Setting base_memory to 512 and extended_memory to 0 results in trouble booting about freebee HOT 7 CLOSED

Call stack on current attempt:
m68k_read_memory_8 (0x00ffff01); // bus error within bus error
m68ki_exception_bus_error ()
m68k_pulse_bus_error () [Page fault?]
m68k_read_memory_8 (0x2FFD86)

From musashi:

/* Exception for bus error */
static inline void m68ki_exception_bus_error(void)
{
    int i;

    /* If we were processing a bus error, address error, or reset,
     * this is a catastrophic failure.
     * Halt the CPU
     */
    if(CPU_RUN_MODE == RUN_MODE_BERR_AERR_RESET)
    {
m68k_read_memory_8(0x00ffff01);
        CPU_STOPPED = STOP_LEVEL_HALT;
        return;
    }
    CPU_RUN_MODE = RUN_MODE_BERR_AERR_RESET;
...

from freebee.

Are you sure this is a supported configuration?

I think 3.50 was the only UNIX which could boot on 512k machines (the 7300 specifically). All 3B1s came with 1MB minimum, and (don't quote me on this!) the 3.51 UNIX.

from freebee.

Ahh this could definitely be the issue -- running out of RAM trying to boot a 3.51 UNIX. If I can find an older OS set, I could try that on the 512k, but it sounds like this is Not A Bug then.

from freebee.

I've spent a lot of time investigating this issue, as I was getting the "double bus error" assert in other instances. My conclusion is that the handling in Musashi is incorrect. On the UNIX PC, there are instances when a page fault will occur during the bus error exception handling of an initial PF. Musashi interprets this as a catastrophic error. However, the only time it's catastrophic is if the bus error occurs during the setup of the stack frame. And the stack frame is setup within m68ki_exception_bus_error() so this should never occur. MAME's Musashi code appears to have been changed to reflect this.

I'll see if I can get Musashi updated with a fix, and then update Freebee to the latest Musashi submodule if that sounds good.

from freebee.

Sounds good to me!

from freebee.

One other quick note (on the original issue), I tried setting the the RAM to 512KB and booting 3.51m with the check for double bus fault disabled in Musashi and sure enough things booted and seemed to work fine. Memory was nearly 100% used and seemed to be hitting the swap file as it was fairly slow, but did seem to work.

from freebee.

Fixed with latest Musashi

from freebee.