01905: Many Sets: Fatal Error crash - Fatal error: Input ports cannot be read at init time! (src/emu/inptport.c:1278)

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
01905	Crash/Freeze	Critical (emulator)	Always	Jun 17, 2008, 16:56	Jun 25, 2008, 02:42

Tester	Tafoid	View Status	Public	Platform	MAME (Self-compiled)
Assigned To	aaron	Resolution	Fixed	OS
Status [?]	Resolved			Driver	misc/cubeqst.cpp
Version	0.125u6	Fixed in Version	0.125u7	Build
		Fixed in Git Commit		Github Pull Request #

Summary	01905: Many Sets: Fatal Error crash - Fatal error: Input ports cannot be read at init time! (src/emu/inptport.c:1278)
Description	Fatal error: Input ports cannot be read at init time! (src/emu/inptport.c:1278) Effects multiple sets and drivers. Report all here per instructions in Additional Information.
Steps To Reproduce
Additional Information	Added sets/drivers found with this as notes.. if you have more to add after initial note, edit your note. Sets effected: 8080bw.c - Most sets acefruit.c - All Sets alg.c - All Sets alpha68k.c - tnexspce arcadia.c - All Sets astrocde.c - demndrgn, ebases astrof.c - All Sets bloodbro.c - All Sets cabal.c - All Sets centiped.c - warlords cischeat.c - bigrun cps1.c - (3wonders, willow) + clones cubocd32.c - All Sets ddragon3.c - ctribe + clones, ddrago3b enigma2.c - All Sets exidy.c - teetert exidy440.c - hitnmiss, hitnmis2 findout.c - gt103+clones, quiz, quiz211 galdrvr.c - gmgalax, azurian, kingball, kingbalj getrivia.c - (gtsers1, sextriv1) + clones grchamp.c - All Sets itech8.c - gtg, gtg2j ksys573.c - Many sets ladybug.c - ladybug, ladybugb, ladybgb2 lethalj.c - cclownz m52.c - All Sets m57.c - All Sets m58.c - All Sets magic10.c - magic10, magic10a mw8080bw.c - Most Sets midzeus.c - crusnexo, thegrid, thegrida mpu4drvr.c - dealem namcofl.c - All Sets namcos22.c - (acedrvrw, cybercomm, raveracw, ridgerac, ridgera2) + clones neodrvr.c - irrmaze, popbounc, svcpcb, svcpcba onetwo.c - All Sets peplus.c - All Sets phoenix.c - All Sets except survivor playmark.c - All Sets but wbeachvl, wbeachv2 powerbal.c - magicstk seibuspi.c - ejanhs shisen.c - All Sets suprridr.c - All Sets taito_f2.c - qcrayon2 taitosj.c - kikstart tempest.c - All Sets toaplan2.c - (batrider, bbakraid, bgaregga) + clones ultratnk.c - All Sets wilytowr.c - wilytowr, atomboy, atomboya zaxxon.c - razmataz findout.c and getrivia.c merged to gei.c in 0.135u1 cubocd32.c changed to cd32.c in 0.141u3 cd32.c changed to cubo.c in 0.154
Github Commit

Flags
Regression Version	0.125u6
Affected Sets / Systems	Many Sets

Attached Files	less-aggressive-check.diff (2,324 bytes) Jun 18, 2008, 02:28 [Show Content]

