02140: cubeqst: cube quest crashes mame when saving state

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
02140	Save/Restore	Minor	Always	Aug 21, 2008, 05:50	Oct 21, 2008, 22:57

Tester	Lord Nightmare	View Status	Public	Platform	MAME (Self-compiled)
Assigned To	Phil Bennett	Resolution	Fixed	OS	Windows XP/Vista 32-bit
Status [?]	Resolved			Driver
Version	0.127	Fixed in Version	0.128u1	Build	Normal
		Fixed in Git Commit		Github Pull Request #

Summary	02140: cubeqst: cube quest crashes mame when saving state
Description	cube quest crashes mame to prompt immediately when saving state
Steps To Reproduce	mame cubeqst hit shift-f7 press any number key
Additional Information	Program received signal SIGSEGV, Segmentation fault. 0x77c46fe2 in msvcrt!memcpy () from C:\WINDOWS\system32\msvcrt.dll (gdb) bt full #0 0x77c46fe2 in msvcrt!memcpy () from C:\WINDOWS\system32\msvcrt.dll No symbol table info available. #1 0x00a34cdd in state_save_save_continue () at src/emu/state.c:685 entry = (ss_entry ) 0x8a60a30 func = (ss_func ) 0x0 count = 0 Backtrace stopped: frame did not save the PC
Github Commit

Flags
Regression Version
Affected Sets / Systems	cubeqst

Attached Files

No.02152 Tafoid Administrator Aug 21, 2008, 08:29	I'm not sure, but doesn't the game expect there to be a video to play? If you have no CHD (you don't have the CHD yet.. correct?) it would have no idea where to leave off and probably cause an error expecting to save a location of video referenced from actual CHD data.. just guessing. It does crash for me without a CHD - that I can confirm.
No.02153 Robbbert Senior Tester Aug 21, 2008, 08:41	Looking at the game driver, save states don't appear to be supported yet.
No.02155 Tafoid Administrator Aug 21, 2008, 08:57	Supported or not, under normal circumstances, attempting to save a game shouldn't cause a fatal error crash of MAME.
No.02166 Phil Bennett Developer Aug 22, 2008, 09:41 edited on: Aug 22, 2008, 09:41	The state saving in the bit-sliced CPU cores is embarrassing; saving RAM pointers before you make the actual allocations isn't wise! A fix should make it into u1/u2.
No.02169 Firewave Senior Tester Aug 22, 2008, 11:00	Missing save state support can cause anything. I am quite sure I will get at least a dozen crashes when I force state saving on each driver in a regtest. IMO if a driver doesn't support savestates, it should not be possible at all to save or load them.
No.02171 Robbbert Senior Tester Aug 22, 2008, 11:18	I'd agree with that last statement. The f-keys and command-line options should be ignored if the game doesn't support save states.
No.02175 robiza Developer Aug 22, 2008, 16:05	i don't agree partial savestate is useful
No.02176 Firewave Senior Tester Aug 22, 2008, 16:08	Maybe for a developer, so maybe it should be only disabled in debug builds and show a warning (yellow or even red background) when using it to avoid invalid or bogus bug reports. This might cause even more havoc and false reports, if people are using the "autosave" option with sets not supporting savestates properly.
No.02181 etabeta Developer Aug 22, 2008, 17:38 edited on: Aug 22, 2008, 17:39	it's already written in the documentation that the option works only for games supporting saves -[no]autosave When enabled, automatically creates a save state file when exiting MAME and automatically attempts to reload it when later starting MAME with the same game. This only works for games that have explicitly enabled save state support in their driver. The default is OFF (-noautosave). and it's off by default. I see no reason to change the current behavior since: 1. there has been no bogus report until now due to autosave used in games not supporting it 2. change an useful option because users are lazy is wrong, imho what can be reasonably done is to add a rule to Mametesters, stating save state reports are only accepted if a game supports save states (if devs agree this should be the rule)
No.02186 Firewave Senior Tester Aug 22, 2008, 20:55	Oh sorry. I did remember I was forcing this option on each driver and not go by the listxml information, but I must have been modifiying the source back then. SOrry for the confusion.