Symptom
motoxgo (World; US/Japan sets presumably alike) shows a black screen immediately after the system warning screens and never reaches POST. The MAME window goes "Not Responding" and the process must be force-killed. Reproduced with -nodrc, with a completely clean cfg/ini environment, and on self-built binaries from unmodified sources (Windows, MSYS2 UCRT64, GCC 16.1) — an official-source rebuild at the mame0288 tag works fine with the same toolchain, so this is not a build-environment issue.
Bisection
All probes built with SOURCES=src/mame/namco/namcos23.cpp on Windows and booted with the same ROM set:
First bad commit: d7e131c — "F2MC-16: External interrupts do not seem to have a iack/vector fetch cycle, according to the docs. And essentially none of the drivers use interrupts in the first place" (src/devices/cpu/f2mc16/f2mc16.cpp, +4/−7).
Why only some System 23 games
The motoxgo sets emulate the "extra" I/O board with a Fujitsu MB90611A (F2MC-16 family) MCU; timecrs2 has no F2MC device and is unaffected. This is consistent with the external-interrupt change wedging the I/O-board MCU (and with it the main CPU's I/O handshake) before POST. Untested prediction: aking (MB90F574 "iocpu2" region) is likely affected as well.
Found while regression-testing PR #15777 against other System 23 titles. Happy to test a fix — the repro loop here is quick.
Symptom
motoxgo(World; US/Japan sets presumably alike) shows a black screen immediately after the system warning screens and never reaches POST. The MAME window goes "Not Responding" and the process must be force-killed. Reproduced with-nodrc, with a completely clean cfg/ini environment, and on self-built binaries from unmodified sources (Windows, MSYS2 UCRT64, GCC 16.1) — an official-source rebuild at the mame0288 tag works fine with the same toolchain, so this is not a build-environment issue.Bisection
All probes built with
SOURCES=src/mame/namco/namcos23.cppon Windows and booted with the same ROM set:First bad commit: d7e131c — "F2MC-16: External interrupts do not seem to have a iack/vector fetch cycle, according to the docs. And essentially none of the drivers use interrupts in the first place" (
src/devices/cpu/f2mc16/f2mc16.cpp, +4/−7).Why only some System 23 games
The
motoxgosets emulate the "extra" I/O board with a Fujitsu MB90611A (F2MC-16 family) MCU;timecrs2has no F2MC device and is unaffected. This is consistent with the external-interrupt change wedging the I/O-board MCU (and with it the main CPU's I/O handshake) before POST. Untested prediction:aking(MB90F574 "iocpu2" region) is likely affected as well.Found while regression-testing PR #15777 against other System 23 titles. Happy to test a fix — the repro loop here is quick.