.\" Contact d_m@plastic-idolatry.com to correct errors or typos.
.THuxntal7"05 Aug 2024""1.0""Uxntal Reference Guide"
.SHNAME
uxntal \- assembly langauge for Varvara virtual machine
.SHDESCRIPTION
Uxntal is an 8-bit instruction set for programming the Varvara virtual machine.
It uses the lower 5-bits to specify an opcode, and the upper 3-bits to specify
optional modes.
ROMs consist of a 16-bit address space of bytes. Any byte can be interpreted as either data or an instruction. A 2-byte program counter (\fIpc\fP) determines the address of the next instruction to decode and run.
Instructions manipulate data using two stacks: a working stack (\fBwst\fP) and a return stack (\fBrst\fP). Each stack consists of 256 bytes, and in the case of overflow or underflow the stack pointer will wrap (the stacks are circular).
There are also 256 bytes of device memory, which are used to interact with the virtual machine and its devices.
Instructions deal with unsigned 8-bit values (\fIbytes\fP) and unsigned 16-bit values (\fIshorts\fP). There are no other built-in data types.
The "complete" opcode's value can be derived by combining the base value with its flags.
For example, \fBADD2k\fP is \fB(ADD | 2 | k)\fP = \fB(0x18 | 0x20 | 0x80)\fP = \fB0xb8\fP.
Unlike other opcodes, \fB0x00\fP (\fBBRK*\fP) is contextual: its meaning depends on the \fImode\fP bits provided:
0x00 \fBBRK\fP 0x80 \fBLIT\fP
0x20 \fBJCI\fP 0xa0 \fBLIT2\fP
0x40 \fBJMI\fP 0xc0 \fBLITr\fP
0x60 \fBJSI\fP 0xe0 \fBLIT2r\fP
.SHSTACKEFFECTS
.BR
.SSNOTATION
Given a stack effect \fB( a^ b^ c^ -- c^ a^ b^ )\fP here is what each symbol means:
\fB(\fP and \fB)\fP are comment delimiters
\fBa^\fP, \fBb^\fP, and \fBc^\fP are values on the stack
\fB^\fP indicates that each value is a \fIbyte\fP (\fB*\fP would indicate \fIshort\fP)
\fB--\fP separates the "before" and "after" of the stack effect
The effect here is to move the top byte of the stack below the next two bytes, which could be achieved with \fBROT ROT\fP.
By default stack effects describe the effect on \fBwst\fP. When \fBrst\fP is involved we use \fB[]\fP to differentiate the stacks. For example \fB( a* [b*] -- a+1* [b+1*] )\fP will increment the top short of both \fBwst\fP and \fBrst\fP.
.SSEFFECTSANDMODES
Regular instructions have a single stack effect which is modified in a predictable way by any additional modes.
For example the generic effect for \fBADD\fP is ( x y -- x+y ). The eight combinations of modes have the following effects:
Thus for regular instructions writing a "generic" effect (leaving sigils off values whose size depends on \fIshort mode\fP) is sufficient to describe its behavior across all eight variations. Note that some instructions always read values of a fixed size. For example the boolean condition read by \fBJCN\fP is always one byte, no matter what modes are used.
In \fIreturn mode\fP the stacks are reversed. Effects on \fBwst\fP will instead affect \fBrst\fP, and effects on \fBrst\fP will instead affect \fBwst\fP. For example, \fBSTH\fP reads a byte from \fBwst\fP and writes it to \fBrst\fP, but \fBSTHr\fP reads a byte from \fBrst\fP and writes it to \fBwst\fP.
In \fIkeep mode\fP all the values on the left-hand side of the stack effect will also appear on the right-hand side before the outputs. For example, \fBSWP\fP is \fB(x y -- y x)\fP but \fBSWPk\fP is \fB(x y -- x y y x)\fP.
.SSTERMINOLOGY
We consider the top of the stack to be the first value of the stack, and count back from there. For example, given the stack effect \fB( a b c -- )\fP we would say that \fBc\fP is the top of the stack, \fBb\fP is the second value (second from the top), and \fBa\fP is the third value (third from the top).
The program counter (\fIpc\fP) is unconditionally updated. When \fIx\fP is a byte, it is treated as relative (\fBpc += x\fP) and when \fIx\fP is a short it is treated as absolute (\fBpc = x\fP).
