Add data_segment feature, refactor

13 files changed, 76 insertions, 127 deletions

diff --git a/atk16_asm/asm_pass0.py b/atk16_asm/asm_pass0.py
index b30ddc0..270dc7a 100644
--- a/atk16_asm/asm_pass0.py
+++ b/atk16_asm/asm_pass0.py
@@ -1,5 +1,6 @@
 from dataclasses import dataclass
 import os.path
+from typing import Callable
 from .asm_ops import *
 from .asm_eval import *
 from .tokenizer import *
@@ -14,10 +15,38 @@ class Result0Line:
 class Result0:
   lines: list[Result0Line]
 
-def pass_0(lines: list[str], file_name: str) -> Result0:
+InsertAtDataSegment = Callable[[str, str, str, int], None]
+
+def pass_0(lines: list[str], file_name: str, insert_at_data_segment: InsertAtDataSegment | None = None) -> Result0:
   file_name = file_name if file_name.endswith(".atk16") else file_name + ".atk16"
   result_lines: list[Result0Line] = []
 
+  if not insert_at_data_segment:
+    def _insert_at_data_segment(label: str, value: str, context_file_name: str, context_line_num: int):
+      data_segment_index = None
+      for idx, result_line in enumerate(result_lines):
+        if result_line.line == "%%data_segment":
+          data_segment_index = idx
+          break
+
+      if data_segment_index is None:
+        raise Exception(f"Error: @data before @data_segment in {context_file_name}:{context_line_num}")
+
+      # Inverted order because of insert semantics
+      result_lines.insert(data_segment_index + 1, Result0Line(
+        src_file=context_file_name,
+        line_num=context_line_num,
+        line=f"${{{value}}}"
+      ))
+
+      result_lines.insert(data_segment_index + 1, Result0Line(
+        src_file=context_file_name,
+        line_num=context_line_num,
+        line=f"@label {label}"
+      ))
+
+    insert_at_data_segment = _insert_at_data_segment
+
   for (line_num, line) in enumerate(lines):
     line_num += 1 # line numbers are 1-based
 
@@ -27,6 +56,23 @@ def pass_0(lines: list[str], file_name: str) -> Result0:
     keyword, *args = tokenize(line)
 
     match keyword:
+      case "@data_segment":
+        data_segment_sentinel = "%%data_segment"
+        result_lines.append(Result0Line(
+          src_file=file_name,
+          line_num=line_num,
+          line=data_segment_sentinel
+        ))
+
+      case "@data":
+        if len(args) != 2:
+          raise Exception(f"Error: @data missing arguments in {file_name}:{line_num}")
+
+        name = args[0]
+        data = args[1]
+
+        insert_at_data_segment(name, data, file_name, line_num)
+
       case "@include":
         asm_file_name = args[0] if args[0].endswith(".atk16") else args[0] + ".atk16"
         # if starts with %, use the path relative to this source file, not the current working directory
@@ -40,7 +86,7 @@ def pass_0(lines: list[str], file_name: str) -> Result0:
         with open(path, "r") as f:
           incl_lines = f.readlines()
 
-        incl_result0 = pass_0(incl_lines, asm_file_name)
+        incl_result0 = pass_0(incl_lines, asm_file_name, insert_at_data_segment)
         for incl_line in incl_result0.lines:
           result_lines.append(Result0Line(
             src_file=incl_line.src_file,
diff --git a/atk16_asm/asm_pass1.py b/atk16_asm/asm_pass1.py
index 4786afa..d192b5e 100644
--- a/atk16_asm/asm_pass1.py
+++ b/atk16_asm/asm_pass1.py
@@ -35,6 +35,9 @@ def pass_1(result0: Result0) -> Result1:
           expansion = mod_expansions[op]
           if ops == "*" or op in ops_split:
             operations[op] = expansion
+      case "%%data_segment":
+        # drop the data_segment sentinel
+        continue
       case _:
         result_lines.append(Result1Line(
           src_file=line.src_file,
diff --git a/atk16_asm/builtin_asm/bootstrap.atk16 b/atk16_asm/builtin_asm/bootstrap.atk16
index 28d8aa1..b3b4254 100644
--- a/atk16_asm/builtin_asm/bootstrap.atk16
+++ b/atk16_asm/builtin_asm/bootstrap.atk16
@@ -31,10 +31,10 @@
 @let vt_text_mem     0x19 ; Text memory buffer address
 
