IT SORTA WORKS LESGOOO

twasm/README.md

@@ -210,11 +210,11 @@ type metadata for the different types is as follows:
 entries are as follows:
 
 ```
-+----------+--------+----------+
++-----------------+-----------------+----------+
 | 31           24 | 23           16 | 15     0 |
-+----------+--------+----------+
++-----------------+-----------------+----------+
-| reserved | opcode | token ID |
+| dest=reg opcode | dest=r/m opcode | token ID |
-+----------+--------+----------+
++-----------------+-----------------+----------+
 ```
 
 note the lack of support for multiple-byte opcodes or multiple opcodes for one token ID; these features will likely be added at some point after the parser accumulates too much jank.

twasm/asm/main.asm

@@ -124,16 +124,20 @@ assemble:
     pop rsi
 
     push rdi
+    push rsi
                 ; di = tte of operator
+    mov sil, 0b ; standard opcode
     call get_opcode
     ; al = opcode
     call .output_byte
+    pop rsi
     pop rdi
 
     pop rax ; rax = number of tokens processed
     pop rdi ; rdi = total number of tokens
     inc rax
     cmp rax, rdi
+    jge .break
     jmp .loop
 
   ; di = tte of operator
@@ -144,10 +148,13 @@ assemble:
     pop rsi
 
     push rdi
+    push rsi
                 ; di = tte of operator
+    mov sil, 0b ; dst=r/m
     call get_opcode
     ; al = opcode
     call .output_byte
+    pop rsi
     pop rdi ; di = tte of operator
 
     pop rax ; rax = number of tokens processed
@@ -214,12 +221,7 @@ assemble:
     call print.debug
     pop rsi
 
-    push rdi
+    mov cx, di ; cx = tte of operator
-    ; di = tte of operator
-    call get_opcode
-    ; al = opcode
-    call .output_byte
-    pop rdi ; di = tte of operator
 
     pop rax
     pop rdi
@@ -237,9 +239,11 @@ assemble:
     pop rdi        ; di = next tte
     je .operator_2_memory_access
 
+    push rcx
     ; di = next tte
     call get_tte_type
     ; al = type of token
+    pop rcx ; cx = tte of operator
 
     cmp al, 0x02 ; type: register
     je .operator_2_register
@@ -256,7 +260,80 @@ assemble:
     mov rsi, .msg_operator_2_memory_access
     call print.debug
     pop rsi
-    jmp .unsupported_memory_access
+
+    cmp di, 0x1000                 ; check if token is addressing a register
+    jne .unsupported_memory_access ; if not, unsupported :/
+
+
+    push rdi
+    mov di, cx ; di = tte of operator
+    mov sil, 0 ; dst = r/m
+    call get_opcode
+    ; al = opcode
+    call .output_byte
+    pop rdi
+
+    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
+
+    mov si, di ; si = dst tte
+
+    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
+
+    push rdi
+    and di, 0xFF00
+    cmp di, 0x1000                ; check if token is a memory address
+    pop rdi                       ; di = next tte
+    je .unsupported_memory_access ; no case of *],[* in asm
+
+    ; di = next tte
+    call get_tte_type
+    ; al = type of token
+
+    cmp al, 0x02
+    je .operator_2_memory_access_register
+
+    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_memory_access_register:
+    push rsi
+    mov rsi, .msg_operator_2_memory_access_register
+    call print.debug
+    pop rsi
+
+    ; si = r/m; dst tte
+    ; di = reg; src tte
+    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:
     push rsi
@@ -264,7 +341,15 @@ assemble:
     call print.debug
     pop rsi
 
-    mov si, di     ; si = `R/M` tte
+    push rdi
+    mov di, cx ; di = tte of operator
+    mov sil, 1 ; dst = reg
+    call get_opcode
+    ; al = opcode
+    call .output_byte
+    pop rdi
+
+    mov si, di ; si = dst tte
 
     pop rax
     pop rdi
@@ -290,7 +375,7 @@ assemble:
     je .operator_2_register_register
 
     pop rax ; rax = number of tokens processed
-    pop rdi ; rdi = total nnumber of tokens
+    pop rdi ; rdi = total number of tokens
     inc rax
     cmp rax, rdi
     jge .break
@@ -338,11 +423,9 @@ assemble:
     call print.debug
     pop rsi
 
-    ; si = `R/M` tte
-    ; di = `reg` tte
     push rsi
-    mov si, di
+    mov si, di  ; si = reg; src tte
-    pop rdi
+    pop rdi     ; di = r/m; dst tte
     mov dl, 11b ; dl = mod bits
     call get_ModRM
     ; al = Mod R/M byte
@@ -388,6 +471,7 @@ assemble:
   .msg_operator_1_register db "operator_1_register", 0x0A, 0x00
   .msg_operator_2 db "operator_2", 0x0A, 0x00
   .msg_operator_2_memory_access db "operator_2_memory_access", 0x0A, 0x00
+  .msg_operator_2_memory_access_register db "operator_2_memory_access_register", 0x0A, 0x00
   .msg_operator_2_register db "operator_2_register", 0x0A, 0x00
   .msg_operator_2_register_memory_access db "operator_2_register_memory_access", 0x0A, 0x00
   .msg_operator_2_register_register db "operator_2_register_register", 0x0A, 0x00
@@ -514,10 +598,13 @@ get_ModRM:
 ; get_opcode
 ;
 ; description:
-; given an operator token, returns its opcode
+; given an operator token, returns its opcode. For operators with multiple
+; opcodes, the variant can be specified.
 ;
 ; parameters:
 ; di = token table entry
+; sil = lower bit: 0: dst=r/m or only opcode
+;                  1: dst=reg or 0x00
 ;
 ; returned:
 ; al = opcode; the rest of rax is zeroed
@@ -525,6 +612,7 @@ get_ModRM:
 
