02808: wswan, wscolor: Debugger access violation with V30MZ cpu

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
02808	Core	Major	Always	Jan 8, 2009, 13:01	Mar 7, 2010, 12:31

Tester	Robbbert	View Status	Public	Platform	Other
Assigned To		Resolution	Fixed	OS
Status [?]	Resolved			Driver
Version	0.129	Fixed in Version		Build	Normal
		Fixed in Git Commit		Github Pull Request #

Summary	02808: wswan, wscolor: Debugger access violation with V30MZ cpu
Description	This CPU is only used in MESS by the Wonderswan driver. Any attempt to use the debugger will cause an immediate access violation. The problem was tracked to src\emu\cpu\nec\necdasm.c There is a variable "v25v35_decryptiontable", which whenever referenced causes the crash. It is used in 3 places in this c file. I'm guessing by the name that the variable should not exist for this cpu, but since it is created at the top of the file, I'm not sure what the programmer's intent is. This is the first crash, but you need to comment out all 3 occurances to make the debugger work again. Program received signal SIGSEGV, Segmentation fault. 0x6b023449 in necv_dasm_one (buffer=0x22f938 "", eip=1048554, oprom=0x22f9dc "ÿ", _config=0x6b131820) at src/emu/cpu/nec/necdasm.c:1541 1541 if (Iconfig->v25v35_decryptiontable) op = Iconfig->v25v35_decryptiontable[op];
Steps To Reproduce
Additional Information	Just noticed another issue: if the lines are commented out as above, after about 20 instructions, an assert occurs, on a debug build only: Assertion failed: bytes <= cpu_get_max_opcode_bytes(device), file src/emu/debug/debugcpu.c, line 434
Github Commit

Flags
Regression Version
Affected Sets / Systems	wswan, wscolor

Attached Files

No.03533 Tafoid Administrator Jan 8, 2009, 15:05 edited on: Jan 8, 2009, 15:07	I can confirm with "mess wswan -debug" command line of the downloadable 0.129 MESS. While not specifically a MAME problem - It's my understanding that if a core in MAME the interaction with with debugger is a core issue, not a MESS issue. A clear direction with Dev would be useful for determining bugs like this as valid or invalid in the future.
No.03536 Haze Senior Tester Jan 8, 2009, 21:36	the decryptiontable thing is for supporting the realtime decryption used in the v25+/v35+ cpus (it's a common internal feature of that cpu, so the table is passed to the cpu core, some games switch it on / off dynamically) I suspect the wonderswan version of the cpu hasn't been set up correctly, maybe some code is shared....I do wonder why it has it's own folder and isn't built on top of the existing core.
No.05842 Robbbert Senior Tester Mar 7, 2010, 11:20	This is now fixed, not sure when it happened.