X-Git-Url: https://git.rrq.au/?a=blobdiff_plain;f=main.asm;h=e510410a5069b239621a1c14edb58893bce05dac;hb=61b721d365efaaae4d5ebb8d13972faa697be54b;hp=e5444b87a58ba20ad0f984627d05ef0e99efecec;hpb=ad2d7300c0ec7b547259fdfd682bb7f0ee6a01ab;p=rrq%2Fjonasforth.git

diff --git a/main.asm b/main.asm
index e5444b8..e510410 100644
--- a/main.asm
+++ b/main.asm
@@ -1,4 +1,31 @@
-format ELF64 executable
+;; vim: syntax=fasm
+
+;; At compile-time we load the module given by the environment variable
+;; OS_INCLUDE. All of the following these procedures should preserve the value
+;; of RSI and RSP. They may use other registers as they like.
+;;
+;; The module should provide the following:
+;;
+;; os_code_section
+;;   Macro to start the text segment.
+;;
+;; os_data_section
+;;   Macro to start the data segment.
+;;
+;; 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 RAX.
+;;
+;; os_terminate
+;;   Shut down the system, returning the error code given in RAX.
+include '%OS_INCLUDE%'
 
 ;; 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
@@ -23,197 +50,684 @@ macro popr x {
   add rbp, 8
 }
 
-segment readable executable
+;; The following macro generates the dictionary header. It updates the
+;; initial_latest_entry variable, which is used as the initial value of the
+;; latest_entry variable that is made available at runtime.
+;;
+;; The header contains a link to the previous entry, the length of the name of
+;; the word and the word itself as a string literal.
+;;
+;; This macro also defines a label LABEL_entry.
+initial_latest_entry = 0
+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:
+label:
+
+initial_latest_entry = label#_entry
+}
+
+;; Define a Forth word that is implemented in assembly. See 'header' for details.
+macro forth_asm label, name, immediate {
+  header label, name, immediate
+  dq .start
+.start:
+}
+
+include "impl.asm"      ; Misc. subroutines
+include "bootstrap.asm" ; Forth words encoded in Assembly
+
+os_code_section
 
 main:
   cld                        ; Clear direction flag so LODSQ does the right thing.
   mov rbp, return_stack_top  ; Initialize return stack
 
-  mov rsi, program
-  next
+  call os_initialize
+
+  mov rax, MAIN
+  jmp qword [rax]
 
 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.
-EXIT:
-  dq .start
-.start:
+;; restore the old value of RSI (saved by 'DOCOL') and resume execution.
+forth_asm EXIT, 'EXIT'
   popr rsi
   next
 
 ;; LIT is a special word that reads the next "word pointer" and causes it to be
 ;; placed on the stack rather than executed.
-LIT:
-  dq .start
-.start:
+forth_asm LIT, 'LIT'
   lodsq
   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'
+  mov [.rsi], rsi
+
+  pop [find.search_length]
+  pop [find.search_buffer]
+  mov rsi, [latest_entry]       ; Start with the last added word
+  call find
+  push rsi
+
+  mov rsi, [.rsi]
+  next
+  push rsi
+
+  mov rsi, [.rsi]
+  next
+
+;; Given an entry in the dictionary, return a pointer to the codeword of that
+;; entry.
+forth_asm TCFA, '>CFA'
+  pop rax
+  add rax, 8 + 1                ; [rax] = length of name
+  movzx rbx, byte [rax]
+  inc rax
+  add rax, rbx                  ; [rax] = codeword
+  push rax
+  next
+
+;; BRANCH is the fundamental mechanism for branching. BRANCH reads the next word
+;; as a signed integer literal and jumps by that offset.
+forth_asm BRANCH, 'BRANCH'
+  add rsi, [rsi] ; [RSI], which is the next word, contains the offset; we add this to the instruction pointer.
+  next           ; Then, we can just continue execution as normal
+
+;; 0BRANCH is like BRANCH, but it jumps only if the top of the stack is zero.
+forth_asm ZBRANCH, '0BRANCH'
+  ;; Compare top of stack to see if we should branch
+  pop rax
+  cmp rax, 0
+  jnz .dont_branch
+.do_branch:
+  jmp BRANCH.start
+.dont_branch:
+  add rsi, 8     ; We need to skip over the next word, which contains the offset.
+  next
+
+;; Duplicate the top of the stack.
+forth_asm DUP_, 'DUP'
+  push qword [rsp]
+  next
+
+;; Execute the codeword at the given address.
+forth_asm EXEC, 'EXEC'
+  pop rax
+  jmp qword [rax]
+
 ;; Expects a character on the stack and prints it to standard output.
-EMIT:
-  dq .start
-.start:
+forth_asm EMIT, 'EMIT'
   pushr rsi
   pushr rax
-  mov rax, 1
-  mov rdi, 1
-  lea rsi, [rsp]
+
+  lea rcx, [rsp]
   mov rdx, 1
-  syscall
+  call os_print_string
+
   add rsp, 8
   popr rax
   popr rsi
   next
 
