Data directive with custom size

flat assembler > Programming Language Design > Data directive with custom size

Goto page 1, 2 Next

Author

Thread

Tomasz Grysztar
Assembly Artist

Joined: 16 Jun 2003
Posts: 7310
Location: Kraków, Poland

Since the design phase I planned fasmg to have a special data directive in addition to fasm-compatible DB/DW/etc. that would allow specify a size of data unit. Let's call it DBX for now. It would be used like:

Code:

dbx 1, 1,2,3 ; same as db 1,2,3
dbx 4, 4,5,6 ; same as dd 4,5,6

The code is ready, adding such directive is a matter of a few minutes. The only reason why I did not enable it is because I could not decide how to name it. Though, now that I think about it, I could just use a semi-random name like above and then just add a synonym if we decide some other name is better (after all, there are already directives with synonyms, like REPT/REPEAT and IRP/ITERATE).

The other name I came up with is DLE, as "define little-endian". It is nicely describing what it does, I'd only be afraid that users would then expect DBE counterpart to be available, and that's outside of my plans. (I honestly don't know how such theoretical DBE should treat string arguments, for example. If you want it, write a macro.)

The obvious advantage of having such directive would be that it would allow to define data units with non-standard sizes. But the other reason for having it is to make overloading data directives easier.
For example, and architecture like ARM needs DH as an equivalent to fasm's DW, while DW should do the same as fasm's standard DD. If we simply defined DH and DW as macros forwarding the arguments to DW and DD, it would not work because DH would call DW macro, which would then pass it to DD. Thus a more complex variant set of macro is needed, based on a scheme like:

Code:

DATA_SIZES_ARM = 1

struc dw? data&
        if DATA_SIZES_ARM
                dd data
        else
                dw data
        end if
end struc

struc dh? data&
        DATA_SIZES_ARM =: 0
        dw data
        restore DATA_SIZES_ARM
end struc

With directive like DBX/DLE redefining standard size would be as simple as:

Code:

struc dw? data&
        dbx 4,data
end struc

struc dh? data&
        dbx 2,data
end struc

As I said, the directive has been ready to fly since the earliest releases of fasmg, all we need is a name. What do you think about the ones I used here? Or do you have a better one?

19 Sep 2016, 12:33

zhak

Joined: 12 Apr 2005
Posts: 490
Location: Belarus

what about naturally aligned data types?

dbx sounds good. but what about RB? to reserve data? do you plan to add it? rbx woudn't work then

19 Sep 2016, 15:00

Tomasz Grysztar
Assembly Artist

Joined: 16 Jun 2003
Posts: 7310
Location: Kraków, Poland

zhak wrote:

what about naturally aligned data types?

Any more high-level concepts are to be implemented through macros. This directive is an example of a low-level tool that macros can access to provide higher level functions.

zhak wrote:

dbx sounds good. but what about RB? to reserve data? do you plan to add it? rbx woudn't work then

I don't think we need the reserving counterpart, since all that it would need to do can be summarized by this simple macro:

Code:

struc rbx? unit*,count*
        label . : unit
        rb (unit) * (count)
end struc

And even if you needed to redefine RB for some purpose, you could still access the same functionality with "DB unit*count dup ?" (or "DBX unit, count dup ?").

19 Sep 2016, 15:17

Tomasz Grysztar
Assembly Artist

Joined: 16 Jun 2003
Posts: 7310
Location: Kraków, Poland

I uploaded the fasmg package that has it implemented under the name DBX. I can still change the name (or add a synonym) later.

19 Sep 2016, 15:29

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 16699
Location: In your JS exploiting you and your system

DV - Define Variable sized value(s)

19 Sep 2016, 15:43

Tomasz Grysztar
Assembly Artist

Joined: 16 Jun 2003
Posts: 7310
Location: Kraków, Poland

revolution wrote:

DV - Define Variable sized value(s)

Wouldn't this one be more fitting for the variant having a size declared independently for each argument?

