03262: Drivers using TMS9928: Using any type of frameskip causes crash

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
03262	DIP/Input	Critical (emulator)	Always	Jun 14, 2009, 18:42	Jun 18, 2009, 01:27

Tester	Tafoid	View Status	Public	Platform	MAME (Official Binary)
Assigned To	etabeta	Resolution	Fixed	OS	Windows 2000
Status [?]	Resolved			Driver
Version	0.132	Fixed in Version	0.132u1	Build	Normal
		Fixed in Git Commit		Github Pull Request #

Summary	03262: Drivers using TMS9928: Using any type of frameskip causes crash
Description	Using a preset -FS value, or even using Fast Forward (Insert), causes MAME to crash.
Steps To Reproduce
Additional Information	Drivers seemingly effected: forte2.c kingpin.c re900.c sg1000a.c cliffhgr.c uses this, but doesn't seem to crash.
Github Commit

Flags
Regression Version	0.132
Affected Sets / Systems	Drivers using TMS9928

Attached Files

No.04503 Tafoid Administrator Jun 14, 2009, 18:45	Program received signal SIGSEGV, Segmentation fault. 0x00cecd7a in draw_sprites (screen=0x0, bitmap=0x0, cliprect=0x0) at src/emu/video/tms9928a.c:675 675 pens = screen->machine->pens;
No.04504 Justin Kerk Developer Jun 14, 2009, 18:47	I debugged the same issue in the coleco driver in MESS, the crash is in src/emu/video/tms9928a.c:675 (draw_sprites()), which seems to be caused by the following call at line 438: draw_sprites (NULL, NULL, NULL); This doesn't work so well when the first thing draw_sprites() does is try to dereference one of those parameters....
No.04509 Robbbert Moderator Jun 15, 2009, 09:17 edited on: Jun 15, 2009, 09:18	Please check MESS svn 5023 for a possible solution. Needs to be extensively tested.
No.04510 etabeta Developer Jun 15, 2009, 09:22 edited on: Jun 15, 2009, 09:24	it should be enough to replace line 438 with draw_sprites (machine->primary_screen, NULL, NULL); EDIT: not sure about Robbbert's change, but my fix is for the frameskipping crash and it works perfectly. never had any soft reset issue otoh, so maybe there are more problems
No.04522 R. Belmont Developer Jun 18, 2009, 01:27 edited on: Jun 18, 2009, 01:29	Yeah, eta's fix is correct. MESS #5023 is wrong, it treats the symptom rather than the problem.