declaration of constant strings in x86 assembly language

Question

Today, while declaring constant strings (using Visual Studio community 2015 in win 10 ) I was facing a problem. Book said,

String BYTE "HELLO",0

while I typed in the same, MASM throws errors like:

syntax error : ,

Then, I removed , and 0 then it showed:

Missing operator in expression

How can I eliminate this error? What is the right way to declare a constant string?

Solution

The "standard" MASM syntax for declaring a constant string is:

VarName DB "String Contents",0

This declares an array of bytes (db == declare byte) with the symbolic name VarName. The contents of that array are given by the value enclosed in quotation marks ("String Contents"), and the array is terminated by a NUL character (0). Obviously, what this gives you is a standard C-style NUL-terminated string that can be used with almost any C API.

I call this the "standard" syntax for MASM because it's the one I use, the one that most compilers/disassemblers use, and the one that you'll see most code samples written in.

However, as Ped7g points out in a comment, modern versions of MASM actually support using the BYTE directive in the declaration. It is effectively a synonym for DB. That would make the book author's code correct, and suitable for use with the version of MASM bundled with any modern version of Visual Studio:

VarName BYTE "String Contents",0

You can use either one you want. If you're learning from a book that uses the latter, you may want to stick with it for convenience. However, you should be aware of the former, too, since as I mentioned, you'll see it all over the place.

---

Here is the complete list of type specifiers (keywords) that you'll see in data definitions:

• DB(Declare Byte), which is equivalent to BYTE
  (an 8-bit value the same size as the AL register)
• DW (Declare Word), which is equivalent to WORD
  (a 16-bit value the same size as the AX register)
• DD (Declare DoubleWord), which is equivalent to DWORD
  (a 32-bit value the same size as the EAX register)
• DF (Declare Float), which is equivalent to FWORD
  (a 48-bit value used to store a single-precision floating-point value)
• DQ (Declare QuadWord), which is equivalent to QWORD
  (a 64-bit value the same size as the RAX register when targeting 64-bit mode x86-64, and also used to store a double-precision floating-point value)
• DT (Declare Ten-byte), which is equivalent to TBYTE
  (an 80-bit value the same size as the x87 FPU's stack-based registers; used to spill a value from there directly to memory without losing precision)

Comments

[Cody Gray]

What book are you reading? Perhaps it is not teaching you MASM syntax? BYTE is illegal here (although it is a keyword used other places). This should be String db "HELLO",0 and probably located in a .data segment.

[Vasanth]

@Cody Gray I am reading x86 assembly programming by kip Irvine. Your notation is given under legacy data directives and as equally correct.It is indeed in the .data segment. Thank you.

[Vasanth]

My problem fixed but rather weirdly.. after updating my PC and rebooting.. it is now accepting both the notations. But what is the standard notation supported by most assemblers? THANK YOU.

[zx485]

There are no 'most assemblers'. Everyone has its own syntax. For MASM, the line should be String db "HELLO",0.

[Ped7g]

a modern version of MASM does support also BYTE directive (as replacement for old good DB), and indeed DB is now "obsolete"... I expect to have plenty of fun with MASM questions on SO in next decade, as if ambiguity between 16/32/64b mode and MASM/TASM vs NASM/FASM/YASM vs gas/AT&T wasn't already enough... Well played MS, well played. :D

[Ross Ridge]

I can't reproduce the problem. The version of MASM included with Visual Studio 2015 assembles String BYTE "HELLO",0 without errors.

[Peter Cordes]

TEXTEQU is an assemble-time directive, like #define in C. It doesn't assemble any bytes into the output file.

[Peter Cordes]

Do you have a citation for DB standing for Declare Byte? I think of it as standing for "Data Byte"; its purpose is to assemble bytes directly into the output file.

[Cody Gray]

No citation on hand, @Peter. Just consistency, the hobgoblin of my simple mind.

[Peter Cordes]

Are the < and > around the data actually needed for EQU or TEXTEQU to work? I don't think that looks like normal MASM syntax.

[John Rudolph]

yes, but usually strings are declared in the .data segment. When the < > are used, the constants are declared in the global scope