 @address vector_table
-  hlt_isr        ; 0x10
-  hlt_isr        ; 0x11
-  hlt_isr        ; 0x12
-  hlt_isr        ; 0x13
+  no_op_isr      ; 0x10
+  no_op_isr      ; 0x11
+  no_op_isr      ; 0x12
+  no_op_isr      ; 0x13
   stack_segment  ; 0x14
   terminal_addr  ; 0x15
   keyboard_addr  ; 0x16
@@ -42,9 +42,13 @@
   sprite_mem     ; 0x18
   text_mem       ; 0x19
 
-@label hlt_isr
-  ldi 0x55 RE
-  hlt
+@label no_op_isr
+  rti
+
+; Define data segment location to this address
+; Programs can use `@data label value` to inject data here
+@label data_segment
+@data_segment
 
 ; Example ISR for keyboard to terminal
 ; @label keyboard_to_terminal_isr
diff --git a/atk16_asm/builtin_asm/std_bump_alloc.atk16 b/atk16_asm/builtin_asm/std_bump_alloc.atk16
index 1d87435..32516bc 100644
--- a/atk16_asm/builtin_asm/std_bump_alloc.atk16
+++ b/atk16_asm/builtin_asm/std_bump_alloc.atk16
@@ -21,10 +21,8 @@
 ; The alloc_next value is a pointer to the next free word in the heap.
 ; It is stored in the first word of the heap.
 
-@label bump_next_p
-  heap_segment
-@label bump_arena_start
-  ${heap_segment - 1}
+@data bump_next_p heap_segment
+@data bump_arena_start ${heap_segment - 1}
 
 @label bump_reset
   ; reset the heap to its initial state (zero allocated blocks)
diff --git a/atk16_asm/builtin_asm/std_math.atk16 b/atk16_asm/builtin_asm/std_math.atk16
index 58b9cdb..6544b32 100644
--- a/atk16_asm/builtin_asm/std_math.atk16
+++ b/atk16_asm/builtin_asm/std_math.atk16
@@ -1,7 +1,5 @@
-@label mul_sign_mask
-  ${2 ** 15}
-@label mul_not_mask
-  0xFFFF
+@data mul_sign_mask ${2 ** 15}
+@data mul_not_mask 0xFFFF
 
 @label mul
   ; multiply a signed (a) and a signed (b)
diff --git a/resources/asm/text_mode.atk16 b/resources/asm/text_mode.atk16
index cab9562..1761361 100644
--- a/resources/asm/text_mode.atk16
+++ b/resources/asm/text_mode.atk16
@@ -3,8 +3,7 @@
 @include %bootstrap
 @include %std_mem
 
