What's new
By:

Midway (Williams) Y unit (Z unit and T unit) Discussion, Serial Numbers etc

Ah, i forgot about that.. i restarted from scratch, and it seems i 'solved' it by copying the SYS.INC from the backup folder... maybe that's my issue?

Basically it errors on linking with references:

1680015705441.png


Maybe includes or my tables have an error, i don't know, it's been a busy work day lol.
Did you set up any folders outside C:\smash\ ?
 
Last edited:
Asure said:
Ah, i forgot about that.. i restarted from scratch, and it seems i 'solved' it by copying the SYS.INC from the backup folder... maybe that's my issue?

Basically it errors on linking with references:

1680015705441.png


Maybe includes or my tables have an error, i don't know, it's been a busy work day lol.
Did you set up any folders outside C:\smash\ ?
Click to expand...
Looking at that screen above the first line is SPAWN_FAT which is a .def in FATMAN.asm not a .lod thing so I'd check here first

I still cannot get the build to run past the cpu board test screen , just exits out of MAME with segmentation error 11. and reboots constantly on real hardware.

I even started from scratch with the full set of files from the backup folder, as I saw that the backup files appear to be rev8 whereas the main files appear to be rev5. (Using the AUDIT file and ROM_NAME def)

Kind of stuck at this point...
 
Also , found some info about checksums (and how to link in SREC) in the trog files that makes sense, but can't find any reference to this type of thing in the smash tv build??

IMG_6447.JPG
 
It is with great joy i can now report that i'm also able to link and build the ROBO.OUT file.
There's an error in the makefile, so just doing gmake -m will produce the error i was reporting.
GSPLNK robo.cmd works just fine to produce the OUT object :)

1680095852281.png
 
Asure said:
It is with great joy i can now report that i'm also able to link and build the ROBO.OUT file.
There's an error in the makefile, so just doing gmake -m will produce the error i was reporting.
GSPLNK robo.cmd works just fine to produce the OUT object :)

1680095852281.png
Click to expand...
Should have sent you my makefile, never mind though we got there!

Interesting warning line in the linker though, same as on mine, so could this be our issue getting it to run correctly?

Let me know when you have use the srec files I sent over and if you get stuck on the same point running it in same or on real hardware.
 
Just tried it, i built with the main copy, not the one from BACKUP, and the main one it just sits on black screen for me.

For checksums, this should work for example:
Code: 
*****                                                                                                                 
FFE00000

2

ROBO.OUT                                                                                                                                                                                                                                      

128                                                                                                                   
34010

A

Srec.exe mentions this in its info when you run it manually, but it prints an error that some ROM part is not empty.. Hrm..

I need to double check the BACKUP build later for you, real life calls :)
 
Asure said:
Just tried it, i built with the main copy, not the one from BACKUP, and the main one it just sits on black screen for me.

For checksums, this should work for example:
Code: 
*****                                                                                                                
FFE00000

2

ROBO.OUT                                                                                                                                                                                                                                     

128                                                                                                                  
34010

A

Srec.exe mentions this in its info when you run it manually, but it prints an error that some ROM part is not empty.. Hrm..

I need to double check the BACKUP build later for you, real life calls :)
Click to expand...

Yes that's what I've been on about, smash tv does not have the cheksum def defined peoperly for srec use and they are just hard codes in there... See my posts from earlier up
 
Asure said:
Just tried it, i built with the main copy, not the one from BACKUP, and the main one it just sits on black screen for me.
Click to expand...
Should still run the diag screen even with the regular files

did you set the file names correct

FE0 -> U105
FE1 -> U89
 
Experimented with adding some better automated checksum generation based on the TROG files

After awhile of creating an updated diag.asm file and some other stuff:

-----------------------------------------------------------------------
Enter FILE address equivilent to EPROM address 0 [0000]: FFE00000

Fill character [FF]:

Interleave Factor [1]:2

Enter Interleave Remainder
(Same as Interleave factor for ALL) [2]:

Input file name [.MOT]: ..\OUTPUT\ROBO.OUT
processing ..\OUTPUT\ROBO.OUT as a COFF file.
Input file name [.MOT]:

Size of output file in Kilobytes [128]: 128

Checksum calculation (part number) [0]:34010

checksum for part 0 = DBC9
Output file name [..\OUTPUT\ROBOfe.0 ]:

outputting 128K (1024K bits) to ..\OUTPUT\ROBOfe.0

processing ..\OUTPUT\ROBO.OUT as a COFF file.
checksum for part 1 = 4CFE
outputting 128K (1024K bits) to ..\OUTPUT\ROBOfe.1
C:\video\SMASHTV\ROMS>

-----------------------------------------------------------------------
 
Ok so i looked at my output more closely, and my roms suck, mostly FFFF filled, which is why they blackscreen. I'm dumb. Sorry.
Compared with my old LRN and mame's trace logs, i see a lot of differences.

