AsmPro 1.19

[a/b]

Yeah, this is understandable, to let the assembler decide. Because, as I said, from the user perspective either will work 99.99%. Now, whether that's a correct decision or not, I'm not conviced either way. What asm-pro does is by chance (or how the 15KB+ of parser code was written ;P), and not by some higher logic. My subjective opinion, of course, based on the state of its source code.
And in some cases I'd prefer to have exactly what I put there. Maybe I'm doing some surgical black magic with the opcodes, maybe it's a 4KB intro where every byte matters and some opcodes compress better, etc. And I don't want to write ugly DC.Ws. I could always write cmpi if I wanted that, that's why it could be pretty important to honor what the programmer wrote. And vasm (since it was mentioned above) goes by that. Phx could say more about it, so I won't make too many assumptions.

[a/b]

v1.20 + source code uploaded to aminet, should be available in an hour or so...

Patch notes:

[ ADDITIONS ]
01: ASM: New symbol REPTN, value is set to 0 to (N-1) while inside a
    REPT block and -1 while outside.
02: ASM: New predefined symbol ASMPRO, value contains ASMPro version
    in the following format: major*256+minor.
03: ASM: New directive EQUA, intended for post-processing. Do not use
    if unsure how it works and what it does. It works similar to EQU
    but the value is absolute (it also "relocates" and can be used to
    access relocatable data from absolute sections).
04: ASM/DISASM: Added support for PLPA instruction (68060).

[ MODIFICATIONS ]
01. UI: reqtools.library v38+ is required instead of any version.
02. DISASM: FNOP is now displayed instead of a 0 displacement FBF.W.
03. ASM: Macro parameter \@ reimplemented to work correctly with
    system includes (pass1/2 conflicts with conditional assembly).
04: ASM: Trailing / is appended to INCDIR path if needed.
05: ASM: Current source directory is used as fallback to find include
    files, and finally the INCLUDE: assign as well.
06. ASM: Both CMP and CMPM with (Ax)+,(Ay)+ are now allowed.
07: ASM: Allow MOVEM and FMOVEM descending register lists.
08: ASM: Allow longword size for FMOVEM with control registers.
09: UI: The ancient prefs toggles and closewb code removed from the
    menu event handler.
10. Over 900 additional micro optimizations, 21KB reduction of the
    executable size compared to v1.18.

[ BUGFIXES ]
01. UI-1.19: File pattern in the read/write source file requester is
    incorrect if Source .ASM is enabled in the general preferences.
02. UI: Crash if diskfont.library couldn't be opened.
03. DISASM: Crash on 68000/010 systems if DBcc.L (Apollo) has a 2-char
    conditional code.
04. EDIT: Continuously pressing page up to reach the top of the source
    code after unsuccessful assembly (and similar variations), would
    freeze the editor.
05. DISASM: CALLM absolute and pc-relative modes don't work, immediate
    mode's hex operand is missing a leading $.
06. DISASM: Several CHK2 and CMP2 modes don't work.
07. ASM: Incorrect opcodes generated or not supported for CHK2 and CMP2.
08. DISASM: CINV and CPUSH specifier NC (neither cache) is displayed as
    BC (both caches).
09. ASM: CINV and CPUSH specifier NC (neither cache) is not implemented.
10. DISASM: MOVES destination absolute long mode doesn't work.
11. DISASM: PMOVE (68030 variant) mixes up registers TT0 and TT1.
12. ASM: FMOVEM and some FMOVE variants allow destination pc-relative
    modes.
13. UI: The changed source prompts are brought up at the end of restart
    instead of right away.
14. ASM: No CPU warning for CALLM (68020 only) if the selected CPU is
    68030 or higher.
15: ASM: INCDIR path is not reset when starting a new assembly.
16: ASM: Including a directory prints out random trash characters.
17: MONI: Scrolling left/right (alt-left/right arrow) hides the cursor.
18: ASM: PMOVE with MMUSR outputs incorrect register.
19: DBG: Debugged program's user and supervisor stacks (and corresponding
    USP/SSP registers) are switched.
20: DISASM: FMOVEM has a trailing "/" in some scenarios.
21: ASM: FMOVEM allows size .X with control registers.
22: DISASM: Infinite loop with some 68040 specific FPU opcodes.
23: ASM: Watchpoints and breakpoints are kept after switching to a
    different source.
24: Public port's name and type are not initialized.
25: ASM: PRINTV doesn't flush the print buffer properly causing the next
    line to be truncated (for example, if followed by INCLUDE or INCBIN
    the path is not printed).
26: UI: System freeze and/or crash if multiple hotkeys are pressed at the
    same time, and the first one is causing the menu strip change (for
    example, pressing amiga-AS in the editor).
27: ASM: Numerous FPU instructions with source immediate longword operand
    are generated as byte, resulting either in the out of 8-bit range error
    or incorrect opcode.
28: DISASM: FSINCOS prints only one destination FPU register.
29: DISASM: Absolute and pc-relative modes don't work with all bitfield
    instructions (68020+).
30: DISASM: CMPI pc-relative index mode (68020+) doesn't work.
31: DISASM: CAS allows several invalid and non-existing modes, absolute
    modes don't work.
32: DISASM: LPSTOP (68060) is not implemented.
33: DISASM: Typos in FPU ROM constant names (appended as a comment to the
    MOVECR instruction), probably due to the same typos in the M68K manual.
    Additionally, TAB changed to a space if Printer Dump is not enabled in
    the preferences to avoid weird characters on the screen.
34: ASM: PFLUSHR doesn't allow address register indirect mode.
35: ASM: MMU register PCSR is not recognized (PCR is checked twice instead).