19 Sep 2016, 15:45

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 16699
Location: In your JS exploiting you and your system

Tomasz Grysztar wrote:

revolution wrote:
DV - Define Variable sized value(s)
Wouldn't this one be more fitting for the variant having a size declared independently for each argument?

Perhaps yes.

DX - Define random sized value(s)

19 Sep 2016, 15:47

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 16699
Location: In your JS exploiting you and your system

DS - Define Specified size value(s)

19 Sep 2016, 15:48

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 16699
Location: In your JS exploiting you and your system

DA - Define arbitrary size value(s)

19 Sep 2016, 19:15

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 16699
Location: In your JS exploiting you and your system

Actually if/when this is implemented then perhaps all other data directives should be eliminated. Since this would be the core definition that others can be derived from.

19 Sep 2016, 19:18

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 16699
Location: In your JS exploiting you and your system

D - Data

DATA - Data

19 Sep 2016, 19:25

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 16699
Location: In your JS exploiting you and your system

Some more suggestions from one of my colleagues:

EMIT
OUTPUT
OUT
PUT

I'll try to stop spamming this topic now, sorry.

19 Sep 2016, 20:01

jmg

Joined: 18 Sep 2016
Posts: 62

Tomasz Grysztar wrote:

.... that would allow specify a size of data unit.
Let's call it DBX for now. It would be used like:
... all we need is a name. What do you think about the ones I used here? Or do you have a better one?

My main aversion to dbx would be that intel code is already a sea of x's...
I also dislike too-terse mnemonics, so 2 letters are out...

Perhaps
dbvw ? - Data Byte Variable Width
dbfw ? - Data Byte Field/Fixed Width

What else can this do ?
Currently db 1,23,"hello",00 is allowed

so, what about a variant syntax using : like

dbfw 1: 1,2,3 ; 00 01 02

dbfw 2: 1,2,3 ; 0000 0001 0002

The colon now also allows this
dbfw 3: 1,2,3, 2:0x55 ; 000000 000001 000002 0055

and even this...

dbfw 9: "StrA","StringB","22" ; right justified strings, " " packed ?
dbfw -9: "StrA","StringB","22" ; left justified strings, " " packed ?

19 Sep 2016, 20:11

Tomasz Grysztar
Assembly Artist

Joined: 16 Jun 2003
Posts: 7310
Location: Kraków, Poland

revolution wrote:

Actually if/when this is implemented then perhaps all other data directives should be eliminated. Since this would be the core definition that others can be derived from.

I had a similar thought, but at the same time fasmg contains predefined values like "word" and "dword" too, even though defining them is even easier. The main point here is that everything can be redefined, and some symbols are there by default just to make the environment a bit more friendly to a fasm user coming to fasmg.

jmg wrote:

My main aversion to dbx would be that intel code is already a sea of x's...
I also dislike too-terse mnemonics, so 2 letters are out...

Perhaps
dbvw ? - Data Byte Variable Width
dbfw ? - Data Byte Field/Fixed Width

What about something mentioning the endianess? DLTTLE? Wink

jmg wrote:

The colon now also allows this
dbfw 3: 1,2,3, 2:0x55 ; 000000 000001 000002 0055

In fasmg macros are your [obligatory] friend:

Code:

macro dbfw? args&
        local unit
        unit = 1
        iterate arg,args
                match size:value, arg
                        unit = size
                        dbx unit,value
                else
                        dbx unit,arg
                end match
        end iterate
end macro

dbfw 3: 1,2,3, 2:0x55

It is part of that "complex solutions with simple features" philosophy I declared for fasm (I even said once that I see fasmg as the ultimate embodiment of that idea).

jmg wrote:

dbfw 9: "StrA","StringB","22" ; right justified strings, " " packed ?
dbfw -9: "StrA","StringB","22" ; left justified strings, " " packed ?

I'm going to leave writing such macro as an excercise to the readers. Wink

It is getting late where I'm sitting.

19 Sep 2016, 20:27

jmg

Joined: 18 Sep 2016
Posts: 62

Tomasz Grysztar wrote:

The main point here is that everything can be redefined, and some symbols are there by default just to make the environment a bit more friendly to a fasm user coming to fasmg.

