;)

twasm/asm/main.asm

@@ -75,6 +75,7 @@ start:
 ;   0x1000 ; memory address, following byte should be a register
 ;   0x1000 ; not a register
 ; - operator tokens followed by the wrong number of arguments
+;   TODO enforce this in `tokenise`
 ;   0x004F ; hlt, expects 0 arguments
 ;   0x0000 ; rax, an argument
 ; - an undefined token is included, like 0x0051
@@ -92,17 +93,6 @@ assemble:
     mov di, [rax * TOKEN_TABLE_ENTRY_SIZE + TOKEN_TABLE_ADDR] ; next tte
     push rax
 
-    ; di = next tte
-    call get_tte_type
-    ; al = type of token
-
-    ; on stack:
-    ; rax = number of tokens processed
-    ; rdi = number of tokens in table
-    cmp al, 0x01           ; check if next tte's type is an operator
-    je .operator           ; if so, handle case of operator
-    jne .continue_operator ; if not, jump past the case
-
   ; di = tte of operator
   .operator: ; if next tte's type is an operator:
     push rdi
@@ -122,7 +112,9 @@ assemble:
     cmp al, 2      ; check if operator has two operands
     je .operator_2 ; if so, handle case of two operands
 
-    jmp .operator_with_args ; of not, handle case of some operands
+    jmp .unexpected_token ; TODO actually check operator type or not first
+                          ; if get_tte_typed_metadata happens to return 0, 1,
+                          ; or 2 on a non-operator, it doesn't get caught
 
   ; di = tte of operator
   .operator_0:
@@ -138,7 +130,11 @@ assemble:
     call .output_byte
     pop rdi
 
-    jmp .continue
+    pop rax ; rax = number of tokens processed
+    pop rdi ; rdi = total number of tokens
+    inc rax
+    cmp rax, rdi
+    jmp .loop
 
   ; di = tte of operator
   .operator_1:
@@ -188,9 +184,9 @@ assemble:
   .operator_1_memory_access:
     push rsi
     mov rsi, .msg_operator_1_memory_access
-    call print.error
+    call print.debug
     pop rsi
-    jmp halt
+    jmp .unsupported_memory_access
 
   .operator_1_register:
     push rsi
@@ -258,9 +254,9 @@ assemble:
   .operator_2_memory_access:
     push rsi
     mov rsi, .msg_operator_2_memory_access
-    call print.error
+    call print.debug
     pop rsi
-    jmp halt
+    jmp .unsupported_memory_access
 
   .operator_2_register:
     push rsi
@@ -303,9 +299,38 @@ assemble:
   .operator_2_register_memory_access:
     push rsi
     mov rsi, .msg_operator_2_register_memory_access
-    call print.error
+    call print.debug
     pop rsi
-    jmp halt
+
+    cmp di, 0x1000                 ; check if token is addressing to a register
+    jne .unsupported_memory_access ; if not, unsupported :/
+
+    pop rax
+    pop rdi
+    inc rax
+    cmp rax, rdi
+    jge .break
+    push rdi
+    push rax
+    xor edi, edi
+    mov di, [rax * TOKEN_TABLE_ENTRY_SIZE + TOKEN_TABLE_ADDR] ; di = next tte
+
+    ; si = `R/M` tte
+    ; di = `reg` tte
+    push rsi
+    mov si, di
+    pop rdi
+    mov dl, 00b ; dl = mod bits
+    call get_ModRM
+    ; al = Mod R/M byte
+    call .output_byte
+
+    pop rax ; rax = number of tokens processed
+    pop rdi ; rdi = total number of tokens
+    inc rax
+    cmp rax, rdi
+    jge .break
+    jmp .loop
 
   .operator_2_register_register:
     push rsi
@@ -315,6 +340,9 @@ assemble:
 
     ; si = `R/M` tte
     ; di = `reg` tte
+    push rsi
+    mov si, di
+    pop rdi
     mov dl, 11b ; dl = mod bits
     call get_ModRM
     ; al = Mod R/M byte
@@ -327,99 +355,18 @@ assemble:
     jge .break
     jmp .loop
 
