X-Git-Url: https://git.rrq.au/?a=blobdiff_plain;f=main.asm;h=5985b997af8095926394802fd7e6c59b636a5510;hb=bcba0c88f4eaddac8e0d8c0fa68264b2e81ee272;hp=edcba0ae9421c712218041e64d4fc0872370f8d9;hpb=2d2c007e5b5e2f9b79271d638863d77338969f38;p=rrq%2Fjonasforth.git

diff --git a/main.asm b/main.asm
index edcba0a..5985b99 100644
--- a/main.asm
+++ b/main.asm
@@ -1,4 +1,75 @@
-format ELF64 executable
+;; vim: syntax=fasm
+
+;; At compile-time we load the module given by the environment variable
+;; OS_INCLUDE. This module should define the following macros:
+;;
+;; Each of these functions should preserve the value of RSI and RSP. They may
+;; use other registers as they like.
+;;
+;; os_initialize
+;;   Called at initialization.
+;;
+;; os_print_string
+;;   Takes a string buffer in RCX and the length in RDX, and prints the string
+;;   to the console.
+;;
+;; os_read_char
+;;   Wait for the user to type a key, and then put the corresponding ASCII byte
+;;   into the buffer pointed to by RCX.
+;;
+;; os_terminate
+;;   Shut down the system.
+include '%OS_INCLUDE%'
+
+;; Print a string of a given length.
+;;
+;; Input:
+;; - RCX = Pointer to buffer
+;; - RDX = Buffer length
+;;
+;; Clobbers: RAX, RCX, R11, RDI, RSI
+macro sys_print_string {
+  push r8
+  push r9
+  push r10
+
+  call os_print_string
+
+  pop r10
+  pop r9
+  pop r8
+}
+
+;; Read a character from the user into the given buffer.
+;;
+;; Input:
+;; - RSI = Character buffer
+;;
+;; Output:
+;; - BYTE [RSI] = Character
+;;
+;; Clobbers: RAX, RCX, R11, RDI, RSI, RDX
+macro sys_read_char {
+  push rbx
+  push r8
+  push r9
+  push r10
+  push r15
+
+  mov rcx, rsi
+  call os_read_char
+
+  pop r15
+  pop r10
+  pop r9
+  pop r8
+  pop rbx
+}
+
+macro sys_terminate code {
+  mov rax, code
+  call os_terminate
+}
 
 ;; The code in this macro is placed at the end of each Forth word. When we are
 ;; executing a definition, this code is what causes execution to resume at the
@@ -32,11 +103,16 @@ macro popr x {
 ;;
 ;; This macro also defines a label LABEL_entry.
 initial_latest_entry = 0
-macro header label, name {
+macro header label, name, immediate {
   local .string_end
 
 label#_entry:
   dq initial_latest_entry
+  if immediate eq
+    db 0
+  else
+    db 1
+  end if
   db .string_end - ($ + 1)
   db name
   .string_end:
@@ -46,31 +122,23 @@ initial_latest_entry = label#_entry
 }
 
 ;; Define a Forth word that is implemented in assembly. See 'header' for details.
-macro forth_asm label, name {
-  header label, name
+macro forth_asm label, name, immediate {
+  header label, name, immediate
   dq .start
 .start:
 }
 
-;; Define a Forth word that is implemented in Forth. (The body will be a list of
-;; 'dq' statements.)
-macro forth label, name {
-  header label, name
-  dq docol
-}
-
-
+section '.text' code readable executable
 
-segment readable executable
-
-entry main
-
-include "impl.asm"
+include "impl.asm"      ; Misc. subroutines
+include "bootstrap.asm" ; Forth words encoded in Assembly
 
 main:
   cld                        ; Clear direction flag so LODSQ does the right thing.
   mov rbp, return_stack_top  ; Initialize return stack
 
+  call os_initialize
+
   mov rax, MAIN
   jmp qword [rax]
 
