# Toy lisp interpreter. Incomplete.
#
# To run:
# $ ./translate_subx init.linux 0*.subx apps/mulisp.subx
# $ ./a.elf
# 42
# => 42
# ^D
# $
== code
Entry: # run tests if necessary, a REPL if not
# . prologue
89/<- %ebp 4/r32/esp
# initialize heap
(new-segment *Heap-size Heap)
{
# if (argc <= 1) break
81 7/subop/compare *ebp 1/imm32
7e/jump-if-lesser-or-equal break/disp8
# if (argv[1] != "test")) break
(kernel-string-equal? *(ebp+8) "test") # => eax
3d/compare-eax-and 0/imm32
74/jump-if-equal break/disp8
#
(run-tests)
# syscall(exit, *Num-test-failures)
8b/-> *Num-test-failures 3/r32/ebx
eb/jump $main:end/disp8
}
(repl Stdin Stdout)
# syscall(exit, 0)
bb/copy-to-ebx 0/imm32
$main:end:
b8/copy-to-eax 1/imm32/exit
cd/syscall 0x80/imm8
# Data structures
#
# Lisp is dynamically typed. Values always carry around knowledge of their
# type.
#
# There's several types of types in the description below, so we need a
# glossary and notational convention to disambiguate:
# lisp type: what Lisp code can see. Looks how you type it at the prompt.
# nil num char string symbol pair array
# type tag: the numeric code for a lisp type. All caps.
# NIL NUM CHAR STRING SYMBOL PAIR ARRAY
# memory type: a type specifying memory layout at the SubX level. Starts
# with a '$'.
# $int $array $(addr _)
#
# Lisp values are represented in memory by the _cell_ data structure. A cell
# is 12 bytes long:
# tag: $int (4 bytes; we're not concerned about wasting space)
# data: 8 bytes whose contents and meaning depend on tag
#
# What values of the different Lisp types look like in memory:
# - nil: cell{ tag: 0/NIL, data: 0 0 }
# - num: cell{ tag: 1/NUM, data: $int 0 }
# data contains the number
# - char: cell{ tag: 2/CHAR, data: $int 0 }
# data contains the utf-8 code of the character (no compound glyphs, no
# modifiers, etc., etc.)
# - string: cell{ tag: 3/STRING, data: $(addr stream byte)
# data contains an (addr array byte) containing the string in utf-8
# - symbol: cell{ tag: 4/SYMBOL, data: $(addr array byte) 0 }
# data contains an (addr array byte) containing the name of the symbol in utf-8
# alternatively, data could contain an index into the table of interned symbols
# - pair: cell{ tag: 5/PAIR, data: $(addr cell) $(addr cell) }
# data contains pointers to car and cdr
# - array: cell{ tag: 6/ARRAY, data: $tag $(addr stream data)
# data contains a pointer to an array of 8-byte data fields and the common
# tag for them all
repl: # in : (addr buffered-file), out : (addr buffered-file)
# . prologue
55/push-ebp
89/<- %ebp 4/r32/esp
# . save registers
50/push-eax
{
(lisp-read Stdin) # => eax : (handle cell)
# if (eax == 0) break
3d/compare-eax-and 0/imm32
74/jump-if-equal break/disp8
#
(lisp-eval %eax) # => eax : (handle cell)
(lisp-print Stdout %eax)
eb/jump loop/disp8
}
$repl:end:
# . restore registers
58/pop-to-eax
# . epilogue
89/<- %esp 5/r32/ebp
5d/pop-to-ebp
c3/return
# numbers start with a digit and are always in hex
# characters start with a backslash
# pairs start with '('
# arrays start with '['
# symbols start with anything else but quote, backquote, unquote or splice
# only one s-expression per line
lisp-read: # in : (addr buffered-file) -> eax : (handle cell)
# . prologue
55/push-ebp
89/<- %ebp 4/r32/esp
# . save registers
51/push-ecx
# var s/ecx : (stream byte 512)
81 5/subop/subtract %esp 0x200/imm32
68/push 0x200/imm32/size
68/push 0/imm32/read
68/push 0/imm32/write
89/<- %ecx 4/r32/esp
{
# read line into s
(clear-stream %ecx)
(read-line-buffered *(ebp+8) %ecx)
# if (s->write == 0) return null
{
81 7/subop/compare *ecx 0/imm32
75/jump-if-not-equal break/disp8
b8/copy-to-eax 0/imm32/eof
eb/jump $lisp-read:end/disp8
}
# ...
#? eb/jump loop/disp8
}
# return s
89/<- %eax 1/r32/ecx
$lisp-read:end:
# . reclaim locals
81 0/subop/add %esp 0x20c/imm32
# . restore registers
59/pop-to-ecx
# . epilogue
89/<- %esp 5/r32/ebp
5d/pop-to-ebp
c3/return
# lisp-read: in : (addr buffered-file) -> (handle cell)
# token tmp = next-mulisp-token(in)
# if is-int(tmp) return cell(tmp)
# if is-string(tmp) return cell(tmp)
# if is-pair(tmp) ...
# if is-array(tmp) ...
next-mulisp-token: # in : (addr buffered-file), line : (addr stream byte), result : (addr slice)
# pseudocode:
# if (line->read >= line->write)
# read-line-buffered(in, line)
# recurse
# if (line->data[line->read] == ' ')
# skip-chars-matching-whitespace(line)
# recurse
# if (line->data[line->read] == '#')
# read-line-buffered(in, line)
# recurse
# eax = line->data[line->read]
# if (eax == '"')
# result->start = &line->data[line->read]
# skip-string(in)
# result->end = &line->data[line->read]
# return
# if (is-digit(eax))
# result->start = &line->data[line->read]
# skip-hex-int(in)
# result->end = &line->data[line->read]
# return
# if (eax in '(' ')' '[' ']')
# result->start = &line->data[line->read]
# ++line->read
# result->en = &line->data[line->read]
# return
# else
# result->start = &line->data[line->read]
# skip-lisp-word(line)
# result->en = &line->data[line->read]
# return
#
# . prologue
55/push-ebp
89/<- %ebp 4/r32/esp
# . save registers
$next-mulisp-token:end:
# . reclaim locals
# . restore registers
# . epilogue
89/<- %esp 5/r32/ebp
5d/pop-to-ebp
c3/return
new-int-cell: # in : (addr slice) -> eax : (handle cell)
new-string-cell: # in : (addr slice) -> eax : (handle cell)
lisp-eval: # in : (addr cell) -> eax : (handle cell)
# . prologue
55/push-ebp
89/<- %ebp 4/r32/esp
# . save registers
8b/-> *(ebp+8) 0/r32/eax
$lisp-eval:end:
# . restore registers
# . epilogue
89/<- %esp 5/r32/ebp
5d/pop-to-ebp
c3/return
lisp-print: # out : (addr buffered-file), x : (addr cell)
# . prologue
55/push-ebp
89/<- %ebp 4/r32/esp
# . save registers
# write(x)
(write-buffered Stdout "=> ")
(write-stream-data Stdout *(ebp+0xc))
(flush Stdout)
$lisp-print:end:
# . restore registers
# . epilogue
89/<- %esp 5/r32/ebp
5d/pop-to-ebp
c3/return
== data
Nil:
0/imm32/tag
0/imm32/data