-  ; di = tte of operator
-  .operator_with_args:
-    mov [.pending_operator_num_args], al ; save # args fttb
-
-    push rdi
-    ; di = next tte
-    call get_opcode
-    ; al = opcode
-    mov [.pending_operator_opcode], al ; save opcode fttb
-    pop rdi
-
-    jmp .continue
-
-  .continue_operator:
-    cmp al, 0x02           ; check if next tte's type is a register
-    je .register           ; if so, handle case of register
-    jne .continue_register ; if not, jump past the case
-
-  ; di MUST be a valid register tte
-  .register: ; if next tte's type is a register:
-    call .dec_num_args ; because we've found an argument, we need 1 fewer noch
-
-    cmp byte [.pending_operator_num_args], 1 ; check if this is 1st of 2 args
-    je .register_one_of_two                  ; if so, jump to handler
-
-    cmp byte [.pending_operator_num_args], 0 ; check if this is the last arg
-    je .register_last                        ; if so, jump to handler
-                                             ; note: not necessarily the last
-                                             ; of 2 args, it could also be the
-                                             ; last of 1
-
-    ; otherwise, discard the token, reset things, and keep going :/
-    push rsi
-    mov rsi, .warn_unexpected_register
-    call print.warn
-    pop rsi
-    call .reset_state
-    jmp .continue_register
-
-  .register_one_of_two: ; if it's the first of 2 arguments:
-    mov [.first_argument], di ; ax = tte
-    jmp .continue_register
-
-  .register_last: ; if it's the last argument:
-    ; swap so the first argument sits in .first_argument
-    push rax
-    mov ax, di
-    mov di, [.first_argument]
-    mov [.first_argument], ax
-    pop rax
-
-    cmp di, UNRECOGNISED_TOKEN_ID ; check if the second argument is defined
-    jne .operator_finalise_2      ; if so, there are 2 arguments
-                                  ; if not, there is just 1
-
-  .operator_finalise_1:
-    mov di, 0x0000 ; id of rax. reg bits 000b
-
-  .operator_finalise_2:
-    ; TODO avoid swapping earlier and now :/
-    mov cx, di
-    mov di, [.first_argument]
-    mov si, cx
-    mov dl, 11b
-    call get_ModRM
-    ; al = ModR/M byte
-    push rax
-    mov al, [.pending_operator_opcode]
-    call .output_byte ; output operator's opcode
-    pop rax
-
-    call .output_byte ; output ModR/M byte
-
-    call .reset_state ; reset all the state parts of this function
-    jmp .continue_register
-
-  .continue_register:
-
-  .continue:
-    pop rax ; rax = number of tokens processed
-    pop rdi ; rdi = total number of tokens
-
-    inc rax
-    cmp rax, rdi
-    jge .break       ; at end of table, break
-    jmp .loop
-
   .break:
     ret
 
-  ; constants
+  .unexpected_token:
+    mov rsi, .msg_unexpected_token
+    call print.error
+    jmp halt
 
-  .warn_unexpected_register db "ignoring unexpected register", 0x0A, 0x00
+  .unsupported_memory_access:
+    mov rsi, .msg_unsupported_memory_access
+    call print.error
+    jmp halt
 
   ; procedures
 
@@ -430,35 +377,11 @@ assemble:
     inc edx                      ; increment address
     mov [.next_output_byte], edx ; put output byte's address
     ret
+  .next_output_byte dd OUTPUT_ADDR ; next empty byte in output
+                                   ; TODO get rid of this sketchy bit of state
 
-  ; runs dec on .pending_operator_num_args
+  .msg_unexpected_token db "unexpected token, aborting", 0x0A, 0x00
-  .dec_num_args:
+  .msg_unsupported_memory_access db "unsupported memory access, aborting", 0x0A, 0x00
-    push rax
-    mov al, [.pending_operator_num_args]
-    dec al
-    mov [.pending_operator_num_args], al
-    pop rax
-    ret
-
-  ; resets sketchy memory-based state
-  ; TODO put this state in an accumulator or something
-  .reset_state:
-    ; I don't actually know if these `word` directives are needed
-    ; TODO check that. I think they are, becasue Nasm doesn't record the size
-    ; of labels?
-    mov word [.pending_operator_opcode], UNRECOGNISED_TOKEN_ID
-    mov [.pending_operator_num_args], 0x00
-    mov word [.first_argument], UNRECOGNISED_TOKEN_ID
-    ret
-
-  ; state variables
-
-  .pending_operator_opcode db 0x00   ; the operator seeking args
-  .pending_operator_num_args db 0x00 ; # of args it takes
-
-  .first_argument dw UNRECOGNISED_TOKEN_ID ; first argument if there are two
-
-  .next_output_byte dd OUTPUT_ADDR             ; next empty byte in output
   .msg_operator_0 db "operator_0", 0x0A, 0x00
   .msg_operator_1 db "operator_1", 0x0A, 0x00
   .msg_operator_1_memory_access db "operator_1_memory_access", 0x0A, 0x00
@@ -1792,7 +1715,7 @@ opcodes:
     db 0x00   ; reserved
 
     dw 0x0056 ; mov
-    db 0x89
+    db 0x8B
     db 0x00   ; reserved
 
     dw 0x004F ; hlt
@@ -1810,6 +1733,7 @@ program:
   db "xor eax, eax", 0x0A
   db "inc rax ; inline comment", 0x0A
   db "; one line comment", 0x0A
+  db "mov rdx, [rax]", 0x0A
   db "mov [rax], rdx", 0x0A
   db "hlt", 0x0A ; TODO make it so it doesn't need to end with a newline; this
                  ; would mean range checking members in operator/operand/com-