@@ -79,13 +147,13 @@ program: dq MAIN
 ;; The codeword is the code that will be executed at the beginning of a forth
 ;; word. It needs to save the old RSI and update it to point to the next word to
 ;; execute.
-docol:
+header DOCOL, 'DOCOL'
   pushr rsi            ; Save old value of RSI on return stack; we will continue execution there after we are done executing this word
   lea rsi, [rax + 8]   ; RAX currently points to the address of the codeword, so we want to continue at RAX+8
   next                 ; Execute word pointed to by RSI
 
 ;; This word is called at the end of a Forth definition. It just needs to
-;; restore the old value of RSI (saved by 'docol') and resume execution.
+;; restore the old value of RSI (saved by 'DOCOL') and resume execution.
 forth_asm EXIT, 'EXIT'
   popr rsi
   next
@@ -97,6 +165,17 @@ forth_asm LIT, 'LIT'
   push rax
   next
 
+;; When LITSTRING is encountered while executing a word, it instead reads a
+;; string from the definition of that word, and places that string on the stack
+;; as (buffer, length).
+forth_asm LITSTRING, 'LITSTRING'
+  lodsb
+  push rsi ; Buffer
+  movzx rax, al
+  push rax ; Length
+  add rsi, rax ; Skip over string before resuming execution
+  next
+
 ;; Given a string (a pointer following by a size), return the location of the
 ;; dictionary entry for that word. If no such word exists, return 0.
 forth_asm FIND, 'FIND'
@@ -119,7 +198,7 @@ forth_asm FIND, 'FIND'
 ;; entry.
 forth_asm TCFA, '>CFA'
   pop rax
-  add rax, 8                    ; [rax] = length of name
+  add rax, 8 + 1                ; [rax] = length of name
   movzx rbx, byte [rax]
   inc rax
   add rax, rbx                  ; [rax] = codeword
@@ -158,45 +237,94 @@ forth_asm EXEC, 'EXEC'
 forth_asm EMIT, 'EMIT'
   pushr rsi
   pushr rax
-  mov rax, 1
-  mov rdi, 1
-  lea rsi, [rsp]
+
+  lea rcx, [rsp]
   mov rdx, 1
-  syscall
+  sys_print_string
+
   add rsp, 8
   popr rax
   popr rsi
   next
 
-;; Prints a newline to standard output.
-forth NEWLINE, 'NEWLINE'
-  dq LIT, $A
-  dq EMIT
-  dq EXIT
+;; Read a single character from the current input stream. Usually, this will wait
+;; for the user to press a key, and then return the corresponding character. When
+;; reading from a special buffer, it will instead return the next characater from
+;; that buffer.
+;;
+;; The ASCII character code is placed on the stack.
+forth_asm KEY, 'KEY'
+  call .impl
+  push rax
+  next
 
-;; Prints a space to standard output.
-forth SPACE, 'SPACE'
-  dq LIT, ' '
-  dq EMIT
-  dq EXIT
+;; Result in RAX
+.impl:
+  ;; Are we reading from user input or from the input buffer?
+  cmp [input_buffer], 0
+  jne .from_buffer
+
+  ;; Reading user input
+  push rsi
+  mov rsi, .buffer
+  sys_read_char
+  pop rsi
+
+  movzx rax, byte [.buffer]
+  ret
 
-;; Read a word from standard input and push it onto the stack as a pointer and a
-;; size. The pointer is valid until the next call to READ_WORD.
+.from_buffer:
+  ;; Reading from buffer
+  mov rax, [input_buffer]
+  movzx rax, byte [rax]
+
+  inc [input_buffer]
+  dec [input_buffer_length]
+  ret
+
+;; Read a word and push it onto the stack as a pointer and a size. The pointer
+;; is valid until the next call to READ_WORD.
 forth_asm READ_WORD, 'READ-WORD'
-  mov [.rsi], rsi
+  push rsi
+.skip_whitespace:
+  ;; Read characters until one of them is not whitespace.
+  call KEY.impl
+  ;; We consider newlines and spaces to be whitespace.
+  cmp al, ' '
+  je .skip_whitespace
+  cmp al, $A
+  je .skip_whitespace
+
+  ;; We got a character that wasn't whitespace. Now read the actual word.
+  mov [.length], 0
 
