Viewing Issue Advanced Details
ID Category [?] Severity [?] Reproducibility Date Submitted Last Update
07016 Crash/Freeze Critical (emulation) Always Jun 12, 2018, 16:06 6 days ago
Tester Kale View Status Public Platform MAME (Self-compiled)
Assigned To AJR Resolution Fixed OS Windows Vista/7/8 (64-bit)
Status [?] Resolved Driver
Version 0.198 Fixed in Version 0.273GIT Build 64-bit
Fixed in Git Commit ca1c02d Github Pull Request #
Summary 07016: astrowar: game resets at start of stage 5
Description Astro Wars resets itself at start of stage 5, right after drawing the background vertical lines and before showing the boss.
Steps To Reproduce Reach stage 5 or load attached save state (need to kill stage 4 last enemy).
Preferably load it after the POST, save states aren't offcially supported by the game.
Additional Information Culprit seems to be this snippet:
1:09F9: loda,0 $35ad ;reads s2636 port
1:09FC: comi,0 $80
1:09FE: bctr,p $2a14
1:0A00: bdrr,3 *$29fa
0:15AD: loda,0 $0000 ;jumps to s2636 area?
Could be either s2650 or s2636 core bug.

Seldomly this doesn't occur, by hitting the last enemy in specific spots?

Seems to happen up to version 0.169
Github Commit
Flags
Regression Version
Affected Sets / Systems astrowar
Attached Files
zip file icon r.zip (1,191 bytes) Jun 12, 2018, 16:06 Uploaded by Kale
png file icon 0017.png (1,978 bytes) Jun 12, 2018, 16:08 Uploaded by Kale
Snapshot of the game prior to reset.
Kale
Relationships
There are no relationship linked to this issue.
Notes
5
User avatar
No.15169
kazblox
Tester
Jun 12, 2018, 20:22
This driver is poor already (among the many things is that video needs a total rewrite to support the actual RAW_PARAMS configuration which is similar to Lazarian), but this does look like a core bug at a glance. Nice catch.
User avatar
No.15179
AJR
Developer
Jun 15, 2018, 15:46
I suspect that this might actually be a bad dump. In context, it seems as if the BDRR,3 opcode (FB) ought to be COMR,3 (EB) instead.
User avatar
No.15337
AJR
Developer
Aug 21, 2018, 18:55
I've decided to mark one ROM as a bad dump for now, in the hope that it can be redumped and verified. There isn't another dumped set for code comparison purposes, the code performs no self-test (contrary to the "POST" mention in the report), and, given how stupidly difficult the game is to play, I fear it might have actually shipped with this bug.
User avatar
No.15338
Osso
Moderator
Aug 22, 2018, 05:56
FWIW I've asked an Italian owner for a redump, he agreed but it won't be in the near future, as he doesn't have access to his games right now.
User avatar
No.22510
hap
Developer
7 days ago
edited on: 7 days ago
I received another ROM dump, but it looks very suspicious to me, I think it was modified by a previous pcb owner when reprogramming 1980 EPROMs that had gone bad.
Suspicious in the sense that only the bad opcode was changed, and changed so much that it can't be flipped bits. It was changed to E4 40.

(...)
1:09F3: 06 27 lodi,2 $27
1:09F5: 07 10 lodi,3 $10
1:09F7: BB 86 zbsr *$0006 ; did not appear to change R3 when i checked live disasm
1:09F9: 0C 15 AD loda,0 $35ad ; random number?
1:09FC: E4 80 comi,0 $80 ; compare rand with $80
1:09FE: 19 14 bctr,p $2a14

1:0A00: EB F8 comr,3 *$29fa ; EB F8 MAME + AJR bitflip, *$29fa = $35ad, so it compares R3($10) with rand
1:0A00: E4 40 comi,0 $40 ; E4 40 hack?

1:0A02: 19 08 bctr,p $2a0c
1:0A04: 04 7B lodi,0 $7b
1:0A06: 05 1A lodi,1 $1a
1:0A08: 06 47 lodi,2 $47
1:0A0A: 1B 1A bctr,a $2a26
1:0A0C: 04 A3 lodi,0 $a3
1:0A0E: 05 1A lodi,1 $1a
1:0A10: 06 E7 lodi,2 $e7
1:0A12: 1B 12 bctr,a $2a26
1:0A14: E4 C0 comi,0 $c0 ; compare rand with $c0
1:0A16: 19 08 bctr,p $2a20
1:0A18: 04 53 lodi,0 $53
1:0A1A: 05 19 lodi,1 $19
1:0A1C: 06 A7 lodi,2 $a7
1:0A1E: 1B 06 bctr,a $2a26
1:0A20: 04 3B lodi,0 $3b
1:0A22: 05 19 lodi,1 $19
1:0A24: 06 47 lodi,2 $47
1:0A26: 07 34 lodi,3 $34

So, at the start of the bossfight, it sets up R0,R1,R2 with 4 possible sets of values, based on a random number.
current MAME: 50% chance it will crash
AJR's bitflip: 25% 25% [43.75% 6.25%](depends on R3)
hack ver: 25% 25% 25% 25%