-;; Prints a newline to standard output.
-NEWLINE:
-  dq docol
-  dq LIT, $A
-  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_WORD:  ; 400170
-  dq .start
-.start:
-  mov [.rsi], rsi
-  mov [.rax], rax
+;; 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
 
+;; 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
+  call os_read_char
+  ret
+
+.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'
+  push rsi
 .skip_whitespace:
-  ;; Read characters into .char_buffer until one of them is not whitespace.
-  mov rax, 0
-  mov rdi, 0
-  mov rsi, .char_buffer
-  mov rdx, 1
-  syscall
-
-  cmp [.char_buffer], ' '
+  ;; 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 [.char_buffer], $A
+  cmp al, $A
   je .skip_whitespace
 
-.alpha:
   ;; We got a character that wasn't whitespace. Now read the actual word.
   mov [.length], 0
 
 .read_alpha:
-  mov al, [.char_buffer]
   movzx rbx, [.length]
   mov rsi, .buffer
   add rsi, rbx
   mov [rsi], al
   inc [.length]
 
-  mov rax, 0
-  mov rdi, 0
-  mov rsi, .char_buffer
-  mov rdx, 1
-  syscall
+  call KEY.impl
 
-  cmp [.char_buffer], ' '
+  cmp al, ' '
   je .end
-  cmp [.char_buffer], $A
+  cmp al, $A
   jne .read_alpha
 
 .end:
+  pop rsi
   push .buffer
   movzx rax, [.length]
   push rax
 
-  mov rsi, [.rsi]
-  mov rax, [.rax]
+  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 rcx     ; Length
+  pop rdi     ; String pointer
+
+  push rsi
+  call parse_number
+  pop rsi
+
+  push rax                      ; Result
   next
 
 ;; Takes a string (in the form of a pointer and a length on the stack) and
 ;; prints it to standard output.
-TYPE:
-  dq .start
-.start:
-  mov rbx, rsi
-  mov rcx, rax
+forth_asm TELL, 'TELL'
+  pushr rax
+  pushr rsi
 
-  mov rax, 1
-  mov rdi, 1
-  pop rdx     ; Length
-  pop rsi     ; Buffer
-  syscall
+  pop rdx ; Length
+  pop rcx ; Buffer
+  call os_print_string
 