I think it is a good idea to have "expected defaults" - ie that any ASM user would anticipate, but being able to overlay/redefine, covers all the bases...

19 Sep 2016, 20:39

bitRAKE

Joined: 21 Jul 2003
Posts: 2791
Location: dank orb

https://en.wikipedia.org/wiki/Word_(computer_architecture)

DWLD = data word length data, or just WLD (wild, lol) Very Happy

(I was thinking FASMG was a little too x86-centric when I was reading the manual. Even though this was the intent, it still feels odd.)

19 Sep 2016, 20:41

Tomasz Grysztar
Assembly Artist

Joined: 16 Jun 2003
Posts: 7310
Location: Kraków, Poland

bitRAKE wrote:

(I was thinking FASMG was a little too x86-centric when I was reading the manual. Even though this was the intent, it still feels odd.)

Yes, and this is mostly because I wanted to preserve as much of fasm as possible. When designing fasmg I also still considered that it could become a base engine for fasm 2. But now I have more fun writing macros than even thinking about fasm 2.

19 Sep 2016, 20:53

jmg

Joined: 18 Sep 2016
Posts: 62

Tomasz Grysztar wrote:

jmg wrote:
The colon now also allows this
dbfw 3: 1,2,3, 2:0x55 ; 000000 000001 000002 0055
In fasmg macros are your [obligatory] friend:
Code:
macro dbfw? args&
        local unit
        unit = 1
        iterate arg,args
                match size:value, arg
                        unit = size
                        dbx unit,value
                else
                        dbx unit,arg
                end match
        end iterate
end macro

dbfw 3: 1,2,3, 2:0x55    
It is part of that "complex solutions with simple features" philosophy I declared for fasm (I even said once that I see fasmg as the ultimate embodiment of that idea).

Yes, I am new to fasmg, but what can be done with macros is frankly amazing.

I just tried the above (of course forgot I do not yet have dbx..)

No surprise it gives this

invert.asm [99] macro ? [1] macro dbfw [6]
Error: illegal instruction.

I'm guessing
[99] is source line#,
[1] is ? macro level ?,
[6] is line# inside macro dbfw,

but the following

Error: illegal instruction.

is rather terse, and could be clearer, & much easier to find if it also indicated the offending word ? - here, dbx

Last edited by jmg on 19 Sep 2016, 20:56; edited 1 time in total

19 Sep 2016, 20:54

JohnFound

Joined: 16 Jun 2003
Posts: 3500
Location: Bulgaria

The name is not very important here. More important is that the suggested syntax does not distinguish between the data and the data size. In order to stay readable enough, it must separate the data size from the data values.

My suggestion is to use only "D" combined with the size of the data:

Code:

d1 1,2,3 ; same as db 1,2,3
d4 4,5,6 ; same as dd 4,5,6

Or if the single "d" is not enough, it can be "dv" from "define variable":

Code:

dv1 1,2,3 ; same as db 1,2,3
dv4 4,5,6 ; same as dd 4,5,6

Another approach is to use another separator for the size, like ":"

Code:

dbx 1: 1,2,3 ; same as db 1,2,3
dbx 4: 4,5,6 ; same as dd 4,5,6

19 Sep 2016, 20:55

jmg

Joined: 18 Sep 2016
Posts: 62

JohnFound wrote:

My suggestion is to use only "D" combined with the size of the data:
Code:
d1 1,2,3 ; same as db 1,2,3
d4 4,5,6 ; same as dd 4,5,6    

I thought of that, but that does not allow an equate to set the field width, which could be useful.

JohnFound wrote:

Another approach is to use another separator for the size, like ":"
Code:
dbx 1: 1,2,3 ; same as db 1,2,3
dbx 4: 4,5,6 ; same as dd 4,5,6    

I agree that is better, (now can be equate set) and as the macro example above hints, this Size: can also now be redefined along the line too.

19 Sep 2016, 20:59

Goto page 1, 2 Next

< Last Thread | Next Thread >

Forum Rules:

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum