X-Git-Url: https://git.rrq.au/?a=blobdiff_plain;f=main.asm;h=a670b343e99c1a1f0154f0eab083ef6eb1e6cc1a;hb=7677dbbdce584a64e63946b39879f5edd114f36a;hp=8133cdc33b284a2ce2b3764df1c2304bbe40e9fa;hpb=89a6a52d6128a4f9ac3c7aa26fc0df111e4160db;p=rrq%2Fjonasforth.git

diff --git a/main.asm b/main.asm
index 8133cdc..a670b34 100644
--- a/main.asm
+++ b/main.asm
@@ -1,5 +1,35 @@
+;; vim: syntax=fasm
+
 format ELF64 executable
 
+;; "Syscalls" {{{
+
+;; [NOTE] Volatile registers Linux (syscalls) vs UEFI
+;;
+;;   Linux syscalls: RAX, RCX, R11
+;;   UEFI:           RAX, RCX, R11, RDX, R8, R9, R10
+
+;; We are in the process of replacing our dependency on Linux with a dependency
+;; on UEFI. The following macros attempt to isolate what would be syscalls in
+;; Linux; thus, we will be able to replace these with UEFI-based implementations,
+;; and in theory we should expect the program to work.
+
+;; 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 {
+  mov rax, 1
+  mov rdi, 1
+  mov rsi, rcx
+  syscall
+}
+
+;; }}}
+
 ;; 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
 ;; next word in that definition.
@@ -37,7 +67,11 @@ macro header label, name, immediate {
 
 label#_entry:
   dq initial_latest_entry
-  dq 0
+  if immediate eq
+    db 0
+  else
+    db 1
+  end if
   db .string_end - ($ + 1)
   db name
   .string_end:
@@ -53,18 +87,12 @@ macro forth_asm label, name, immediate {
 .start:
 }
 
-;; Define a Forth word that is implemented in Forth. (The body will be a list of
-;; 'dq' statements.)
-macro forth label, name, immediate {
-  header label, name, immediate
-  dq DOCOL
-}
-
 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.
@@ -118,7 +146,7 @@ forth_asm FIND, 'FIND'
 ;; entry.
 forth_asm TCFA, '>CFA'
   pop rax
-  add rax, 16                   ; [rax] = length of name
+  add rax, 8 + 1                ; [rax] = length of name
   movzx rbx, byte [rax]
   inc rax
   add rax, rbx                  ; [rax] = codeword
@@ -157,31 +185,24 @@ 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
-
-;; Prints a space to standard output.
-forth SPACE, 'SPACE'
-  dq LIT, ' '
-  dq EMIT
-  dq EXIT
-
-;; 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.
+;; 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'
+  ;; Are we reading from user input or from the input buffer?
+  cmp [input_buffer], 0
+  jne .from_buffer
+
+  ;; Reading user input
   mov [.rsi], rsi
 
   call read_word
@@ -191,11 +212,28 @@ forth_asm READ_WORD, 'READ-WORD'
   mov rsi, [.rsi]
   next
 
+.from_buffer:
+  ;; Reading from buffer
+  mov [.rsi], rsi
+
+  mov rsi, [input_buffer]
+  mov rcx, [input_buffer_length]
+
+  call pop_word
+
+  mov [input_buffer], rsi        ; Updated buffer
+  mov [input_buffer_length], rcx ; Length of updated buffer
+  push rdi                       ; Word buffer
+  push rdx                       ; Length of word buffer
+
+  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
@@ -204,25 +242,18 @@ 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.
@@ -231,16 +262,6 @@ forth_asm TERMINATE, 'TERMINATE'
   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
-
 ;; Duplicate a pair of elements.
 forth_asm PAIRDUP, '2DUP'
   pop rbx
@@ -264,31 +285,16 @@ forth_asm DROP, 'DROP'
   add rsp, 8
   next
 
-;; 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'
-  ;; Read word
-  dq READ_WORD
-  dq PAIRDUP
-  ;; Stack is (word length word length).
-  dq FIND                       ; Try to find word
-  dq DUP_
-  dq ZBRANCH, 8 * 8             ; Check if word is found
-
-  ;; Word is found, execute it
-  dq TCFA
-  ;; Stack is (word length addr)
-  dq SWAP, DROP
-  dq SWAP, DROP
-  ;; Stack is (addr)
-  dq EXEC
-  dq EXIT
-
-  ;; No word is found, assume it is an integer literal
-  ;; Stack is (word length addr)
-  dq DROP
-  dq PARSE_NUMBER
-  dq EXIT
+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'
@@ -341,11 +347,9 @@ 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
@@ -365,6 +369,18 @@ forth_asm GET, '@'
   push rax
   next
 
+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
@@ -382,35 +398,25 @@ forth_asm MINUS, '-'
   push rbx
   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
-
-;; Get the location of the LATEST variable. It can be set with '!' and read with
-;; '@'.
-forth LATEST, 'LATEST'
-  dq LIT, latest_entry
-  dq EXIT
-
-;; Get the location at which compiled words are expected to be added. This
-;; pointer is usually modified automatically when calling ',', but we can also
-;; read it manually with 'HERE'.
-forth HERE, 'HERE'
-  dq LIT, here
-  dq EXIT
-
-forth COMMA, ','
-  dq HERE, GET, PUT             ; Set the memory at the address pointed to by HERE
-  dq HERE, GET, LIT, 8, PLUS    ; Calculate new address for HERE to point to
-  dq HERE, PUT                  ; Update HERE to point to the new address
-  dq EXIT
+;; 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 user input until next " character is found. Push a string containing the
+;; 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 character can be read.
+;; 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
@@ -440,6 +446,40 @@ forth_asm READ_STRING, 'S"'
 
   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.
@@ -455,11 +495,10 @@ forth_asm CREATE, 'CREATE'
   mov [latest_entry], rdi       ; Update LATEST to point to this word
 
   add rdi, 8
-  mov rax, 0
-  mov [rdi], rax                ; Set immediate flag
+  mov [rdi], byte 0             ; Insert immediate flag
 
-  add rdi, 8
-  mov [rdi], rcx                ; Insert length
+  add rdi, 1
+  mov [rdi], byte cl            ; Insert length
 
   ;; Insert word string
   add rdi, 1
@@ -474,27 +513,73 @@ forth_asm CREATE, 'CREATE'
 
   next
 
-;; Mark the last added word as immediate.
-forth IMMEDIATE, 'IMMEDIATE', 1
-  dq LIT, 1
-  dq LATEST, GET
-  dq LIT, 8, PLUS
-  dq PUT
-  dq EXIT
+forth_asm TICK, "'"
+  lodsq
+  push rax
+  next
 
-;; Return 0 if the given word is not immediate.
-forth IS_IMMEDIATE, 'IMMEDIATE?'
-  dq FIND
-  dq LIT, 8, PLUS
-  dq GET
-  dq EXIT
+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
 
+;; 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
+
 segment readable writable
 
 ;; The LATEST variable holds a pointer to the word that was last added to the
@@ -506,6 +591,12 @@ latest_entry dq initial_latest_entry
 ;; 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 ?
@@ -523,8 +614,17 @@ DOTU.printed_length dq ?
 
 ;; Reserve space for compiled words, accessed through HERE.
 here dq here_top
-here_top rq $2000
+here_top rq $4000
 
 ;; Return stack
 rq $2000
 return_stack_top:
+
+segment readable
+
+;; 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
+