09146: pokemini: If the divisor to the DIV instruction is zero, CPU should raise an exception

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
09146	Crash/Freeze	Minor	Always	Mar 30, 2025, 10:59	Mar 30, 2025, 13:47

Tester	Robbbert	View Status	Public	Platform	MAME (Self-compiled)
Assigned To		Resolution	Open	OS	Windows 10/11 (64-bit)
Status [?]	Confirmed			Driver	nintendo/pokemini.cpp
Version	0.275	Fixed in Version		Build	64-bit
		Fixed in Git Commit		Github Pull Request #

Summary	09146: pokemini: If the divisor to the DIV instruction is zero, CPU should raise an exception
Description	The instruction DIV HL/A (CE D9) will crash MAME if A is zero. According to the information at https://www.pokemon-mini.net/documentation/minx-cpu/ , the CPU should hard-lock instead. That doesn't mean that MAME should stop responding, just that the CPU should no longer advance through the program. ~~~~~~~ MAME crash was fixed, but no divide by 0 exception handling
Steps To Reproduce	The program that causes this issue is attached. C:\MAME>mame pokemini -cart e:\data\nintendo\pokemini\ok\3dcubetest.min
Additional Information	----------------------------------------------------- Exception at EIP=00007ff786dd6061 (minx_cpu_device::execute_one_ce()+0x4ab1): INTEGER DIVIDE BY ZERO ----------------------------------------------------- RAX=0000000000000910 RBX=000002c2be442f90 RCX=0000000000000910 RDX=0000000000000000 RSI=00000000000000d9 RDI=0000000000000000 RBP=00007ff792a25ce0 RSP=000000dd613c98e0 R8=0000000000000000 R9=0000000000000000 R10=0000000000000000 R11=0000003a35294400 R12=0de0b6b3a7640000 R13=00007ff792a25ce0 R14=0000000000000353 R15=000000dd613c9d90 ----------------------------------------------------- Stack crawl: 000000dd613c9930: 00007ff786dd6061 (minx_cpu_device::execute_one_ce()+0x4ab1) 000000dd613c9990: 00007ff786ddc57a (minx_cpu_device::execute_one()+0x117a) 000000dd613c99d0: 00007ff786de194d (minx_cpu_device::execute_run()+0x003d) 000000dd613c9a50: 00007ff78476433c (device_scheduler::timeslice()+0x04ac) 000000dd613c9bd0: 00007ff784761225 (running_machine::run(bool)+0x01c5) 000000dd613cf1f0: 00007ff78790142c (mame_machine_manager::execute()+0x024c) 000000dd613cf5e0: 00007ff78b65fcda (cli_frontend::start_execution(mame_machine_manager*, std::vector<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > > const&)+0x03ea) 000000dd613cf8f0: 00007ff78b66030a (cli_frontend::execute(std::vector<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > >&)+0x007a) 000000dd613cf950: 00007ff7878fc1d7 (emulator_info::start_frontend(emu_options&, osd_interface&, std::vector<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > >&)+0x0027) 000000dd613cfd20: 00007ff78d320271 (luaopen_lfs+0xc6bd91) 000000dd613cfd70: 00007ff7809912ee (__tmainCRTStartup+0x016e) 000000dd613cfda0: 00007ff780991406 (mainCRTStartup+0x0016) 000000dd613cfdd0: 00007ffda2847374 (BaseThreadInitThunk+0x0014) 000000dd613cfe50: 00007ffda2afcc91 (RtlUserThreadStart+0x0021)
Github Commit

Flags	Noted in Source
Regression Version
Affected Sets / Systems	pokemini

Attached Files	3Dcubetest.zip (4,099 bytes) Mar 30, 2025, 10:59 Uploaded by Robbbert 3D Cube Test

No.23169 hap Developer Mar 30, 2025, 12:36 edited on: Mar 30, 2025, 12:45	from TODO note: - DIV (CE D9) division by zero handling - is supposed to raise a EX4 exception. A real Pokemini unit will freeze. MAME currently will crash. I removed the crash here, still no exception handling but at least it doesn't crash MAME anymore. https://github.com/mamedev/mame/commit/04edb1b5543d7b961f51b96bdb6a3ee3be91465b
No.23170 Robbbert Moderator Mar 30, 2025, 13:05 edited on: Mar 30, 2025, 13:06	Ah, hadn't known about the todo. Thanks for fixing the crash. I'll change this to confirmed and let it be until someone eventually can make it the same as the real unit. I guess it can become the same as a Z80 HALT instruction, but without an interrupt to escape it.
No.23171 hap Developer Mar 30, 2025, 13:45	I didn't look into it further. I assume the Pokemon Mini BIOS exception handler does a soft-lock, and it's not likely that the CPU itself locks up.