ASM Question

[Maine]

now this might be a completely stupid question but i am gonna ask it anyway :P

Ok in x86 ASM i see they have

General Registers
Segment Registers
Pointer Registers

My question is these in regular programming language (VB,C++ etc)

General Registers would be like global variables?

Segment Registers would be variables that only exist in that function or in ASM the code block?

Pointer Registers would be pointers to data inside the code blocks that alter the segment registers/general registers?

now this might be wayyyyyy off the mark but i figure i would take a stab at it to see if i have a general idea of this ASM stuff

i read c++ on and off no pro but i like to do it as a hobby to to know

[CornedBee]

You're 100% wrong. Registers are always for pretty transient variables, and you cannot really compare them to variables of a higher-level language in any way.

[Raedwulf]

You're 100% wrong. Registers are always for pretty transient variables, and you cannot really compare them to variables of a higher-level language in any way.

Yes that is true. But when you use assembly language, especially when you translate a high-level algorithm -
You basically use eax, ecx, edx - and if you preserve ebx, ebp, esi, edi those can become in handy as well for being 'variables'. However these as 'global variables' is not quite correct as these 'variables' are global to the entire computer OS, drivers and all.
Preservation is the key.

Segment registers are generally unused in 32-bit usermode programming with the exception of the FS segment register which is used for Structured Exception Handling (SEH)

Common conventions are ESI, EDI for arrays & strings
ESP, EBP for stack
EAX for putting in a result (especially on function returns)
ECX for counter
EDX for data
EBX (I can't remember off the top of my head )

Cheers.

[CornedBee]

But when you use assembly language, especially when you translate a high-level algorithm -
You basically use eax, ecx, edx - and if you preserve ebx, ebp, esi, edi those can become in handy as well for being 'variables'.

Yes, but that's all function-local stuff. And registers are, as I said, extremely volatile. You must preserve some of them yourself across function calls, for example.

The global variables of hlls are only representable by reserving some real memory for them.

However these as 'global variables' is not quite correct as these 'variables' are global to the entire computer OS, drivers and all.

Not quite. Each execution context has its own set of registers - the physical registers will probably be shared, but logically each process has its own set.

EBX (I can't remember off the top of my head )

Used as the base for address calculations, usually.

[Raedwulf]

Not quite. Each execution context has its own set of registers - the physical registers will probably be shared, but logically each process has its own set.

Yeah i forgot to include that - the OS preserves them if i am correct.

Cheers.

[peachville]

Hi guys

I haven't programmed ASM in years [8086 code on a 33Mhz 286 in 1993...lol] but I found using the registers as defined [eg: cx for counts, ax, bx for math] was just good coding. They can be utilized any way you wish but when performing LIFO pushes and pops it just kept everything a little more organized.

[minor28]

Forget registers as variables. Assembler has four directives what's called sections, namely .DATA-section, .DATA?-section, .CONST-section and .CODE-section.

.DATA-section contains initialized data of your program. These data is stored in your program and will therefor affect the size of your executable file. Example

szAppName db "MyProgram",0

.DATA?-section contains uninitialized data of your program. In this section you just preallocate memory but not initialize it. The advantage of uninitialized data is it doesn't take space in the executable file. You only tell the assembler how much space you need when the program is loaded into memory.

.CONST-section contains declaration of constants used by your program. Constants in this section can never be modified in your program. It also affect the size of the executable file.

.CODE-section is where your codes reside.

I would compare .data, .data? and .const with VB Global or Public declarations.

In assembler you also have LOCALS. I would compare it with VB Dim in a Sub or Function or Private declarations. These locals ar volatile variables.

Locals are just preloocating stack space. For example I have a process

MyProcess proc hWin:dword
	LOCAL	lineno:DWORD
	LOCAL	linecount:dword
	LOCAL	buffer[32]:BYTE
	LOCAL	hDC:HDC
	LOCAL	rect:RECT
	LOCAL	hRgn:DWORD

There are 4 dwords, a buffer of 32 bytes and a rect structure of 4 dwords, totaly 64 bytes. Now you have to prelocate stack space for those locals. First save the stackframe with

push ebp
mov ebp,esp

and then prelocate stack space for your locals by decreasing the stackpointer with the number of bytes[/code]
sub esp,64[/code]

As you understand these locals hold garbage before you have put your values to them.

When you leave your process you must restore the frame pointer and release stack space

mov esp,ebp
pop ebp

and finally find the return address. In this case there is on process argument so you have to write

ret 4

.

This seems to be very complicated but if you use masm there are masm syntax that will do this work for you.

[CornedBee]

The .data? section is called .bss in many assemblers. Never assume that any two assemblers work even approximately the same, even if they're for the same platform.

[minor28]

Well, I'm working with masm32 but know some about GOASM, FASM, TASM or whatever they are called. But I have not come across .bss section as coding directive for uninitiated data. however I agree the directives differs between different assemblers.

In masm you have the directive .CODE [name]. If name is not specified the segment is called _TEXT.
The .DATA directive gives the segment name _DATA and .DATA? directive gives _BSS. For example if you want your dll lib to share uninitialized data you give the link.exe option /section:.bss,s.

Correct me if I am wrong.

Edit: Sorry Fasm directive is "section '.bss'" I recall.