update readme

README.md

@@ -10,4 +10,4 @@ status: gets to long mode, loads+jumps to kernel, starts idt and gdt... :)
 
 hobby self-hosted assembler
 
-status: tokenises a lot of common stuff, compiles a lot of common stuff... not good enough for anything real-world yet
+status: tokenises a lot of common stuff, compiles a lot of common stuff... not good enough for anything real-world yet. It can assemble its own print function!!!

twasm/README.md

@@ -267,8 +267,8 @@ entries are as follows:
 +----------+----------+----------+---------+
 | 31    24 | 23    16 | 15               0 |
 +----------+----------+--------------------+
-| reserved | opcode   | token ID           |
-|          | dst=r/m  |                    |
+| opcode+r | opcode   | token ID           |
+| dst=r    | dst=r/m  |                    |
 +----------+----------+--------------------+
 
 16 bytes
@@ -306,18 +306,19 @@ entries are as follows:
 +----------+------+
 
 1 byte
-+----------------------------------------------------+
++--------------------------------------------------------------+
 | flagsX byte                                                  |
-+----------+-----------+-------------+---------------+
-| 7      5 |     4     |      3      | 2           0 |
-+----------+-----------+-------------+---------------+
-| reserved | no ModR/M | 0x0F prefix | operator flag |
-+----------+-----------+-------------+---------------+
++----------+---------+-----------+-------------+---------------+
+| 7      6 |       5 |         4 |           3 | 2           0 |
++----------+---------+-----------+-------------+---------------+
+| reserved | +r flag | no ModR/M | 0x0F prefix | operator flag |
++----------+---------+-----------+-------------+---------------+
 
 ; flags key:
 8bit ; tte has opcodes for r/m8 and r8 instead of r/m and r respectively
 
 ; flagsX key:
++r flag       ; there is a +r variation of this opcode
 no ModR/M     ; there is no ModR/M byte for this opcode
 0x0F prefix   ; there is a 0x0F prefix for this opcode
 operator flag ; contents of `reg` if applicable
@@ -526,70 +527,3 @@ hlt
 
