well done, now we are good!!!

Many thanks; I know DamienD will also be very happy indeed

[Toni Wilen]

https://download.abime.net/winuae/fi...uae_4900b43.7z
https://download.abime.net/winuae/fi...e64_4900b43.7z

Beta 43 (RC4): Beta tags removed. Final should be out by 07.12.2021. (and a remainder: it is called 4.9 because this version is expected to have stupid bugs due to huge amount of changes and 68000 IPL timing is not yet correct. 5.0 should be fully A500 compatible)

- CDFS mounting used wrong character set translation functions, file names with character codes larger than 128 could not be opened depending on Windows regional settings.
- Vertical centering sometimes decided to use max offset, pushing the image down as much as possible.
- Filter Overscan blanking used old horizontal origin (which changed to earlier, "true" horizontal=0 position, due to chipset updates). Workaround added.
- Blacker than black in non-HDR mode was not RGB=000000 black.

[paraj]

Originally Posted by Toni Wilen

I still don't know what causes JIT do to this, I have never seen it happening.. Not very important because it isn't recoverable.

(disassembler output fixed in this condition).

I agree that it's probably not important, so feel free to ignore/move from 4.9.0 thread. A bit strange that you're not seeing the same behavior as me. It's not fixed in b34.

Callstack when I get the disassembly (sorry I did not provide this in my previous report) is:

>    winuae64.exe!ShowEA(void * f, unsigned int pc, unsigned short opcode, int reg, amodes mode, wordsizes size, wchar_t * buf, unsigned int * eaddr, int * actualea, int safemode) Line 526    C++
     winuae64.exe!m68k_disasm_2(wchar_t * buf, int bufsize, unsigned int pc, unsigned short * bufpc, int bufpcsize, unsigned int * nextpc, int cnt, unsigned int * seaddr, unsigned int * deaddr, unsigned int lastpc, int safemode) Line 2319    C++
     winuae64.exe!m68k_disasm(unsigned int addr, unsigned int * nextpc, unsigned int lastpc, int cnt) Line 6488    C++
     winuae64.exe!m68k_dumpstate(unsigned int * nextpc, unsigned int prevpc) Line 6579    C++
     winuae64.exe!default_xlate(unsigned int addr) Line 1165    C++
     winuae64.exe!memory_get_real_address(unsigned int addr) Line 3802    C++
     winuae64.exe!op_4ea8_40_ff(unsigned int opcode) Line 18415    C++
     winuae64.exe!execute_normal() Line 5277    C++
     winuae64.exe!m68k_run_jit() Line 5352    C++
     winuae64.exe!m68k_go(int may_quit) Line 6457    C++

Execution continues normally for a while (don't know how long, I only stepped over ~15 times) after the access violation (Access violation reading location 0x0000000080F00000) that causes exit from pushall_call_handler.


It's reproducible for me with a debug build of latest beta as well, so if you have any thing you want me to try out/extra debug stuff that can be enabled, let me know. (Normal log attached)

[Toni Wilen]

Ah, it tries to jump to address that does not exist. But it should not start debugger unless it was already started previously. It isn't anything serrious because it is not recoverable. Not sure what to do in this case except to cause some exception (bus error) or just stop it completely (HALT).

I'll check this after 4.9 is out..

[paraj]

Originally Posted by Toni Wilen

Ah, it tries to jump to address that does not exist. But it should not start debugger unless it was already started previously. It isn't anything serrious because it is not recoverable. Not sure what to do in this case except to cause some exception (bus error) or just stop it completely (HALT).

I'll check this after 4.9 is out..

Yeah, you deserve to have 4.9 done before Christmas after all your hard work



(Just FYI bisection shows that the debug window started showing up from commit d0e56711422 "4500b16" due to change in memory.cpp, specifically the unconditional call to m68k_dumpstate in default_xlate)

[heavy]

no more compatible with windows xp sp3 ? kernel32.dll error "CompareStringEx" entry point at start.

[MrDBUG]

Congrats on big release !!! Most excited I am right now about Floppybridge !
Any idea where I can direct questions about Floppybridge on Greaseweazle with Winuae under Wine ? :-/ I am struggling to get the /dev/ttyACM0 device to show up as a com port. I've put it in Wine/Crossover registry :-/

[Toni Wilen]

Create separate threads, thanks. I forgot to close this thread

Originally Posted by heavy

no more compatible with windows xp sp3 ? kernel32.dll error "CompareStringEx" entry point at start.

4.3 was last official supported version and apparently 4.4 still worked "accidentally". AFAIK some Visual Studio update released after 4.4 removed XP support completely.

Originally Posted by MrDBUG

Congrats on big release !!! Most excited I am right now about Floppybridge !
Any idea where I can direct questions about Floppybridge on Greaseweazle with Winuae under Wine ? :-/ I am struggling to get the /dev/ttyACM0 device to show up as a com port. I've put it in Wine/Crossover registry :-/

No idea. Probably better to ask FB author. WinUAE does not do any direct port accesses, everything happens using FB API.