The SunOS 5.x SPARC assembler takes assembly language programs, as specified in this document, and produces relocatable object files for processing by the SunOS 5.x SPARC link editor. The assembly language described in this document corresponds to the SPARC instruction set defined in the SPARC Architecture Manual (Version 8 and Version 9) and is intended for use on machines that use the SPARC architecture.
This chapter is organized into the following sections:
In the descriptions of assembly language syntax in this chapter:
Brackets ([ ]) enclose optional items.
Asterisks (*) indicate items to be repeated zero or more times.
Braces ({ }) enclose alternate item choices, which are separated from each other by vertical bars (|).
Wherever blanks are allowed, arbitrary numbers of blanks and horizontal tabs may be used. Newline characters are not allowed in place of blanks.
The syntax of assembly language files is:
[line]*
The syntax of assembly language lines is:
[statement [ ; statement]*] [!comment]
The syntax of an assembly language statement is:
[label:] [instruction]
where:
label
is a symbol name.
instruction
is an encoded pseudo-op, synthetic instruction, or instruction.
This section describes the lexical features of the assembler syntax.
Uppercase and lowercase letters are distinct everywhere except in the names of special symbols. Special symbol names have no case distinction.
A comment is preceded by an exclamation mark character (!); the exclamation mark character and all following characters up to the end of the line are ignored. C language-style comments (``/*…*/'') are also permitted and may span multiple lines.
A label is either a symbol or a single decimal digit n (0…9). A label is immediately followed by a colon ( : ).
Numeric labels may be defined repeatedly in an assembly file; normal symbolic labels may be defined only once.
A numeric label n is referenced after its definition (backward reference) as nb, and before its definition (forward reference) as nf.
Decimal, hexadecimal, and octal numeric constants are recognized and are written as in the C language. However, integer suffixes (such as L) are not recognized.
For floating-point pseudo-operations, floating-point constants are written with 0r or 0R (where r or R means REAL) followed by a string acceptable to atof(3); that is, an optional sign followed by a non-empty string of digits with optional decimal point and optional exponent.
The special names 0rnan and 0rinf represent the special floating-point values Not-A-Number (NaN) and INFinity. Negative Not-A-Number and Negative INFinity are specified as 0r-nan and 0r-inf.
The names of these floating-point constants begin with the digit zero, not the letter “O.”
A string is a sequence of characters quoted with either double-quote mark (") or single-quote mark (') characters. The sequence must not include a newline character. When used in an expression, the numeric value of a string is the numeric value of the ASCII representation of its first character.
The suggested style is to use single quote mark characters for the ASCII value of a single character, and double quote mark characters for quoted-string operands such as used by pseudo-ops. An example of assembly code in the suggested style is:
add %g1,'a'-'A',%g1 ! g1 + ('a' - 'A') --> g1
The escape codes described in Table 2–1, derived from ANSI C, are recognized in strings.
Table 2–1
Escape Code |
Description |
---|---|
\a |
Alert |
\b |
Backspace |
\f |
Form feed |
\n |
Newline (line feed) |
\r |
Carriage return |
\t |
Horizontal tab |
\v |
Vertical tab |
\nnn |
Octal value nnn |
\xnn... |
Hexadecimal value nn... |
The syntax for a symbol name is:
{ letter | _ | $ | . } { letter | _ | $ | . | digit }*
In the above syntax:
Uppercase and lowercase letters are distinct; the underscore ( _ ), dollar sign ($), and dot ( . ) are treated as alphabetic characters.
Symbol names that begin with a dot ( . ) are assumed to be local symbols. To simplify debugging, avoid using this type of symbol name in hand-coded assembly language routines.
The symbol dot ( . ) is predefined and always refers to the address of the beginning of the current assembly language statement.
External variable names beginning with the underscore character are reserved by the ANSI C Standard. Do not begin these names with the underscore; otherwise, the program will not conform to ANSI C and unpredictable behavior may result.
Special symbol names begin with a percentage sign (%) to avoid conflict with user symbols. Table 2–2 lists these special symbol names.
Table 2–2
Symbol Object |
Name |
Comment |
---|---|---|
General-purpose registers |
%r0 … %r31 |
|
General-purpose global registers |
%g0 … %g7 |
Same as %r0 … %r7 |
General-purpose out registers |
%o0 … %o7 |
Same as %r8 … %r15 |
General-purpose local registers |
%l0 … %l7 |
Same as %r16 … %r23 |
General-purpose in registers |
%i0 … %i7 |
Same as %r24 … %r31 |
Stack-pointer register |
%sp |
(%sp = %o6 = %r14) |
Frame-pointer register |
%fp |
(%fp = %i6 = %r30) |
Floating-point registers |
%f0 … %f31 |
|
Floating-point status register |
%fsr |
|
Front of floating-point queue |
%fq |
|
Coprocessor registers |
%c0 … %c31 |
|
Coprocessor status register |
%csr |
|
Coprocessor queue |
%cq |
|
Program status register |
%psr |
|
Trap vector base address register |
%tbr |
|
Window invalid mask |
%wim |
|
Y register |
%y |
|
Unary operators |
%lo |
Extracts least significant 10 bits |
|
%hi |
Extracts most significant 22 bits |
|
%r_disp32 |
Used only in Sun compiler-generated code. |
|
%r_plt32 |
Used only in Sun compiler-generated code. |
Ancillary state registers |
%asr1 … %asr31 |
|
There is no case distinction in special symbols; for example,
%PSR
is equivalent to
%psr
The suggested style is to use lowercase letters.