 0xF4 ; HLT
 ```
-
-#### program output with the function `print`, each comma-seperated `db` value put onto its own line
-
-editted output of `x/512xb 0x00070000` in [gdb](https://www.sourceware.org/gdb/)
-
-the following is somewhat correct! I just need to a) null-terminate 8-byte chars and b) define all the addresses currently represented as `0xff 0xff 0xff 0xff`
-
-```
-0x48 0xff 0xf2
-0x48 0xff 0xf0
-0x48 0xff 0xf6
-0xc7 0xc2 0xf8 0x03 0x00 0x00
-0x8a 0x06
-0x80 0xf8 0x00
-0x0f 0x84 0xff 0xff 0xff 0xff
-0x66 0xee
-0x48 0xff 0xc6
-0xe9 0xff 0xff 0xff 0xff
-0x48 0x8f 0xc6
-0x48 0x8f 0xc0
-0x48 0x8f 0xc2
-0xc3
-0x48 0xff 0xf6
-0xc7 0xc6 0xff 0xff 0xff 0xff
-0xe8 0xff 0xff 0xff 0xff
-0x48 0x8f 0xc6
-0xe9 0xff 0xff 0xff 0xff
-0x48 0xff 0xf6
-0xc7 0xc6 0xff 0xff 0xff 0xff
-0xe8 0xff 0xff 0xff 0xff
-0x48 0x8f 0xc6
-0xe9 0xff 0xff 0xff 0xff
-0x48 0xff 0xf6
-0xc7 0xc6 0xff 0xff 0xff 0xff
-0xe8 0xff 0xff 0xff 0xff
-0x48 0x8f 0xc6
-0xe9 0xff 0xff 0xff 0xff
-0x48 0xff 0xf6
-0xc7 0xc6 0xff 0xff 0xff 0xff
-0xe8 0xff 0xff 0xff 0xff
-0x48 0x8f 0xc6
-0xe9 0xff 0xff 0xff 0xff
-0x1b 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-0x5b 0x33 0x36 0x6d 0x00 0x00 0x00 0x00
-0x7b 0x44 0x45 0x42 0x55 0x47 0x5d 0x3a
-0x1b 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-0x5b 0x30 0x6d 0x00 0x00 0x00 0x00 0x00
-0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-0x1b 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-0x5b 0x31 0x3b 0x33 0x31 0x6d 0x00 0x00
-0x7b 0x45 0x52 0x52 0x4f 0x52 0x5d 0x3a
-0x1b 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-0x5b 0x30 0x6d 0x00 0x00 0x00 0x00 0x00
-0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-0x1b 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-0x5b 0x31 0x3b 0x33 0x33 0x6d 0x00 0x00
-0x5b 0x54 0x45 0x53 0x54 0x5d 0x3a 0x20
-0x1b 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-0x5b 0x30 0x6d 0x00 0x00 0x00 0x00 0x00
-0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-0x1b 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-0x5b 0x31 0x3b 0x33 0x35 0x6d 0x00 0x00
-0x5b 0x57 0x41 0x52 0x4e 0x5d 0x3a 0x20
-0x1b 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-0x5b 0x30 0x6d 0x00 0x00 0x00 0x00 0x00
-0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
-```

twasm/asm/main.asm

@@ -63,6 +63,9 @@ start:
   pop rdi ; rdi = number of tokens in token table
   call assemble
 
+  mov rsi, msg_welcome_1
+  call OUTPUT_ADDR
+
   jmp halt
 
 ; ------------------------------------------------------------------------------
@@ -120,6 +123,8 @@ assemble:
     call print.debug
     pop rsi
 
+    mov esi, [.buffer_pointer]
+    sub esi, .buffer
     mov esi, [.next_output_byte]
     sub esi, OUTPUT_ADDR ; esi = relative address of label to start of program
     and edi, 0x0FFF      ; edi = index to add address hash to
@@ -207,7 +212,8 @@ assemble:
     call .get_next_tte
     mov rcx, rdi ; cx = operand tte
     pop rdi
-    push rcx     ; pushes until after write_prefix_continue
+    push rcx     ; pushes until after write_prefix_continue or before
+                 ; plusr_continue
 
                  ; di = tte of operator
     mov sil, 0   ; dst=r/m
@@ -222,9 +228,40 @@ assemble:
   .operator_1_get_opcode_label:
     mov sil, 4 ; dst=rel
   .operator_1_get_opcode_continue:
+    push rdi
+    push rbx
     call get_opcode
     ; al = opcode
     ; dl = op flag
+    pop rbx ; flags
+    pop rdi ; di
+
+    push rdx
+    and dl, 0x20
+    cmp dl, 0x20
+    pop rdx
+    jne .operator_1_plusr_continue
+
+    mov sil, 1 ; dst=r
+    ; TODO only do this if we've confirmed the next token's a register xD
+    call .clear_write_buffer
+    call get_opcode
+    ; al = opcode
+    ; dl = op flag
+    pop rdi ; di = next tte. from early in .operator_1
+    push rax
+    ; di = tte
+    call get_tte_typed_metadata
+    ; al = register typed metadata
+    mov ecx, eax ; cl = register typed metadata
+    shr ecx, 2
+    and ecx, 111b ; cl = reg value
+    pop rax ; al = opcode
+    add eax, ecx ; opcode + r
+    call .write_byte
+
+    jmp .loop_next_token
+  .operator_1_plusr_continue:
     push rdx
     and dl, 1000b
     cmp dl, 1000b
@@ -237,8 +274,7 @@ assemble:
   .operator_1_write_prefix_continue
     call .write_byte
     pop rdx ; dl = op flag
