 No.05074
etabeta Developer
Oct 27, 2009, 21:49
|
I'm not sure this is a bug: isn't the UNEMULATED_PROTECTION flag set exactly because the MCU is not emulated yet? |
|---|---|
 No.05075
Tafoid Administrator
Oct 27, 2009, 22:39
|
The flag was added, yes. I'm not sure if Aaron intended this set and clone to become unplayable when it used to work before the emu was needed. It was simulated before, I think. |
|
No.05986
AWJ Developer
Apr 17, 2010, 23:18
edited on: Apr 17, 2010, 23:36
|
dumpmtmt has the same problem. This is happening because even though the i8751 MCU is disabled, it writes to each of its special I/O ports during its CPU_RESET and in the case of segas16a.c this causes the 68000 RESET line to get set and never cleared: static CPU_RESET( mcs51 ) { (snip) /* set the port configurations to all 1's */ SET_P3(0xff); SET_P2(0xff); SET_P1(0xff); SET_P0(0xff); To fix the problem, two things need to be done: First, either the MAME core needs to be made not to call CPU_RESET functions for CPUs that are disabled, or individual CPU cores such as mcs51 have to be made to check whether they are disabled and if they are, not call any functions at reset that can affect the machine at large. I'm not sure which is the right approach--Aaron? Second, in drivers that have an i8751 that needs to be disabled (because we don't have a dump and are using simulation) the i8751 needs to be disabled at the driver level using MDRV_CPU_FLAGS(CPU_DISABLE) rather than at runtime using cpu_suspend(). Suspending the CPU at runtime doesn't happen until after the various reset functions have already been called, so even if one of the fixes above is applied it won't help quartet or dumpmtmt. |