| View previous topic :: View next topic |
| Author |
Message |
ccovell
Regular
Joined: 19 Dec 2005
Posts: 100
Location: Japan
|
 Posted: Fri Dec 08, 2006 3:14 pm Post subject: Bug in MagicKit's branch handling |
 |
|
Hi, as long as I'm on the subject of bugs (see the BRAM topic), I thought I might as well point out a bug that I came across a long time ago while working on my game.
Here is a short listing from a section where I discovered MagicKit assembling my code incorrectly:
| Code: | 351 01:DA72 .pmsprite_loop1
352 01:DA72 spr_set sprite_no,satb
353 01:DA8B spr_x sx
354 01:DA98 spr_y sy
355 01:DAA4 spr_pattern spr_vram_point
356 01:DAB1 spr_ctrl #SIZE_MASK|FLIP_MASK,#SIZE_32x64|NO_FLIP
357 01:DABD spr_pri #1
358 01:DACB spr_pal #1
359
360 01:DAD5 EE CD 26 inc sprite_no
361 01:DAD8 addw #$20,sx
362 01:DAE9 EA nop
363 01:DAEA EA nop
364 01:DAEB EA nop
365 01:DAEC addw #$0010,spr_vram_point
366
367 01:DAFD AD CD 26 lda sprite_no
368 01:DB00 C9 3C cmp #$3C
369 01:DB02 D0 22 bne .nomovePauseSpr
370 01:DB04 subw #$80,sx ; that will center it on the screen
371 01:DB15 addw #$40,sy
372 01:DB26 .nomovePauseSpr
373 01:DB26 AD CD 26 lda sprite_no
374 01:DB29 C9 40 cmp #$40 ;Make sure this matches!!!!
375 01:DB2B D0 45 bne .pmsprite_loop1
376 01:DB2D 60 rts |
You can see that despite my .pmsprite_loop1 label being at a lower address in ROM, the branch that is generated branches forward to an invalid address. I can post other examples if anyone wants to see.
Before I found this bug, my game was crashing every time I made some loops longer (usually during sprite-drawing routines), and I couldn't figure out why. (Branches that are too long usually halt assembly and produce an error.) I guess this bug is caused by MagicKit not counting the bytes produced by macro routines (or functions) properly. David, any comments?
_________________
http://www.chrismcovell.com |
|
| Back to top |
|
 |
dmichel
Admin
Joined: 04 Apr 2002
Posts: 1166
Location: France
|
| Posted: Sat Dec 09, 2006 12:28 am Post subject: |
|
|
I remember such a bug but this was a long long time ago. I tried to reproduce it with the current version of pceas but the assembler generated an 'out of range' error in all cases. What version are you using? The latest is 3.21. If you have the source code of pceas check the class2 function in code.c file.
_________________
David Michel |
|
| Back to top |
|
|
ccovell
Regular
Joined: 19 Dec 2005
Posts: 100
Location: Japan
|
| Posted: Sat Dec 09, 2006 5:50 pm Post subject: |
|
|
I have the latest version that I got from here: http://www.magicengine.com/mkit/
| Quote: | 09/09/2000 :
New MagicKit and HuC released!
Assembler 2.5.0 -> 2.5.1 |
If you've been hoarding a much newer version, please release it! 
_________________
http://www.chrismcovell.com |
|
| Back to top |
|
|
Tomaitheous
Elder
Joined: 27 Sep 2005
Posts: 306
Location: Tucson
|
| Posted: Sat Dec 09, 2006 11:36 pm Post subject: |
|
|
Pceas 3.2.1 or whatever it is, comes with HuC. But it has two bugs that I found so far - all equates must not have any space between the const name and the beginning of the line:
junk .equ $4321 ; right
junk .equ $4321 ;wrong
And all instructions need to have atleast one space between them and the beginning of the line (reversed of the above).
Btw, doesn't Pceas expand all macros on the first pass?
-Rich
_________________
www.pcedev.net |
|
| Back to top |
|
|
dmichel
Admin
Joined: 04 Apr 2002
Posts: 1166
Location: France
|
| Posted: Sun Dec 10, 2006 1:41 am Post subject: |
|
|
Yup, 3.21 is part of HuC. It's not really a new version thought, it just has a few extra stuff that was needed by HuC, I think 2.51 already had the 'out of range' buf fix, it's strange, may be you will have the same problem with 3.21 too, let me know if this is the case.
_________________
David Michel |
|
| Back to top |
|
|
dmichel
Admin
Joined: 04 Apr 2002
Posts: 1166
Location: France
|
| Posted: Sun Dec 10, 2006 1:50 am Post subject: |
|
|
| Tomaitheous wrote: | | Pceas 3.2.1 or whatever it is, comes with HuC. But it has two bugs that I found so far [...] |
Those are not bugs, they are features. 
You can go around the first one by adding a colon to the label :
junk: .equ $4321
For the second one there's no solution thought, anything that start at the beginning of the line is considered a label.
| Quote: | | Btw, doesn't Pceas expand all macros on the first pass? |
Yup, it does.
_________________
David Michel |
|
| Back to top |
|
|
Tomaitheous
Elder
Joined: 27 Sep 2005
Posts: 306
Location: Tucson
|
| Posted: Mon Dec 11, 2006 5:58 am Post subject: |
|
|
| Quote: | | Those are not bugs, they are features. |
_________________
www.pcedev.net |
|
| Back to top |
|
|
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
