add get_reg_bits and test, streamline ModRM calculations, correct opcodes, fix typos...

2 files changed, 199 insertions, 64 deletions

diff --git a/twasm/asm/main.asm b/twasm/asm/main.asm
index 12f54fd..96d15bc 100644
--- a/twasm/asm/main.asm
+++ b/twasm/asm/main.asm
@@ -53,7 +53,7 @@ start:
 
 ; ------------------------------------------------------------------------------
 ; assemble
-; TODO write testsr
+; TODO write tests
 ; TODO make it work :/ putting the cart before the horse
 ;
 ; description:
@@ -73,50 +73,111 @@ assemble:
 
     push rdi
     xor edi, edi
-    mov di, [rax * TOKEN_TABLE_ENTRY_SIZE + TOKEN_TABLE_ADDR] ; rdi = next tte
+    mov di, [rax * TOKEN_TABLE_ENTRY_SIZE + TOKEN_TABLE_ADDR] ; next tte
     push rax
-    xor eax, eax
-    call get_tte_type
 
-    cmp ax, 0x01 ; check if it's an operator
-    je .operator
-    jne .continue_operator
+    ; di = next tte
+    call get_tte_type
+    ; al = type of token
+    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
 
-  .operator:
-    push rsi
-    mov rsi, .msg_found_operator
-    call print
-    pop rsi
+  .operator: ; if next tte's type is an operator:
+    push rax ; MUST be popped BEFORE returning to .continue_operator; it
+             ; contains the type of token, which still needs to be used.
 
     push rdi
-
     ; di = tte
     call get_tte_typed_metadata
     ; al = tte typed metadata
-
     pop rdi
 
     and al, 11b ; mask for # operands
 
-    cmp al, 0 ; 0 operands
-    je .operator_0
-    jne .continue_operator0
+    cmp al, 0               ; check if operator has no operands
+    je .operator_0          ; if so, handle case of no operands
+    jne .operator_with_args ; if not, jump to case of multiple operands
 
   .operator_0:
+    push rdi
+    ; di = next tte
+    call get_opcode
+    ; al = opcode
+    call .output_byte
+    pop rdi
+
+    pop rax ; from start of label .operator
     jmp .continue_operator
 
-  .continue_operator0:
+  .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 it fttb
+    pop rdi
+
+    pop rax ; from start of label .operator
 
   .continue_operator:
-    cmp ax, 0x02 ; check if it's a register
-    je .register
-    jne .continue_register
+    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
 
-  .register:
-    push rsi
-    mov rsi, .msg_found_register
-    call print
-    pop rsi
+  .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, quietly discard the token, reset things, and keep going :/
+    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
+    call get_direct_addressing_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:
     pop rax ; incrementer
@@ -127,10 +188,43 @@ assemble:
 
   .break:
     ret
-  .msg_found_operator db "found operator", 0x0A, 0x00
-  .msg_found_register db "found register", 0x0A, 0x00
-  .next_register dw UNRECOGNISED_TOKEN_ID
-  .next_next_register dw UNRECOGNISED_TOKEN_ID
+
+  ; procedures
+
+  ; al = byte to write
+  .output_byte:
+    mov edx, [.next_output_byte] ; get output byte's address
+    mov [edx], al                ; write byte to that address
+    inc edx                      ; increment address
+    mov [.next_output_byte], edx ; put output byte's address
+    ret
+
+  ; runs dec on .pending_operator_num_args
+  .dec_num_args
+    push rax
+    mov al, [.pending_operator_num_args]
+    dec al
+    mov [.pending_operator_num_args], al
+    pop rax
+    ret
+
+  .reset_state
+    ; I don't actually know if these `word` and `byte` 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 byte [.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
 
 ; ------------------------------------------------------------------------------
 ; get_tte_type
@@ -226,44 +320,24 @@ get_tte_typed_metadata:
 ; ------------------------------------------------------------------------------
 
 get_direct_addressing_ModRM:
-  ; TODO something is backwards in this function but I don't see it. If the test
-  ; suite fails, it's too far gone; rewrite it.
-  push rdi
-  push rsi
-  ; get metadata of reg
-  call get_tte_typed_metadata
-  ; al = typed metadata of reg
-  pop rsi
-  pop rdi
+  ; di = tte
+  call get_reg_bits
+  ; al = reg bits
+  mov bl, al
+  shl bl, 3
 
