Aira Force - an interactive reassembler

[hop]

Originally Posted by kamelito

Defender Microsoft

Thanks. I should be able to submit the file to Microsoft for analysis.

[kamelito]

@hop take Resource for example you can set the datatype of hex code to bytes, word, long, ascii, code this is what I meant. In IRA you just set data but you can’t force it to be byte or word or long.
https://retro-commodore.eu/files/dow...Manual-ENG.pdf

[hop]

Originally Posted by copse

I wrote up notes in my wiki.

Thanks very much. Looks useful.

Originally Posted by copse

I also tried all the compilers and assemblers I could get my hands on and documented the debug formats for symbols (it is very possible IRA already incorporates all of these perhaps even better) and that is in that hunk wiki page:

http://amiga-dev.wikidot.com/file-format:hunk#toc24

I had expected to write code to parse debug hunks, and was pleasantly surprised when ira already added symbols for debug hunks, at least in my very limited test cases. I'm not sure which of these formats ira supports. If you have a set of test debug binaries that would be much appreciated, else I'll build my own.

[kamelito]

@hop .device have the same structures as .library so it should be disassembled too.

[phx]

Originally Posted by hop

I had expected to write code to parse debug hunks, and was pleasantly surprised when ira already added symbols for debug hunks, at least in my very limited test cases. I'm not sure which of these formats ira supports.

Just symbols from

HUNK_SYMBOL

. The only well-documented

HUNK_DEBUG

format are

LINE

debug blocks, which do not make much sense, when you don't have the original source. Although, you can make vasm automatically generate them for the assembler source with option

-linedebug

.

Regarding byte/word/long data: Could be implemented, but it was never so important for me, as the size of data is already shaped by labels, relocations and recognised text.

[copse]

Originally Posted by hop

I had expected to write code to parse debug hunks, and was pleasantly surprised when ira already added symbols for debug hunks, at least in my very limited test cases. I'm not sure which of these formats ira supports. If you have a set of test debug binaries that would be much appreciated, else I'll build my own.

Sadly I was not that organised.

[kamelito]

Originally Posted by phx

Just symbols from

HUNK_SYMBOL

. The only well-documented

HUNK_DEBUG

format are

LINE

debug blocks, which do not make much sense, when you don't have the original source. Although, you can make vasm automatically generate them for the assembler source with option

-linedebug

.

Regarding byte/word/long data: Could be implemented, but it was never so important for me, as the size of data is already shaped by labels, relocations and recognised text.

How do you handle for example a long word where the first contains a word (it is not code) and the next one is code? You can’t force IRA to start the code at the second word as IRA just give you a long…

[hop]

Originally Posted by phx

Just symbols from

HUNK_SYMBOL

. The only well-documented

HUNK_DEBUG

format are

LINE

debug blocks, which do not make much sense, when you don't have the original source. Although, you can make vasm automatically generate them for the assembler source with option

-linedebug

.

Thanks for clearing that up. I was getting my hunks mixed up.

Originally Posted by phx

Regarding byte/word/long data: Could be implemented, but it was never so important for me, as the size of data is already shaped by labels, relocations and recognised text.

Redundant metadata. I guess it could maybe be handy for data referenced by undiscovered code, or debug/dev data unreferenced by "release" code.

[phx]

Originally Posted by kamelito

How do you handle for example a long word where the first contains a word (it is not code) and the next one is code? You can’t force IRA to start the code at the second word as IRA just give you a long…

Why not? You can start code at any even offset:

frank@altair cat tst.s 
        dc.w    -1
        rts
frank@altair vasmm68k_mot -Fhunkexe -quiet -o tst tst.s 
frank@altair cat tst.cnf 
MACHINE 68000
ENTRY $00000000
OFFSET $00000000
CODE $00000002 - $00000004
END
frank@altair ira -a -config tst

IRA V2.09 (Jun 29 2017)
(c)1993-1995 Tim Ruehsen (SiliconSurfer/PHANTASM)
(c)2009-2015 Frank Wille
(c)2014-2017 Nicolas Bastien