-
-    pop rdi ; di = next tte
+    pop rdi ; di = next tte. from early in .operator_1
 
     push rdi
     and di, 0xFF00
@@ -315,7 +351,9 @@ assemble:
     call print.debug
     pop rsi
 
-    mov esi, [.next_output_byte]
+    mov esi, [.buffer_pointer]
+    sub esi, .buffer
+    add esi, [.next_output_byte]
     sub esi, OUTPUT_ADDR ; esi = relative address of label reference to start
                          ;       of program
     mov eax, 0x04        ; al = first 4 bits: # bytes reserved
@@ -593,7 +631,9 @@ assemble:
 
     jmp .operator_2_register_label_32
   .operator_2_register_label_8:
-    mov esi, [.next_output_byte]
+    mov esi, [.buffer_pointer]
+    sub esi, .buffer
+    add esi, [.next_output_byte]
     sub esi, OUTPUT_ADDR ; esi = relative address of label reference to start
     mov eax, 0x11        ; al = first 4 bits: # bits reserved
                          ;      5th bit:      abs flag
@@ -603,7 +643,9 @@ assemble:
     call .write_byte
     jmp .loop_next_token
   .operator_2_register_label_16:
-    mov esi, [.next_output_byte]
+    mov esi, [.buffer_pointer]
+    sub esi, .buffer
+    add esi, [.next_output_byte]
     sub esi, OUTPUT_ADDR ; esi = relative address of label reference to start
     mov eax, 0x12        ; al = first 4 bits: # bits reserved
                          ;      5th bit:      abs flag
@@ -614,7 +656,9 @@ assemble:
     call .write_byte
     jmp .loop_next_token
   .operator_2_register_label_32:
-    mov esi, [.next_output_byte]
+    mov esi, [.buffer_pointer]
+    sub esi, .buffer
+    add esi, [.next_output_byte]
     sub esi, OUTPUT_ADDR ; esi = relative address of label reference to start
     mov eax, 0x14        ; al = first 4 bits: # bits reserved
                          ;      5th bit:      abs flag
@@ -941,7 +985,7 @@ assemble:
     call print.debug
     pop rsi
     mov rsp, rbp
-    ret
+    jmp resolve_awaiting_labels ; tail call
 
   .unexpected_token:
     call .flush_write_buffer
@@ -987,19 +1031,8 @@ assemble:
   .tokens_processed dd 0
   .tokens_total dd 0
 
-  ; al = byte to write
-  .output_byte:
-    push rdx
-    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
-    pop rdx
-    ret
-  .next_output_byte dd OUTPUT_ADDR ; next empty byte in output
-                                   ; TODO get rid of this sketchy bit of state
-
   ; al = byte to push
+  ; buffered push
   .push_byte:
     push rcx
     mov ecx, [.buffer_pointer]
@@ -1024,6 +1057,7 @@ assemble:
     ret
 
   ; al = byte to write
+  ; buffered write
   .write_byte:
     push rdx
     mov edx, [.buffer_pointer]
@@ -1033,6 +1067,16 @@ assemble:
     pop rdx
     ret
 
+  .clear_write_buffer:
+    push rax
+    xor eax, eax
+    mov [.buffer], rax
+    mov [.buffer + 8], rax
+    mov eax, .buffer
+    mov [.buffer_pointer], eax
+    pop rax
+    ret
+
   .flush_write_buffer:
     push rcx
     push rax
@@ -1051,10 +1095,23 @@ assemble:
       pop rax
       pop rcx
       ret
-  .buffer dq 0, 0            ; octo word of space for max of 8 bytes per write
+  .buffer dq 0, 0            ; octo word of space for max of 16 bytes per write
   .buffer_end:
   .buffer_pointer dd .buffer ; points to current byte in buffer
 
+  ; al = byte to write
+  ; unbuffered output; prefer `.write_byte`
+  .output_byte:
+    push rdx
+    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
+    pop rdx
+    ret
+  .next_output_byte dd OUTPUT_ADDR ; next empty byte in output
+
+
   .msg_unexpected_token db "unexpected token, aborting", 0x0A, 0x00
   .msg_unsupported_memory_access db "unsupported memory access, aborting", 0x0A, 0x00
   .msg_size_mismatch db "size mismatch, aborting", 0x0A, 0x00
@@ -1075,6 +1132,64 @@ assemble:
   .msg_operator_2_register_const db "operator_2_register_const", 0x0A, 0x00
   .msg_pseudo_operator db "pseudo_operator", 0x0A, 0x00
 
+; ------------------------------------------------------------------------------
+; resolve_awaiting_labels
+;
+; description:
+; resolves placeholder references in the program at OUTPUT_ADDR using the
+; label table and awaiting label table
+; ------------------------------------------------------------------------------
+
+resolve_awaiting_labels:
+  mov rbp, rsp
+  xor eax, eax
+  .loop:
+    shl rax, 4