-  call read_word
-  push rdi                      ; Buffer
-  push rdx                      ; Length
+.read_alpha:
+  movzx rbx, [.length]
+  mov rsi, .buffer
+  add rsi, rbx
+  mov [rsi], al
+  inc [.length]
+
+  call KEY.impl
+
+  cmp al, ' '
+  je .end
+  cmp al, $A
+  jne .read_alpha
+
+.end:
+  pop rsi
+  push .buffer
+  movzx rax, [.length]
+  push rax
 
-  mov rsi, [.rsi]
   next
 
 ;; Takes a string on the stack and replaces it with the decimal number that the
 ;; string represents.
 forth_asm PARSE_NUMBER, 'PARSE-NUMBER'
-  pop [parse_number.length]     ; Length
-  pop [parse_number.buffer]     ; String pointer
+  pop rcx     ; Length
+  pop rdi     ; String pointer
 
   push rsi
   call parse_number
@@ -205,51 +333,57 @@ forth_asm PARSE_NUMBER, 'PARSE-NUMBER'
   push rax                      ; Result
   next
 
-forth READ_NUMBER, 'READ-NUMBER'
-  dq READ_WORD
-  dq PARSE_NUMBER
-  dq EXIT
-
 ;; Takes a string (in the form of a pointer and a length on the stack) and
 ;; prints it to standard output.
 forth_asm TELL, 'TELL'
-  mov rbx, rsi
-  mov rcx, rax
+  pushr rax
+  pushr rsi
 
-  mov rax, 1
-  mov rdi, 1
-  pop rdx     ; Length
-  pop rsi     ; Buffer
-  syscall
+  pop rdx ; Length
+  pop rcx ; Buffer
+  sys_print_string
 
-  mov rax, rcx
-  mov rsi, rbx
+  popr rsi
+  popr rax
   next
 
 ;; Exit the program cleanly.
 forth_asm TERMINATE, 'TERMINATE'
-  mov rax, $3C
-  mov rdi, 0
-  syscall
-
-forth HELLO, 'HELLO'
-  dq LIT, 'H', EMIT
-  dq LIT, 'e', EMIT
-  dq LIT, 'l', EMIT
-  dq LIT, 'l', EMIT
-  dq LIT, 'o', EMIT
-  dq LIT, '!', EMIT
-  dq NEWLINE
-  dq EXIT
+  sys_terminate 0
 
-;; The INTERPRET word reads and interprets user input. It's behavior depends on
-;; the current STATE. It provides special handling for integers. (TODO)
-forth INTERPRET, 'INTERPRET'
-  dq READ_WORD
-  dq FIND
-  dq TCFA
-  dq EXEC
-  dq EXIT
+;; Duplicate a pair of elements.
+forth_asm PAIRDUP, '2DUP'
+  pop rbx
+  pop rax
+  push rax
+  push rbx
+  push rax
+  push rbx
+  next
+
+;; Swap the top two elements on the stack.
+forth_asm SWAP, 'SWAP'
+  pop rax
+  pop rbx
+  push rax
+  push rbx
+  next
+
+;; Remove the top element from the stack.
+forth_asm DROP, 'DROP'
+  add rsp, 8
+  next
+
+forth_asm NOT_, 'NOT'
+  pop rax
+  cmp rax, 0
+  jz .false
+.true:
+  push 0
+  next
+.false:
+  push 1
+  next
 