It is common to \fBJMP\fP with boolean bytes (0-1) to handle simple conditionals. For example:
The program counter (\fIpc\fP) is updated when \fIbool\fP is non-zero. When \fIx\fP is a byte, it is treated as relative (\fBpc += x\fP) and when \fIx\fP is a short it is treated as absolute (\fBpc = x\fP).
Jump to a location, saving a reference to return to.
Stores the next address to execute before unconditionally updating the program counter (\fIpc\fP). This instruction is usually used to invoke subroutines, which use the \fBJMP2r\fP to return. When \fIx\fP is a byte, it is treated as relative (\fBpc += x\fP) and when \fIx\fP is a short it is treated as absolute (\fBpc = x\fP).
The saved address will always be a short regardless of \fIshort mode\fP.
Note that unlike \fBLDZk\fP and \fBLDAk\fP the \fBLDRk\fP instruction is not very useful, since a relative address is usually only meaningful when run from a particular address (i.e. for a particular \fIpc\fP value).
Note that unlike \fBSTZk\fP and \fBSTAk\fP the \fBSTRk\fP instruction is not very useful, since a relative address is usually only meaningful when run from a particular address (i.e. for a particular \fIpc\fP value).
Reading from some ports may have an effect on the underlying VM; in other cases it will simply read values from device memory. See Varvara device documentation for more details.
Writing to some ports may have an effect on the underlying VM; in other cases it will simply write values to device memory. See Varvara device documentation for more details.
Divide the second value of the stack by the top of the stack.
\fBDIV\fP implements \fIEuclidean division\fP, which is also known as \fIinteger division\fP. It returns whole numbers, so \fB#08 #09 DIV\fP evaluates to \fB0x00\fP.
Division by zero will return zero (instead of signaling an error).
Unlike \fBADD\fP, \fBSUB\fP, and \fBMUL\fP, \fBDIV\fP does not behave correctly for numbers which should be treated as signed. For example, the signed byte representation of \fB-2\fP is \fB0xfe\fP, but \fB#06 #fe DIV\fP evaluates to \fB0x00\fP (\fB6 / 254 = 0\fP). For signed values the correct result should instead be \fB0xfd\fP (\fB6 / -2 = -3\fP).
There is no \fIremainder\fP instruction, but the phrase \fBDIVk MUL SUB\fP can be used to compute the remainder.
Compute a bit shift of the second value of the stack; the directions and distances are determined by the top value of the stack.
Given a byte \fIrl\fP consisting of a low nibble (\fIl\fP) and a high nibble (\fIr\fP), this instruction shifts \fIx\fP left by \fIl\fP and then right by \fIr\fP.
Right shifts are unsigned (they introduce zero bits). There are no signed shifts.
For 16-bit (and 8-bit) values, one nibble (\fB0x0 - 0xf\fP) is sufficient to express all useful left or right shifts.
Right: \fB#ff #03 SFT\fP evaluates to \fB0x1f\fP
Left: \fB#ff #20 SFT\fP evaluates to \fB0xfc\fP
Both: \fB#ff #23 SFT\fP evaluates to \fB0x7c\fP
.SHSPECIALINSTRUCTIONS
These instructions do not accept all mode flags (some do not accept any).
.SSBRK
The break instruction is used to end a vector call and return control to the virtual machine.
.SSJCI,JMI,andJSI
The "immediate jump" instructions are produced by the assembler. They interpret the next 2 bytes of the ROM as a relative address (\fIaddr\fP) and have the following effects:
\fBJMI\fP ( -- ) jump to \fIaddr\fP unconditionally
\fBJCI\fP ( bool^ -- ) jump to \fIaddr\fP if \fIbool\fP is non-zero
(The instruction pointer will be moved forward 2 bytes, past the relative address.)
These instructions are created by the assembler from special syntax:
\fB!dest\fP produces \fBJMI wx yz\fP
\fB?dest\fP produces \fBJCI wx yz\fP
\fBdest\fP produces \fBJSI wx yz\fP (assuming \fBdest\fP is not a macro or reserved)
.SSLIT,LIT2,LITr,andLIT2r
Push a literal value on the stack.
The "literal" instructions are used to push new data onto the stacks. They interpret the next 1-2 bytes of the ROM (\fIwx\fP, \fIwxyz\fP) as data and push it onto the corresponding stack:
