02806: astrof, spfghmk2, spfghmk22: Access Violation with -video none -str 2

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
02806	Crash/Freeze	Critical (emulator)	Sometimes	Jan 7, 2009, 14:21	Feb 25, 2010, 21:27

Tester	Firewave	View Status	Public	Platform	MAME (Official Binary)
Assigned To		Resolution	Unable to reproduce	OS
Status [?]	Closed			Driver	dataeast/astrof.cpp
Version	0.129	Fixed in Version		Build	Normal
		Fixed in Git Commit		Github Pull Request #

Summary	02806: astrof, spfghmk2, spfghmk22: Access Violation with -video none -str 2
Description	----------------------------------------------------- Exception at EIP=004B6CF5: ACCESS VIOLATION While attempting to read memory at 0023008C ----------------------------------------------------- EAX=000000F9 EBX=04EB2FF8 ECX=000142E0 EDX=000000F8 ESI=00000006 EDI=04AC9020 EBP=0022FCF8 ESP=0022FC80
Steps To Reproduce
Additional Information	spfgmk22 -> spfghmk22 in 0.133u1
Github Commit

Flags
Regression Version
Affected Sets / Systems	astrof, spfghmk2, spfghmk22

Attached Files	valgrind_7789.zip (1,433 bytes) Jan 8, 2009, 15:26

No.03514 Tafoid Administrator Jan 7, 2009, 14:50	Doesn't crash/exit in a 32-bit build (from MAMEDEV) on my machine.
No.03517 Mamesick Senior Tester Jan 7, 2009, 15:24	fine here too. 64-bit specific?
No.03518 Osso Moderator Jan 7, 2009, 16:20	They work fine on Vista64-MameUI64 too.
No.03520 aaron Developer Jan 7, 2009, 18:59	I'm guessing this is a debug build with the -debug switch.
No.03521 Tafoid Administrator Jan 7, 2009, 19:04	I tried the MAMEDEV produced 32-bit Debug build with -debug with no crash. Not sure where to go with this..
No.03530 Fujix Administrator Jan 8, 2009, 11:33	Firewave, your reports almost always need some options enabled or special binary or something extra to reproduce them, requiring time and efforts of many other testers. Could you mention if you use something special to test with, or test it on normal setup (such like a normal 32 binary on Win32) before you post it here?
No.03531 Firewave Senior Tester Jan 8, 2009, 13:18	My reports are normally very clear and I do state what you need to do. This is just one I didn't have time to properly investigate, so I just dropped it off (hence the missing backtrace). And I doubt Tafoid required a lot of time and effort on most of the reports I reported during the last u-cycle... But this is indeed very weird. It is caused by using the additional options "-video none -str 2" and it doesn't always crash.
No.03532 Firewave Senior Tester Jan 8, 2009, 14:34	I wasn't able to reproduce the crash in debug. It also doesn't crash with symbols (debug, non-debug and optimized). I will fire up valgrind once again.
No.03534 Firewave Senior Tester Jan 8, 2009, 15:26	valgrind reported a few warnings. Attaching the log.
No.03556 R. Belmont Developer Jan 11, 2009, 07:31	The valgrind trace does indicate a possible bug in the driver's video update. Dunno why it'd only manifest with those switches though.
No.03570 aaron Developer Jan 13, 2009, 19:07	The other thing that could be bad is that -str 2 takes a snapshot at the end, and we could be dying in the snapshot rendering.
No.03571 Firewave Senior Tester Jan 13, 2009, 19:39	Yeah, I also should have thought of that... I can't reproduce this anymore. It's really weird.
No.05765 Tafoid Administrator Feb 25, 2010, 21:24 edited on: Feb 25, 2010, 21:25	Current SVN 64-bit and 32-bit (pre -.136u4) cannot replicate this. Closing. If it happens again or more information can be obtained for these, we'll reopen.