Hello everyone, I know this is a Linux forum but I was wondering if anyone could tell me why the below code acts oddly with DOS printing function? Program is compiled using NASM with nasm -o test.com test.asm. I am running the program in a command prompt on windows 7 but have also tested it with a win98 boot disk and I get the same result?

Here is the code
Here is the output

Is it a .com file? Or an .exe? Shouldn't you set the segment registers first?

Shouldn't you load the address of the message string with lea, instead of mov?

1 members found this post helpful.

This is a .com file and the org 100 sets all the registers to the correct values. I could change them if I wanted to.


It does not matter. Both statements in NASM do the same thing. I think lea creates slower code. You can change the mov dx, message to lea dx, [message] in the code and it does exactly the same thing.

See this post for more clarity on this if you care

1 members found this post helpful.

Thanks for the pointer. Never knew that

I'd prefer org 256 or org 0x100

Just out of curiosity, what's the "$" for?

Are you sure it shouldn't be mov dx, [message] ? i.e. with big brackets? (been a while since I dabbled in stuff like this, would be good to get my skills up again! )

That loads into dx the offset of message from the beginning of the segment (.data) not its offset from the location pointed to by the ds register in a .com file.

It has been too many years since I did such things for me to remember the correct syntax in nasm to refer to offsets within a .com

In masm, you would declare a group (typically named dgroup) containing all the segments and then specify the offset relative to dgroup.

1 members found this post helpful.

'$' is the string-terminator character in some obsolete INT 21H call (CP/M legacy).

Alright I figured out what I was doing wrong :P and I am stupid. NevemTeve had it right when I should have changed the org to org 0x100 instead of org 100 as they are different origin addresses and the way NASM works is to use that origin as the offset for everything else in the program.

johnsfine - NASM works a bit different then MASM with the org directive so I think that's what might have been throwing you off there.

NASM will load the offset from the Origin specified from the ORG directive rather then the beginning of the segment.

It is the register pair DS : DX that is the pointer to the string. Its been a while, and I never really did use NASM, but I think you need to explicitly load DS with the segment address of the .data segment.

--- rod.

That is a DOS issue, not a NASM issue. Your suggestion could be used, but doing it that way is neither practical nor common.

In a .com file in DOS (or DOS mode of other OS's) the PSP (program segment prefix), code segment, data segment, stack, etc. are all treated at run time as if they were merged together into a single segment. The OS loads the address of that segment into CS, DS and SS (I think ES as well, but I don't recall) at the start of execution. It is common and practical for a .com program to leave CS, DS, and SS unchanged through the entire execution.

So DX should be the offset of the string from the beginning of the PSP, not from the beginning of the data segment. There are multiple ways to accomplish that in NASM, depending on details of the full build process from source code to .com file. Using a group directive (similar to what you might do in MASM) is the most general method, suitable for situations in which you assemble to .obj files and then use a linker to create the .com or use a linker to create a .exe and another tool to convert to .com.

Using just the correct org directive without a group directive can handle some of the simpler situations and/or build sequences.

"It's all com - ing back to me now ..."

(click here if you can stand to here Céline Dione actually singing ... I'll pass. )