SOURCE : "tst"
TARGET : "tst.asm"
CONFIG : "tst.cnf"
MACHINE: MC68000
OFFSET : $00000000
codeAdrs: 0   codeAdrMax: 16
CodeArea[0]: 00000000 - 00000000
CodeArea[1]: 00000002 - 00000004


Pass 1: 100%
Pass 2: writing mnemonics
100%

frank@altair cat tst.asm 
; IRA V2.09 (Jun 29 2017) (c)1993-1995 Tim Ruehsen
; (c)2009-2015 Frank Wille, (c)2014-2017 Nicolas Bastien




        SECTION S_0,CODE

SECSTRT_0:
        DC.W    $ffff                   ;0
        RTS                             ;2: 4e75
        END

[kamelito]

@phx @hop yeah, sorry I meant under Aira Force is that possible?

[hop]

Originally Posted by kamelito

@phx @hop yeah, sorry I meant under Aira Force is that possible?

You can tweak CODE start and end addresses in the ira config window.

[kamelito]

Thanks

[hop]

Originally Posted by kamelito

Thanks

I've been meaning to do a few videos demonstrating the basic features. It's just hard to find the time.

[Vairn]

This is pretty sweet, well done.
Always like to see more stuff using ImGUI as a user interface, it is a cool gui system.

One issue I came across was that opening the extract window, made it come up in the small default imgui window size of like 40x40, and since you disable the resize feature, I had to go edit the window size in the ImGUI.ini file to make it visable.

[hop]

Originally Posted by Vairn

This is pretty sweet, well done.
Always like to see more stuff using ImGUI as a user interface, it is a cool gui system.

One issue I came across was that opening the extract window, made it come up in the small default imgui window size of like 40x40, and since you disable the resize feature, I had to go edit the window size in the ImGUI.ini file to make it visable.

Thanks and thanks for letting me know. I'll look into this for the next patch.

[Honitos]

@hop:
Your tool has the potential to be *very* useful!

One think I discovered: AiraForce - or one of the tools it uses - is not able to handle path names with spaces.
The path "TTE" in the screenshot *does* exist.

[hop]

Originally Posted by Honitos

@hop:
Your tool has the potential to be *very* useful!

One think I discovered: AiraForce - or one of the tools it uses - is not able to handle path names with spaces.
The path "TTE" in the screenshot *does* exist.

Hi. I hope so. Thanks for letting me know about this bug - I'll investigate.

[hop]

Originally Posted by Honitos

One think I discovered: AiraForce - or one of the tools it uses - is not able to handle path names with spaces.
The path "TTE" in the screenshot *does* exist.

Hi. Thanks for reporting this. It turns out that is wasn't the spaces, but the unicode character (ö) in your path.

I think that the Windows std::filesystem implementation uses wide chars, but SDL uses UTF-8, so I've updated the code to convert as required.

The fix will be available in the next patch.

[hop]

Originally Posted by Vairn

One issue I came across was that opening the extract window, made it come up in the small default imgui window size of like 40x40, and since you disable the resize feature, I had to go edit the window size in the ImGUI.ini file to make it visable.

I couldn't reproduce this problem, but I've removed the

ImGuiWindowFlags_NoResize

flag from the modal dialogue for the next release.

[copse]

I recorded a hapless user, myself, using it in case it highlighted anything useful for you. It crashed.

[ Show youtube player ]

Here are my notes:

1. Downloaded miniblast.lha from aminet and extracted the miniblast executable.
2. Downloaded latest version of aira force (0.6.3 I believe).
3. Loaded miniblast.
4. Clicked analyse.
5. Wasn't sure what to do next.
6. Scrolled down and found some code that had not been disassembled, including a section.
7. Tried to work out how to use hotkeys to jump to the thing referring to that code section. Ended up just searching for the symbol.
8. Converted the jump to code then tried to follow the reference back to the section by pressing right cursor.
9. Instant crash.