-  mov rax, rcx
-  mov rsi, rbx
+  popr rsi
+  popr rax
   next
 
 ;; Exit the program cleanly.
-TERMINATE:
-  dq .start
-.start:
-  mov rax, $3C
-  mov rdi, 0
-  syscall
-
-PUSH_HELLO_CHARS:
-  dq docol
-  dq LIT, $A
-  dq LIT, 'o'
-  dq LIT, 'l'
-  dq LIT, 'l'
-  dq LIT, 'e'
-  dq LIT, 'H'
+forth_asm TERMINATE, 'TERMINATE'
+  mov rax, 0
+  call os_terminate
+
+;; 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'
+  mov [.length], 0
+  mov [.printed_length], 1
+  pop rax                       ; RAX = value to print
+  push rsi                      ; Save value of RSI
+
+  ;; We start by constructing the buffer to print in reverse
+
+.loop:
+  mov rdx, 0
+  mov rbx, $10
+  div rbx                       ; Put remainer in RDX and quotient in RAX
+
+  ;; Place the appropriate character in the buffer
+  mov rsi, .chars
+  add rsi, rdx
+  mov bl, [rsi]
+  mov rdi, .rbuffer
+  add rdi, [.length]
+  mov [rdi], bl
+  inc [.length]
+
+  ;; .printed_length is the number of characters that we ulitmately want to
+  ;; print. If we have printed a non-zero character, then we should update
+  ;; .printed_length.
+  cmp bl, '0'
+  je .skip_updating_real_length
+  mov rbx, [.length]
+  mov [.printed_length], rbx
+.skip_updating_real_length:
+
+  cmp [.length], 16
+  jle .loop
+
+  ;; Flip buffer around, since it is currently reversed
+  mov rcx, [.printed_length]
+.flip:
+  mov rsi, .rbuffer
+  add rsi, rcx
+  dec rsi
+  mov al, [rsi]
+
+  mov rdi, .buffer
+  add rdi, [.printed_length]
+  sub rdi, rcx
+  mov [rdi], al
+
+  loop .flip
+
+  ;; Print the buffer
+  mov rcx, .buffer
+  mov rdx, [.printed_length]
+  call os_print_string
+
+  ;; Restore RSI and continue execution
+  pop rsi
+  next
+
+;; Takes a value and an address, and stores the value at the given address.
+forth_asm PUT, '!'
+  pop rbx                       ; Address
+  pop rax                       ; Value
+  mov [rbx], rax
+  next
+
+;; Takes an address and returns the value at the given address.
+forth_asm GET, '@'
+  pop rax
+  mov rax, [rax]
+  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
+  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:
+  call os_read_char
+  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 SYSCODE
+  dq INTERPRET_STRING
+  dq INTERPRET
+  dq BRANCH, -8 * 2
+  dq TERMINATE
+
+;; EFI:
+
+forth EFI_SYSTEM_TABLE_CONSTANT, 'SystemTable'
+  dq LIT, system_table, GET
   dq EXIT
 
-PUSH_YOU_TYPED:
-  dq .start
-.start:
-  push you_typed_string
-  push you_typed_string.length
-  next
-
-HELLO:
-  dq docol
-  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
+forth_asm EFICALL2, 'EFICALL2'
+  pop rax ; function pointer
+  pop rdx ; 2nd argument
+  pop rcx ; 1st argument
+
+  sub rsp, 32
+  call rax
+  add rsp, 32
+
+  next
+
+forth_asm EFICALL3, 'EFICALL3'
+  pop rax ; function pointer
+  pop r8  ; 3rd argument
+  pop rdx ; 2nd argument
+  pop rcx ; 1st argument
+
+  sub rsp, 32
+  call rax
+  add rsp, 32
+
+  push rax
+
+  next
+
+forth_asm EFICALL10, 'EFICALL10'
+  pop rax ; function pointer
+
+  mov rcx, [rsp + 8 * 9]
+  mov rdx, [rsp + 8 * 8]
+  mov r8, [rsp + 8 * 7]
+  mov r9, [rsp + 8 * 6]
+
+  ;; Reverse order of stack arguments
+  mov r10, [rsp + 8 * 5]
+  mov r11, [rsp + 8 * 0]
+  mov [rsp + 8 * 5], r11
+  mov [rsp + 8 * 0], r10
+
+  mov r10, [rsp + 8 * 4]
+  mov r11, [rsp + 8 * 1]
+  mov [rsp + 8 * 4], r11
+  mov [rsp + 8 * 1], r10
+
+  mov r10, [rsp + 8 * 3]
+  mov r11, [rsp + 8 * 2]
+  mov [rsp + 8 * 3], r11
+  mov [rsp + 8 * 2], r10
+
+  sub rsp, 32
+  call rax
+  add rsp, 32 + 8 * 10
+
+  push rax
+
+  next
+
+;; Built-in variables:
+
+forth STATE, 'STATE'
+  dq LIT, var_STATE
   dq EXIT
 
-MAIN:
-  dq docol
-  dq HELLO
-  dq READ_WORD
-  dq LIT, you_typed_string
-  dq LIT, you_typed_string.length
-  dq TYPE
-  dq TYPE
-  dq NEWLINE
-  dq HELLO
-  dq TERMINATE
+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
 
-segment readable writable
+forth INPUT_BUFFER, 'INPUT-BUFFER'
+  dq LIT, input_buffer
+  dq EXIT
+
+forth INPUT_LENGTH, 'INPUT-LENGTH'
+  dq LIT, input_buffer_length
+  dq EXIT
 
-you_typed_string db 'You typed: '
-.length = $ - you_typed_string
+os_data_section
+
+;; 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.rax dq ?
-READ_WORD.max_size = $FF
-READ_WORD.buffer rb READ_WORD.max_size
+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 ?
-READ_WORD.char_buffer 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'
+file 'uefi.f'
+sysf.len = $ - sysf
+