No.01295 etabeta Developer Jun 17, 2008, 23:15	can any dev give more hints about what is wrong with these? empty input port? duplicate entries? what kind of sanity checks have been added? (I have no way to test these directly until SDLMAME is released, so any info would be useful :) )
No.01296 M.A.S.H. Senior Tester Jun 18, 2008, 00:57	To fix the Fatal Error crash use the old 0.125u5 emu\inptport.c file. Now you can play Wily Tower with sound
No.01297 aaron Developer Jun 18, 2008, 02:02	They are due to input_port_read being called during DRIVER_INIT/MACHINE_START/VIDEO_START or other early initialization time. When input ports are read at these times, the data is not valid, so the driver is acting on invalid information.
No.01299 aaron Developer Jun 18, 2008, 02:29	Actually, the check in 0.125u6 is too aggressive. Can we verify the reported crashes against the less aggressive check in the attached .diff?
No.01300 Tafoid Administrator Jun 18, 2008, 02:33 edited on: Jun 18, 2008, 05:05	As soon as I'm done the first sweep.. sure. Second thought, I'll try it when I get up in the morning - Checking 4k sets in one day is enough :) If someone feels compelled to check the list I compiled after applying that DIFF from Aaron - please do. If it's not done when I get up and around, I'll do it.
No.01303 aaron Developer Jun 18, 2008, 05:29	Thanks for being so thorough. :)
No.01307 etabeta Developer Jun 18, 2008, 07:23 edited on: Jun 18, 2008, 07:25	with less-strict checks we're down to 7 drivers with issues, and only namcos22 with more than 2 parents involved (mind that I only tested drivers in the list above) 8080bw.c: schaser 8080bw.c: polaris 8080bw.c: polarisa centiped.c: warlords grchamp.c: grchamp namcofl.c: speedrcr namcofl.c: finalapr namcofl.c: finalapb namcofl.c: finalapo namcos22.c: ridgerac namcos22.c: ridgerab namcos22.c: ridgeraj namcos22.c: acedrvrw namcos22.c: ridgera2 namcos22.c: ridger2a namcos22.c: ridger2b namcos22.c: raveracw namcos22.c: raveracj namcos22.c: raveraja namcos22.c: cybrcomm namcos22.c: victlapw neodrvr.c: svcpcb neodrvr.c: svcpcba ultratnk.c: ultratnk
No.01308 abelenki Tester Jun 18, 2008, 07:53 edited on: May 17, 2009, 19:49	etabeta, does less-aggressive-check.diff fix this bug?: 01906
No.01309 Layne Tester Jun 18, 2008, 08:27	ddragon3.c: ddrago3b
No.01310 Firewave Senior Tester Jun 18, 2008, 08:57	I am running a complete test right now and I hope it won't crash, so I get complete results with the net two days.
No.01312 aaron Developer Jun 18, 2008, 10:34	abelenki: No, that's a separate issue
No.01313 etabeta Developer Jun 18, 2008, 11:34	would you mind posting the patch for that issue as well, either here or in bug 1906? thanks
No.01314 Firewave Senior Tester Jun 18, 2008, 16:27	Additional sets so far (with the less restrictive patch): mw8080bw.c: boothill mw8080bw.c: checkmat mw8080bw.c: clowns mw8080bw.c: clowns1 mw8080bw.c: desertgu mw8080bw.c: dplay mw8080bw.c: einning mw8080bw.c: roadrunm dkong.c: 8ballact dkong.c: 8ballat2 dkong.c: dkong dkong.c: dkongjo dkong.c: dkongjo1 dkong.c: dkongjp dkong.c: dkongo dkong.c: drakton dkong.c: herbiedk dkong.c: herodk dkong.c: herodku dkong.c: hunchbkd dkong.c: radarsc1 dkong.c: radarscp dkong.c: sbdk dkong.c: shootgal dkong.c: spcfrcii dkong.c: spclforc dkong.c: strtheat vicdual.c: frogs crbaloon.c: crbalon2 crbaloon.c: crbaloon
No.01315 Tafoid Administrator Jun 18, 2008, 16:35 edited on: Jun 18, 2008, 16:36	Boy, that's odd that dkong didn't trigger on the run with more stringent testing and now it does? MAME is a strange beast :) Appreciate the complete run, Firewave!
No.01327 Firewave Senior Tester Jun 19, 2008, 18:05	Run is complete. Here are the remaining: canyon.c: canyon canyon.c: canyonp circus.c: crash crbaloon.c: crbalon2 crbaloon.c: crbaloon sprint2.c: dominos sprint2.c: sprint1 sprint2.c: sprint2 sprint2.c: sprint2a dragrace.c: dragrace firetrk.c: firetrk firetrk.c: montecar firetrk.c: superbug nitedrvr.c: nitedrvr skydiver.c: skydiver sprint4.c: sprint4 sprint4.c: sprint4a tank8.c: tank8 tank8.c: tank8a tank8.c: tank8b tank8.c: tank8c tank8.c: tank8d triplhnt.c: triplhnt
No.01376 hotcoke Tester Jun 25, 2008, 02:42	great!!