-  mov bl, al ; bl = metadata of reg
+  mov rdi, rsi ; do the other one
 
-  push rdi
-  push rsi
-  push rbx
-
-  ; get metadata of R/M
-  mov di, si
-  call get_tte_typed_metadata
-  ; al = typed metadata of R/M
-
-  pop rbx
-  pop rsi
-  pop rdi
+  ; di = tte
+  call get_reg_bits
+  ; al = reg bits
 
   mov dl, al
 
-  shr dl, 2
-  and dl, 111b ; mask
-
-  shr bl, 2
-  and bl, 111b ; mask
-  shl bl, 3
-
   xor eax, eax
   or al, 11b << 6 ; mod bits
-  or al, dl ; reg bits
-  or al, bl ; R/M bits
+  or al, bl       ; reg bits
+  or al, dl       ; R/M bits
   and rax, 0xFF ; mask for byte
   ret
 
@@ -305,6 +379,28 @@ get_opcode:
     ret
 
 ; ------------------------------------------------------------------------------
+; get_reg_bits
+;
+; description:
+; given a register token, returns its reg bits metadata
+;
+; parameters:
+; di = token table entry
+;
+; returned:
+; al = register token; the rest of rax, including the upper 5 bits of al, are
+;      zeroed.
+; ------------------------------------------------------------------------------
+
+get_reg_bits:
+  ; di = tte
+  call get_tte_typed_metadata
+  ; al = typed metadata
+  shr al, 2    ; discard type data
+  and al, 111b ; mask
+  ret
+
+; ------------------------------------------------------------------------------
 ; tokenising
 ; ------------------------------------------------------------------------------
 
@@ -1024,7 +1120,7 @@ tokens:
 opcodes:
   .by_id:
     dw 0x0053 ; xor
-    db 0x33
+    db 0x31
     db 0x00   ; reserved
 
     dw 0x0054 ; inc
@@ -1032,7 +1128,7 @@ opcodes:
     db 0x00   ; reserved
 
     dw 0x0056 ; mov
-    db 0x8B
+    db 0x89
     db 0x00   ; reserved
 
     dw 0x004F ; hlt
diff --git a/twasm/asm/tests.asm b/twasm/asm/tests.asm
index 22d626a..4705829 100644
--- a/twasm/asm/tests.asm
+++ b/twasm/asm/tests.asm
@@ -40,6 +40,9 @@ run_tests:
   call clear_test_arena
   call test_get_opcode
 
+  call clear_test_arena
+  call test_get_reg_bits
+
   ret
   .msg db "running test suite...", 0x0A, 0x00
 
@@ -529,7 +532,7 @@ test_get_opcode:
 
   mov di, 0x0053 ; xor
   call get_opcode
-  cmp al, 0x33
+  cmp al, 0x31
   jne .fail
 
   mov di, 0x0054  ; inc
@@ -557,6 +560,42 @@ test_get_opcode:
     ret
   .msg db "test_get_opcode...", 0x00
 
+; ------------------------------------------------------------------------------
+; test_get_reg_bits
+;
+; description:
+; tests get_reg_bits described functionality
+; ------------------------------------------------------------------------------
+
+test_get_reg_bits:
+  mov rsi, .msg
+  call print
+
+  mov di, 0x0000    ; rax
+  call get_reg_bits
+  cmp al, 000b
+  jne .fail
+
+  mov di, 0x0010    ; eax
+  call get_reg_bits
+  cmp al, 000b
+  jne .fail
+
+  mov di, 0x0003    ; rdx
+  call get_reg_bits
+  cmp al, 010b
+  jne .fail
+
+  .pass:
+    mov rsi, msg_pass
+    call print
+    ret
+  .fail:
+    mov rsi, msg_fail
+    call print
+    ret
+  .msg db "test_get_reg_bits...", 0x00
+
 msg_pass:
   db 0x0A
   times (TEST_LINE_LENGTH + .start - .end) db " ", ; right align