 ;; .U prints the value on the stack as an unsigned integer in hexadecimal.
 forth_asm DOTU, '.U'
@@ -302,11 +436,9 @@ forth_asm DOTU, '.U'
   loop .flip
 
   ;; Print the buffer
-  mov rax, 1
-  mov rdi, 1
-  mov rsi, .buffer
+  mov rcx, .buffer
   mov rdx, [.printed_length]
-  syscall
+  sys_print_string
 
   ;; Restore RSI and continue execution
   pop rsi
@@ -326,37 +458,262 @@ forth_asm GET, '@'
   push rax
   next
 
-;; Get the location of the STATE variable. It can be set with '!' and read with
-;; '@'.
-forth STATE, 'STATE'
-  dq LIT, var_STATE
-  dq EXIT
+forth_asm PUT_BYTE, 'C!'
+  pop rbx
+  pop rax                       ; Value
+  mov [rbx], al
+  next
+
+forth_asm GET_BYTE, 'C@'
+  pop rax
+  movzx rax, byte [rax]
+  push rax
+  next
+
+;; Add two integers on the stack.
+forth_asm PLUS, '+'
+  pop rax
+  pop rbx
+  add rax, rbx
+  push rax
+  next
+
+;; Calculate difference between two integers on the stack. The second number is
+;; subtracted from the first.
+forth_asm MINUS, '-'
+  pop rax
+  pop rbx
+  sub rbx, rax
+  push rbx
+  next
+
+;; Given two integers a and b on the stack, pushes the quotient and remainder of
+;; division of a by b.
+forth_asm TIMESMOD, '/MOD'
+  pop rbx                       ; b
+  pop rax                       ; a
+  mov rdx, 0
+  div rbx
+  push rax                      ; a / b
+  push rdx                      ; a % b
+  next
+
+;; Read input until next " character is found. Push a string containing the
+;; input on the stack as (buffer length). Note that the buffer is only valid
+;; until the next call to S" and that no more than 255 characters can be read.
+forth_asm READ_STRING, 'S"'
+  ;; If the input buffer is set, we should read from there instead.
+  cmp [input_buffer], 0
+  jne read_string_buffer
+
+  push rsi
+
+  mov [.length], 0
+
+.read_char:
+  mov rsi, .char_buffer
+  sys_read_char
+
+  mov al, [.char_buffer]
+  cmp al, '"'
+  je .done
+
+  mov rdx, .buffer
+  add rdx, [.length]
+  mov [rdx], al
+  inc [.length]
+  jmp .read_char
+
+.done:
+  pop rsi
+
+  push .buffer
+  push [.length]
+
+  next
+
+read_string_buffer:
+  push rsi
+
+  ;; We borrow READ_STRING's buffer. They won't mind.
+  mov [READ_STRING.length], 0
+
+.read_char:
+  mov rbx, [input_buffer]
+  mov al, [rbx]
+  cmp al, '"'
+  je .done
+
+  mov rdx, READ_STRING.buffer
+  add rdx, [READ_STRING.length]
+  mov [rdx], al
+  inc [READ_STRING.length]
+
+  inc [input_buffer]
+  dec [input_buffer_length]
+
+  jmp .read_char
+
+.done:
+  pop rsi
+
+  ;; Skip closing "
+  inc [input_buffer]
+  dec [input_buffer_length]
+
+  push READ_STRING.buffer
+  push [READ_STRING.length]
+
+  next
+
+;; CREATE inserts a new header in the dictionary, and updates LATEST so that it
+;; points to the header. To compile a word, the user can then call ',' to
+;; continue to append data after the header.
+;;
+;; It takes the name of the word as a string (address length) on the stack.
+forth_asm CREATE, 'CREATE'
+  pop rcx                       ; Word string length
+  pop rdx                       ; Word string pointer
+
+  mov rdi, [here]               ; rdi = Address at which to insert this entry
+  mov rax, [latest_entry]       ; rax = Address of the previous entry
+  mov [rdi], rax                ; Insert link to previous entry
+  mov [latest_entry], rdi       ; Update LATEST to point to this word
+
+  add rdi, 8
+  mov [rdi], byte 0             ; Insert immediate flag
+
+  add rdi, 1
+  mov [rdi], byte cl            ; Insert length
+
+  ;; Insert word string
+  add rdi, 1
+
+  push rsi
+  mov rsi, rdx                  ; rsi = Word string pointer
+  rep movsb
+  pop rsi
+
+  ;; Update HERE
+  mov [here], rdi
+
+  next
+
+forth_asm TICK, "'"
+  lodsq
+  push rax
+  next
+
+forth_asm ROT, 'ROT'
+  pop rax
+  pop rbx
+  pop rdx
+  push rax
+  push rdx
+  push rbx
+  next
+
+forth_asm PICK, 'PICK'
+  pop rax
+  lea rax, [rsp + 8 * rax]
+  mov rax, [rax]
+  push rax
+  next
+
+forth_asm EQL, '='
+  pop rax
+  pop rbx
+  cmp rax, rbx
+  je .eq
+.noteq:
+  push 0
+  next
+.eq:
+  push 1
+  next
 