[kamelito]

Great work thanks for sharing !

[StingRay]

04. EDIT: Continuously pressing page up to reach the top of the source
code after unsuccessful assembly (and similar variations), would
freeze the editor.

That really is a super annoying bug, I often had to start certain "rescue operations", i.e. attach ASM-Pro to a debugger to get out of the endless loop it was stuck in, just to not lose my unsaved source. Thanks for fixing that bug, I hope it will never bother me again.

[FlynnTheAvatar]

Thank you for the new version. But are there new requirements for 1.20?

I tried to run it on WB 2.1 (Kickstart 37.175, Workbench 38.26, 68000, 1MB Chip, 8MB Fast, no RTG), and it crashed after opening a new screen. I see the "Asm-Pro OS V1.20 By Genetic. Source 0 >>", and then a 8000 0003 GURU. Same crash with KS 3.1.

Following libs are installed:
mathtrans.library 37.1
asl.library 38.36
reqtools.library 38.1436

Update: It seems that ASMPro 1.20 needs at least a 68020. Is this intended?

Thank you,
Flynn

[a/b]

Thanks for the report, I'll look into it.
v1.19/1.20 should run on 68000. Part of my testing was to have asm-pro build itself on A500/68000/KS2, and making sure the output is 100% the same to what was produced on 68020+/KS3 and some other combos including what was produced by the original 1.18 (using WO instead of WX, of course).
The only change is the reqtools requirement of 38+ just in case someone is using an ancient (KS1.x era) version that doesn't support a screen mode requester. So in practical terms nothing changed.

I'm not able to reproduce the problem on my side, using a similar config with the same asl and reqtools version (mathtrans doesn't matter if you don't have an FPU). Could you please attach or PM me your Asm-Pro.pref file?

[FlynnTheAvatar]

@a/b:

Thank you for looking into it. I sent you a link to my preferences in a PM. Please not that ASMPro also crashes if there is no preferences file in env: .

It only works if I start the E-UAE with 68020 processor. So I guess there is an odd memory access somewhere.

[Don_Adan]

Some (or maybe most) reqtools.library versions doesnt works on 68000 (odd address error). I dont know if this is problem here.
I mean about latest versions.

[FlynnTheAvatar]

Good point. I am using 38.1436 that should be still working on 68000. At least the documentation does not say anything on requiring a 68020 or higher. The newer 39.3 is 68020+ though.

And I tried to remove the reqtools.library to be sure. In this case ASMPro opens a lowres screen but still crashes.

So, does not look like it is a reqtools.library issue.

[FlynnTheAvatar]

Sorry, it seems that the issue on my side.

The crash only happens when running on E-UAE (PPC JIT) on Amiga OS 4.1.

I tested the same HDFs and a similar configuration on Amiberry, and ASMPro did not crash there.

[kamelito]

@a/b
Version 1.19a works fine on my Vampire but 1.20 give me a Sotware failure #8000000B from WB #8000000A from cli
Without SS it give me #8000002F

[FlynnTheAvatar]

Similar here:

ASMPro 1.19c works fine with E-UAE. Just not ASMPro 1.20.

[a/b]

Where/when exactly does the crash happen? I've received another report for morphos ppc with some kind of a crashdump report, the problem there is related to gadtools library and the start-up window (where you allocate workspace memory).
I did some digging and found a potential problem, imsg is used after being replied (and eventually deallocated by gadtools during reply).

But it's maybe, maybe not... Do you both have that window open when you start or do you have it disabled (prefs->show startup win.), in which case the most recent workspace config is used? If disabled, the above problem shouldn't occur.

I don't have any exotic HW, only the actual legacy HW, access to PiStorm via friend, and also winuae, so any extra specifics and details about the problems you might have are welcome.

[kamelito]

The screen is opened if I close the software failure requester the menu can be opened but none work.

[paraj]

Very recently got a pistorm32, so usual a1260 is not hooked up, but I see the same issue.

Asmpro 1.19c starts with "select screenmode" and seems to work fine (though some RTG modes with very high resolution seem a bit borked, but no big deal, I'd never use them).

v1.20 also starts with "select screenmode" requester, after selecting a screenmode (all I've tried), I get $80000004 dead end guru PC=$3.

KS: 45.57, WB: 45.5, reqtools: 39.3, asl: 45.4

EDIT: Same if I boot cold from a floppy that has reqtools 38.1436 in libs

EDIT2: Works on my a1260 (KS 3.1)

EDIT3: Did not work w/ KS3.1 on pistorm booting from floppy

[FlynnTheAvatar]

For me it crashes right after opening the screen and before the ASM-Pro image and startup window is displayed. Please see this image on OneDrive for details:
https://1drv.ms/i/s!ApOoV0ole8uGrf5f...5NhtA?e=b7H9dp

Disabling the startup window via preferences does not prevent the crash for me.

[a/b]

v1.20a is in the zone, please let me know if that one works. My fix for "System freeze and/or crash if multiple hotkeys are pressed at the same time..." has side-effects on some configs .

[tomcat666]

1.20a working on pistorm32 now

[paraj]

Yep, 1.20a fixed it for me as well.

[kamelito]

1.20a fixed the issue for me too, you can reupload the all thing to Aminet.
I had to find tools to unpack 7z on Amiga

I have a source code that display « ** Include jam » every time
I assemble the source while on version 1.19a it does this once then works.
Dunno exactly why this message is displayed though.