I checked the MAP files and compared CMD files, i noticed yours are quite different from the original.
(I couldn't get my roms to build unless i adapted the LRN file to my output and started at 0xFFC00000)

Checked sys.inc and they all have it set for 0xFFC00000
STACK_ST .SET 0FFFC0000H

Mame confirms the original production code for LA8 starts at 0XFFE00000

Wondering what kind of hardware this code was for. (Some even have the YUNIT EQU 0) set..
Total carnage has the same thing, 0xFFC00000
I cannot check against your SYS.INC ..

Trog has similar funnythings. STACK_ST .set 0FFFC0000h in the sys include, but FFFE in the linker CMD.

Maybe a closer look at SYS.INC and the CMD file for the linker is needed.
I had some rom(tried a lot of things tonight..) booting badly with a blue bordered message screen on top, so palette was wrong and no text in it..
 
well, I am still working through things and I am learning an awful lot about how this system is written and compiled. I tested some roms on real hardware today and the thing is trying to run but still crashes out when the attract starts. I have traced through the main program to see how it starts up runs though the diags then to cmos checks and then to attract etc.

I also found this when I run on real hardware. Different to mame. You can see the text being outputted but all corrupt. Which is better than mame, as I thought it wasn't displaying at all. Makes more sense.

IMG_6466.JPG
IMG_6467.JPG


Then this in attract before it crashes and reboots

IMG_6469.JPG


But you can enter the test mode if you are quick enough !!!


IMG_6471.JPG


Which works even though the text is all messed up.

But looks at this!!! Some tests are working fine!!!

IMG_6472.JPG
IMG_6473.JPG
IMG_6475.JPG


So I need to spend more time understanding what is the different between the switch test text and the main diag test text. This will hopefully give me a clue as to the issue area or the area that's got a wrong address or makeup in it.

For ref, The start address for the roms that works here is the 0XFFE00000 one
 
I think i've solved it.

Our problem in the the way load2 outputs and we are missing original img/lod parser.
After the tests, the attract screen is the first place we get actual gfx on screen.. i figured it might be this last night as i couldn't sleep.

I took the trog source, it has IMGTBL.ASM already built in it.
I used load2 to generate it again from trog.lod. What could go wrong. Well, a lot apparently.

Quick screenshot, left is load2, right is original included with trog src.

1680355320942.png


However, i think we only need the GLO output from load2, both IMGTBL.ASM and BGNDTBL.ASM are included in the BACKUP folder.. :)

1680356050547.png


On the left side is the one we are currently using built by load2/newload exe. On the right side the original IMGTBL.ASM from SmashTV backup folder.

You can see the problem where it puts part of the data in a second .WORD and the location .LONG does not match.
So the build we have going now reads from wrong gfx locations and crashes in Mame, and shows random corruptness on real H/W.
Might be some misalignment with gfx even in the files from backup, i don't know. Also they need hex fixings >xxxx to 0xxxxH and bla bla.

If we fix this and build, we still have an issue where we are missing COBRA.TBL, FLATFACE, GALAGAS etc tables...
But at least title screen should look better/work? Maybe the game will crash once those sprites go on screen.
If i had the time, i would be building it right now, but soon our daughter will wake from her nap so nope. sry.

Looking forward to some new screenshots from you hopefully!
I'll be back later tonight!

P.S. i found the source for load2 if you want to fix this or build a new one from scratch. Check PM.
 
I'll just leave this here:

View: https://www.youtube.com/watch?v=y1fOi09K728


Respect to @mypinballs , without the things he found we would have never gotten here. Git updated with this build.
It's v5 from the root of the source leak. It's not 100% and spawns no enemies for me for some reason.
It crashes when it wants GFX from one of the TBL files we cannot generate properly and that are missing from the source leak.
Also it does wierd things with the inputs where start buttons give me coins, and i can't shoot.

At least we got this far :)
 
Asure said:
I'll just leave this here:

View: https://www.youtube.com/watch?v=y1fOi09K728


Respect to @mypinballs , without the things he found we would have never gotten here. Git updated with this build.
It's v5 from the root of the source leak. It's not 100% and spawns no enemies for me for some reason.
It crashes when it wants GFX from one of the TBL files we cannot generate properly and that are missing from the source leak.
Also it does wierd things with the inputs where start buttons give me coins, and i can't shoot.

At least we got this far :)
Click to expand...
Nice one. Good find with the load to load2 program output differences. Annoying that we can’t find the original load program executable but with that source for the later one maybe we can work out how to make a new one that will work for load1 lod files better. It’s amazing that this affects the text generation too. I need to see why this is.

I was out on a days easter drinking with friends yesterday so just picking this up today/tomorrow :)
 
Note I have changed the version to v9 for any changes I make. I am using the backup folder files which I think are the v8 version instead of the v5 files.

It is interesting that a lot of the files built from the load2 and .lod file are the same and work ok. Only the .tbl and *tbl.asm files seem to be the ones that have an issue. I am looking into it to see if I can work out how to make a load1 program again
 
You must log in or register to reply here.
Back
Top