 forth MAIN, 'MAIN'
-  dq HELLO
+  dq SYSCODE
+  dq INTERPRET_STRING
   dq INTERPRET
   dq BRANCH, -8 * 2
   dq TERMINATE
 
-segment readable writable
+;; Built-in variables:
 
+forth STATE, 'STATE'
+  dq LIT, var_STATE
+  dq EXIT
+
+forth LATEST, 'LATEST'
+  dq LIT, latest_entry
+  dq EXIT
+
+forth HERE, 'HERE'
+  dq LIT, here
+  dq EXIT
+
+forth SYSCODE, 'SYSCODE'
+  dq LIT, sysf
+  dq LIT, sysf.len
+  dq EXIT
+
+forth INPUT_BUFFER, 'INPUT-BUFFER'
+  dq LIT, input_buffer
+  dq EXIT
+
+forth INPUT_LENGTH, 'INPUT-LENGTH'
+  dq LIT, input_buffer_length
+  dq EXIT
+
+section '.data' readable writable
+
+;; The LATEST variable holds a pointer to the word that was last added to the
+;; dictionary. This pointer is updated as new words are added, and its value is
+;; used by FIND to look up words.
 latest_entry dq initial_latest_entry
 
 ;; The STATE variable is 0 when the interpreter is executing, and non-zero when
 ;; it is compiling.
 var_STATE dq 0
 
+;; The interpreter can read either from standard input or from a buffer. When
+;; input-buffer is set (non-null), words like READ-WORD and S" will use this
+;; buffer instead of reading user input.
+input_buffer dq 0
+input_buffer_length dq 0
+
 FIND.rsi dq ?
 
 READ_WORD.rsi dq ?
 READ_WORD.rbp dq ?
 
+READ_STRING.char_buffer db ?
+READ_STRING.buffer rb $FF
+READ_STRING.length dq ?
+
 DOTU.chars db '0123456789ABCDEF'
 DOTU.buffer rq 16               ; 64-bit number has no more than 16 digits in hex
 DOTU.rbuffer rq 16
 DOTU.length dq ?
 DOTU.printed_length dq ?
 
+KEY.buffer dq ?
+
+READ_WORD.buffer rb $FF
+READ_WORD.length db ?
+
+;; Reserve space for compiled words, accessed through HERE.
+here dq here_top
+here_top rq $4000
+
 ;; Return stack
 rq $2000
 return_stack_top:
+
+;; We store some Forth code in sys.f that defined common words that the user
+;; would expect to have available at startup. To execute these words, we just
+;; include the file directly in the binary, and then interpret it at startup.
+sysf file 'sys.f'
+sysf.len = $ - sysf
+