+    mov rdi, [AWAITING_LABEL_TABLE_ADDR + rax]
+    ; rdi = raxth hash in awaiting label table
+
+    cmp rdi, 0
+    je .break
+
+    push rax
+    ; rdi = raxth hash in awaiting label table
+    call add_label_hash
+    ; rax = index in label table
+    shl rax, 4
+    mov edi, [LABEL_TABLE_ADDR + rax + 8] ; edi = address of label
+    pop rax ; rax = count
+
+    mov esi, [AWAITING_LABEL_TABLE_ADDR + rax + 8] ; esi = address of reference
+
+    sub edi, esi ; edi = offset
+    ; also sub length of reference TODO handle non-4-length refs 
+    sub edi, 4
+
+    mov dl, [AWAITING_LABEL_TABLE_ADDR + rax + 12] ; dl = flags
+
+    ; make sure it's 4 byte reference
+    push rdx
+    and dl, 0x0F
+    cmp dl, 4
+    pop rdx
+    jne .break
+
+    ; make sure it's relative reference
+    push rdx
+    and dl, 0x10
+    cmp dl, 0x00
+    pop rdx
+    jne .break
+
+    mov [OUTPUT_ADDR + rsi], edi
+
+    shr rax, 4
+    inc rax
+    jmp .loop
+  .break:
+    mov rsp, rbp
+    ret
+
 ; ------------------------------------------------------------------------------
 ; get_tte_type
 ;
@@ -1252,13 +1367,13 @@ get_ModRM:
 ; returned:
 ; al = opcode; the rest of rax is zeroed.
 ; dl = flagsX
-; +----------------------------------------------------+
+; +--------------------------------------------------------------+
 ; | flagsX byte                                                  |
-; +----------+-----------+-------------+---------------+
-; | 7      5 |     4     |      3      | 2           0 |
-; +----------+-----------+-------------+---------------+
-; | reserved | no ModR/M | 0x0F prefix | operator flag |
-; +----------+-----------+-------------+---------------+
+; +----------+---------+-----------+-------------+---------------+
+; | 7      6 |       5 |         4 |           3 | 2           0 |
+; +----------+---------+-----------+-------------+---------------+
+; | reserved | +r flag | no ModR/M | 0x0F prefix | operator flag |
+; +----------+---------+-----------+-------------+---------------+
 ;
 ; ------------------------------------------------------------------------------
 
@@ -2149,25 +2264,6 @@ print:
   .test_msg db 0x1B, "[1;33m", "[TEST]: ", 0x1B, "[0m", 0x00
   .warn_msg db 0x1B, "[1;35m", "[WARN]: ", 0x1B, "[0m", 0x00
 
-; ------------------------------------------------------------------------------
-; print_least_4_bits
-;
-; description:
-; prints the least significant 4 bits of rax for debugging reasons
-; ------------------------------------------------------------------------------
-
-print_least_4_bits:
-  push rax
-  push rsi
-  add eax, 0x30
-  mov [.byte], al
-  mov rsi, .byte
-  call print.warn
-  pop rsi
-  pop rax
-  ret
-  .byte db 0x00, 0x0A, 0x00
-
 ; ------------------------------------------------------------------------------
 ; halt
 ;
@@ -2392,7 +2488,7 @@ add_awaiting_label:
     mov rdx, [LABEL_TABLE_ADDR + rdi]
     mov [AWAITING_LABEL_TABLE_ADDR + rax], rdx ; hash
     mov [AWAITING_LABEL_TABLE_ADDR + 8 + rax], esi ; address
-    mov [AWAITING_LABEL_TABLE_ADDR + 13 + rax], cl ; bits passed to al
+    mov [AWAITING_LABEL_TABLE_ADDR + 12 + rax], cl ; bits passed to al
     ret
 
 ; ------------------------------------------------------------------------------
@@ -3119,30 +3215,30 @@ opcodes:
     dw 0x0000
 
     ; push
-    ; TODO add support for the +r variation
     dw 0x005F
     db 0xFF       ; r/m
-    db 0x00
+    db 0x50       ; +r
 
     db 0x68       ; imm16/32
     db 0x6A       ; imm8
     dw 0x0000
 
-    db 0x06       ; 6: r/m
+    db 0x26       ; 26: +r flag, r/m
     db 0x00
     dw 0x0000
 
     dd 0x00000000
 
     ; pop
-    ; TODO add support for the +r variation
     dw 0x0060
     db 0x8F       ; r/m
+    db 0x58       ; +r
+
+    dd 0x00000000
+
+    db 0x20       ; 20: +r flag, r/m
     db 0x00
-
-    dd 0x00000000
-
-    dd 0x00000000
+    dw 0x0000
 
     dd 0x00000000
 
@@ -3162,12 +3258,12 @@ opcodes:
   .by_id_end:
 
 msg_welcome db 0x1B, "[35m", "Welcome to Twasm", 0x1B, "[0m", 0x0A, 0x00
+msg_welcome_1 db 0x1B, "[35m", "Welcome to Twasm (from the next level)", 0x1B, "[0m", 0x0A, 0x00
 msg_halt db "halted.", 0x0A, 0x00
 
 whitespace_2 db " ", 0x0D
 
 ; test program
-align 128
 program:
   db "print:", 0x0A
   db "  push rdx", 0x0A