Viewing Issue Advanced Details
|ID||Category [?]||Severity [?]||Reproducibility||Date Submitted||Last Update|
|04333||Save/Restore||Minor||Always||May 7, 2011, 10:11||Jun 5, 2011, 17:02|
|Version||0.142u3||Fixed in Version||0.142u5||Build||64-bit|
|Summary||04333: Many games in m92.c: Save states made while raster interrupts are active do not work|
First of all, I'm aware that m92.c does not officially support save states. However, as far as I can tell (and I've been hacking on the driver for the past week) everything necessary to support save states in the driver is already done; only the actual GAME_SUPPORTS_SAVE tag is missing.
If you save while a game in this driver is using raster interrupts (i.e. any time "n partial updates" is displayed under the frame rate), then when you load that save, the screen and music will be correctly restored, but the game will immediately freeze (except for the music, which continues to play).
inthunt - save during POST - loads normally
inthunt - save any time after POST finishes - freezes
nbbatman - save any time from POST up to Irem splash screen - loads normally
nbbatman - save any time after that - freezes
dsoccr94j - save any time - loads normally
majtitl2 - save any time - loads normally
|Steps To Reproduce||
1. Save state at any time when "n partial updates" is displayed
2. Load the state
3. The game freezes
This bug did not occur in 0.141 (I tested with inthunt and nbbatman; in both games, you could save at any time and load the save without it freezing)
Probably related to 4330 and 4331 (emu_timer object-ification)
|Affected Sets / Systems||Many games in m92.c|
May 7, 2011, 18:06
edited on: May 7, 2011, 18:07
This is an issue to be addressed when u3 is released.
Posting issue now based on changes from your submission to SVN does not make it an allowable bug report here. In 0.142u2, the driver is not flagged with GAME_SUPPORTS_SAVE and by the strictest rules, we do not load/save bugs for games which are not set up to handle them (are flagged).
I'm choosing to suspend this report it until it's decided how or when to continue with it.
May 7, 2011, 20:00
The report has nothing to do with my submission. I didn't add a single state->save_item() call to m92.c. All of them were already there, and have been for many years. All I did was add the tag.
I've just done my own testing and confirmed that the regression version is r11464. Save states work perfectly in r11463. In r11464 you can save during inthunt POST sequence, but if you save during gameplay it freezes when you load the state.
May 7, 2011, 21:24
edited on: May 7, 2011, 21:25
That may be so, but this site only deals with release to release regressions.
With current policy, as I stated before, since U2 is not flagged, we will not accept bugs for it. I'm not deleting it - just suspending it until such time that U3 is released which does have the GAME_SUPPORTS_SAVE for the games.
As a contributor, if you knew there is a problem with save states, you shouldn't have added the flags until they could be proven working. I know, it's weak reasoning and I do understand why you posted this as there is a clear regression, but without that flag in a RELEASE version, it's not accepted. That has always been the policy here. Presumably, this will be resolved as soon as the other similar corruption issues are dealt with surrounding the other save state reports which I'm sure are more widespread.