Can't boot Mac OS 9.22 iso on Linux (Debian 13)

[diegov-au]

Hi there - thanks for all the work on this emulator.

I’m running into an intermittent boot issue with the Beige G3 emulation while trying to boot a Mac OS 9.2.2 install ISO. Roughly 1 out of 10 boot attempts succeeds; the rest freeze at the “Starting Up” splash screen. I can also report that same build for Windows works fine, so it appears to be specific to Linux.

A few observations:

The emulator process itself does not appear to crash or hang completely - I can still access the debug console after the freeze.
Deleting nvram.bin sometimes allows a successful boot attempt.

The warning I see repeatedly in the logs during failed boots is:

( 16.775s) viacuda.cpp:555 WARN| Cuda transaction probably in progress

Startup command:

./dingusppc -b g3.rom -r --rambank_size=256 --gfxmem_size=6 --hdd_img=macos922.dsk --cdr_img=macos_922.iso

I've also tried booting with the cd-rom only (no hdd attached)

Relevant log output:

(   0.000s)              loguru.cpp:851   INFO| Logging to 'dingusppc.log', mode: 'a', verbosity: 0
(   0.001s)      machinefactory.cpp:461   INFO| Found match (1/1):
(   0.001s)      machinefactory.cpp:463   INFO|     ROM description: Power Mac G3 (v3)
(   0.001s)      machinefactory.cpp:465   INFO|     Machine identified from ROM: Power Mac G3 Beige
(   0.001s)      machinefactory.cpp:537   INFO|     ROM Version: 077d.45f2
(   0.001s)      machinefactory.cpp:541   INFO|     ConfigInfo.BootstrapVersion: "Boot Gossamer 0."
(   0.001s)      machinefactory.cpp:567   INFO|     ROM Checksum: 0x78f57389
...
(   0.134s)             viacuda.cpp:196   WARN| Attempted write to VIA Port A!
(   0.135s)             viacuda.cpp:749   INFO| Cuda: send 1 to PB0
(   0.135s)             viacuda.cpp:843   WARN| Unsupported I2C device 0x56
(   0.136s)             viacuda.cpp:843   WARN| Unsupported I2C device 0x55
...
(   2.667s)             viacuda.cpp:614   INFO| Cuda: autopoll started, rate: 11 ms
...
(  16.775s)             viacuda.cpp:555   WARN| Cuda transaction probably in progress

I’m not sure whether this is related to CUDA/VIA emulation, NVRAM state persistence, or the OS 9.2.2 boot process itself, but the behavior is very reproducible on my end.

Let me know if there’s any additional logging or testing I can provide.

Thank you!