 get_opcode:
   and rdi, 0xFFFF
+  and rsi, 1
   xor eax, eax
 
   .loop:
@@ -543,7 +631,7 @@ get_opcode:
     mov al, UNRECOGNISED_ID_OPCODE
     ret
   .found:
-    mov al, [2 + opcodes.by_id + rax * 4]
+    mov al, [rsi + 2 + opcodes.by_id + rax * 4]
     and rax, 0xFF ; mask
     ret
 
@@ -817,7 +905,7 @@ tokenise:
     jmp .operand_break_continue
 
   ; cx = token ID
-  .operand_addr_register
+  .operand_addr_register:
     mov word [TOKEN_TABLE_ADDR + rax * TOKEN_TABLE_ENTRY_SIZE], 0x1000
     inc rax ; 0x1000: addr reg token, next token is the register
     mov [TOKEN_TABLE_ADDR + rax * TOKEN_TABLE_ENTRY_SIZE], cx
@@ -1417,12 +1505,12 @@ trim_trailing_whitespace:
   cmp al, 0 ; if last element whitespace
   je .done  ; then break
 
-  .trim     ; otherwise one shorter
+  .trim:    ; otherwise one shorter
     dec rdi
     call trim_trailing_whitespace
     ret
 
-  .done
+  .done:
     mov rax, rdi
     ret
 
@@ -1707,20 +1795,20 @@ tokens:
 opcodes:
   .by_id:
     dw 0x0053 ; xor
-    db 0x31
+    db 0x31   ; r/m <- reg
-    db 0x00   ; reserved
+    db 0x33   ; reg <- r/m
 
     dw 0x0054 ; inc
-    db 0xFF
+    db 0xFF   ; r/m
-    db 0x00   ; reserved
+    db 0x00   ;
 
     dw 0x0056 ; mov
-    db 0x8B
+    db 0x89   ; r/m <- reg
-    db 0x00   ; reserved
+    db 0x8B   ; reg <- r/m
 
     dw 0x004F ; hlt
-    db 0xF4
+    db 0xF4   ; .
-    db 0x00   ; reserved
+    db 0x00   ;
   .by_id_end:
 
 msg_welcome db "Welcome to Twasm", 0x0A, 0x00
@@ -1735,7 +1823,7 @@ program:
   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
+  db "hlt", 0x0A
-                 ; would mean range checking members in operator/operand/com-
+  .size dq $ - program
-                 ; ment loops at the end of the program
+
-  .size db $ - program
+msg_end db "end of the binary ->|", 0x0A, 0x00

twasm/asm/tests.asm

@@ -13,6 +13,8 @@ run_tests:
   mov rsi, .msg
   call print.test
 
+  call test_completeness
+
   call clear_test_arena
   call test_djb2
 
@@ -46,6 +48,27 @@ run_tests:
   ret
   .msg db "running test suite...", 0x0A, 0x00
 
+; ------------------------------------------------------------------------------
+; test_completeness
+;
+; description:
+; visual confirmation of binary integrity
+; ------------------------------------------------------------------------------
+
+test_completeness:
+  mov rsi, .msg
+  call print.test
+  mov rsi, .msg_content
+  call print
+  mov rsi, msg_end
+  call print
+  mov rsi, .msg_confirm
+  call print
+  ret
+  .msg db "test_completeness...", 0x0A, 0x00
+  .msg_content db "  here is the ", 0x00
+  .msg_confirm db "  here is the end of the binary ->|", 0x0A, " assert: the previous 2 lines are identical", 0x0A, 0x00
+
 ; ------------------------------------------------------------------------------
 ; test_elemb
 ;
@@ -228,6 +251,11 @@ test_get_tte_typed_metadata:
   cmp al, 0x02                ; # operands
   jne .fail
 
+  mov di, 0x004F              ; hlt
+  call get_tte_typed_metadata
+  cmp al, 0x00                ; # operands
+  jne .fail
+
   mov di, 0x0003              ; rdx
   call get_tte_typed_metadata
   cmp al, 00001011b           ; reg: 010b
@@ -266,21 +294,32 @@ test_get_opcode:
   call print.test
 
   mov di, 0x0053 ; xor
+  mov sil, 0b
   call get_opcode
   cmp al, 0x31
   jne .fail
 
+  mov di, 0x0053 ; xor
+  mov sil, 1b
+  call get_opcode
+  cmp al, 0x33
+  jne .fail
+
+
   mov di, 0x0054  ; inc
+  mov sil, 0b
   call get_opcode
   cmp al, 0xFF
   jne .fail
 
   mov di, 0x004F  ; hlt
+  mov sil, 0b
   call get_opcode
   cmp al, 0xF4
   jne .fail
 
   mov di, 0x0003  ; rdx (not an operator)
+  mov sil, 0b
   call get_opcode
   cmp al, UNRECOGNISED_ID_OPCODE
   jne .fail