The lack of case distinction allows for the use of non-recursive preprocessor substitutions, for example:
#define psr %PSR
The special symbols %hi and %lo are true unary operators which can be used in any expression and, as other unary operators, have higher precedence than binary operations. For example:
%hi a+b = (%hi a)+b %lo a+b = (%lo a)+b
To avoid ambiguity, enclose operands of the %hi or %lo operators in parentheses. For example:
%hi(a) + b
The operators described in Table 2–3 are recognized in constant expressions.
Table 2–3
Binary |
Operators |
Unary |
Operators |
---|---|---|---|
+ |
Integer addition |
+ |
(No effect) |
– |
Integer subtraction |
– |
2's Complement |
* |
Integer multiplication |
~ |
1's Complement |
/ |
Integer division |
%lo(address) |
Extract least significant 10 bits as computed by: (address & 0x3ff) |
% |
Modulo |
%hi(address) |
Extract most significant 22 bits as computed by: (address >>10) |
^ |
Exclusive OR |
%r_disp32 %r_disp64 |
Used in Sun compiler-generated code only to instruct the assembler to generate specific relocation information for the given expression. |
<< |
Left shift |
%r_plt32 %r_plt64 |
Used in Sun compiler-generated code only to instruct the assembler to generate specific relocation information for the given expression. |
>> |
Right shift |
|
|
& |
Bitwise AND |
|
|
| |
Bitwise OR |
|
|
Since these operators have the same precedence as in the C language, put expressions in parentheses to avoid ambiguity.
To avoid confusion with register names or with the %hi, %lo, %r_disp32/64, or %r_plt32/64 operators, the modulo operator % must not be immediately followed by a letter or digit. The modulo operator is typically followed by a space or left parenthesis character.
The following V9 64-bit operators and expressions in Table 2–4 ease the task of converting from V8/V8plus assembly code to V9 assembly code.
Table 2–4
Unary |
Calculation |
Operators |
---|---|---|
%hh |
(address) >> 42 |
Extract bits 42-63 of a 64-bit word |
%hm |
((address) >> 32) & 0x3ff |
Extract bits 32-41 of a 64-bit word |
%lm |
(((address) >> 10) & 0x3fffff) |
Extract bits 10-31 of a 64-bit word |
For example:::
sethi %hh (address), %l1 or %l1, %hm (address), %l1
sethi %lm (address), %12 or %12, %lo (address), %12
sllx %l1, 32, %l1 or %l1, %12, %l1
The V9 high 32-bit operators and expressions are identified in Table 2–5.
Table 2–5
Unary |
Calculation |
Operators |
---|---|---|
%hix |
((((address) ^ 0xffffffffffffffff >> 10) &0x4fffff) |
Invert every bit and extract bits 10-31 |
%lox |
((address) & 0x3ff | 0x1c00 |
Extract bits 0-9 and sign extend that to 13 bits |
For example:
%sethi %hix (address), %l1 or %l1, %lox (address), %l1
The V9 low 44-bit operators and expressions are identified in Table 2–6.
Table 2–6
Unary |
Calculation |
Operators |
---|---|---|
%h44 |
((address) >> 22) |
Extract bits 22-43 of a 64-bit word |
%m44 |
((address) >> 12) & 0x3ff |
Extract bits 12-21 of a 64-bit word |
l44 |
(address) & 0xfff |
Extract bits 0-11 of a 64-bit word |
For example::
%sethi %h44 (address), %l1 or %l1, %m44 (address), %l1 sllx %l1, 12, %l1 or %l1, %144 (address), %l1
Messages generated by the assembler are generally self-explanatory and give sufficient information to allow correction of a problem.
Certain conditions will cause the assembler to issue warnings associated with delay slots following Control Transfer Instructions (CTI). These warnings are:
Set synthetic instructions in delay slots
Labels in delay slots
Segments that end in control transfer instructions
These warnings point to places where a problem could exist. If you have intentionally written code this way, you can insert an .empty pseudo-operation immediately after the control transfer instruction.
The .empty pseudo-operation in a delay slot tells the assembler that the delay slot can be empty or can contain whatever follows because you have verified that either the code is correct or the content of the delay slot does not matter.