-@label text_buffer_size
-  2048
+@data text_buffer_size 2048
 
 @label main
   ; use memset to clear the text buffer
diff --git a/test/e2e/call_stack/lib.atk16 b/test/e2e/call_stack/lib.atk16
deleted file mode 100644
index 58b9cdb..0000000
--- a/test/e2e/call_stack/lib.atk16
+++ /dev/null
@@ -1,92 +0,0 @@
-@label mul_sign_mask
-  ${2 ** 15}
-@label mul_not_mask
-  0xFFFF
-
-@label mul
-  ; multiply a signed (a) and a signed (b)
-  ; parameters RA = a
-  ;            RB = b
-  ; return     RG = a * b
-
-  stack_stash RA RB RC RD
-  ldi 0 RG                ; initialize return value in RG to 0
-
-  ldi mul_sign_mask RD
-  ldr RD RD               ; RD := 2 ** 16
-  and RA RD RC            ; RC := 2 ** 16 if a is negative, 0 otherwise
-  and RB RD RD            ; RD := 2 ** 16 if b is negative, 0 otherwise
-  xor RC RD RC            ; RC := 2 ** 16 if exactly one of a or b is negative, 0 otherwise
-
-  ldi mul_not_mask RD
-  ldr RD RD               ; RD := 0xFFFF
-
-  ; RA := abs(a)
-  subi RA 0 RA
-  bri sign mul_a_negative
-  jpi mul_a_negative_done
-@label mul_a_negative
-  xor RA RD RA            ; RA := ~a
-  inc RA                  ; negate twos complement a
-@label mul_a_negative_done
-
-  ; RB := abs(b)
-  subi RB 0 RB
-  bri sign mul_b_negative
-  jpi mul_b_negative_done
-@label mul_b_negative
-  xor RB RD RB            ; RB := ~b
-  inc RB                  ; negate twos complement b
-@label mul_b_negative_done
-
-  ; the algorithms iterates b times, adding a to RG each time
-  ; so b should be as small as possible
-  sub RB RA RD            ; if a <= b, swap a and b
-  bri sign mul_loop
-
-  mov RA RD
-  mov RB RA
-  mov RD RB
-@label mul_loop
-  ; if b == 0, we are done
-  subi RB 0 RB
-  bri zero mul_done
-  ; RG += RA
-  add RA RG RG
-  dec RB
-  jpi mul_loop
-@label mul_done
-  ; set the most significant bit to the precomputed sign
-  subi RC 0 RC
-  bri sign mul_negate_result
-  jpi mul_return
-@label mul_negate_result
-  ldi mul_not_mask RD
-  ldr RD RD               ; RD := 0xFFFF
-  xor RG RD RG
-  inc RG
-@label mul_return
-  stack_restore RA RB RC RD
-  return
-
-; Recursive impl of the factorial
-; Calling convention is that arguments are in registers RA..RG (max 7 arguments since RH = SP)
-; and the return value is in RG (note: overwriting arg in RG)
-
-@label fact
-  ; parameters RA = n
-  ; return     RG = factorial(n)
-  subi RA 0 RA
-  bri zero fact_basecase
-@label fact_reccase
-  stack_stash RA RB
-  mov RA RB             ; RB := n
-  subi RA 1 RA          ; RA := n - 1
-  calli fact            ; RG := fact(n - 1)
-  mov RG RA             ; RA := fact(n - 1)
-  calli mul             ; RG := n * fact(n - 1)
-  stack_restore RA RB
-  return
-@label fact_basecase
-  ldi 1 RG
-  return
diff --git a/test/e2e/call_stack/run_fact.atk16 b/test/e2e/call_stack/run_fact.atk16
index e0b8bf1..9a282ff 100644
--- a/test/e2e/call_stack/run_fact.atk16
+++ b/test/e2e/call_stack/run_fact.atk16
@@ -1,5 +1,5 @@
 @include %bootstrap
-@include lib
+@include %std_math
 
 @let INPUT %INPUT% ; value replaced by test driver code
 
diff --git a/test/e2e/call_stack/run_mul.atk16 b/test/e2e/call_stack/run_mul.atk16
index 377e61c..f47e19b 100644
--- a/test/e2e/call_stack/run_mul.atk16
+++ b/test/e2e/call_stack/run_mul.atk16
@@ -1,11 +1,9 @@
 @include %bootstrap
-@include lib
+@include %std_math
 
 ; values replaced by test driver code
-@label INPUT_A
-  %INPUT_A%
-@label INPUT_B
-  %INPUT_B%
+@data INPUT_A %INPUT_A%
+@data INPUT_B %INPUT_B%
 
 @label main
   ldi INPUT_A RA
diff --git a/test/e2e/fibo/fibo.atk16 b/test/e2e/fibo/fibo.atk16
index a817a1c..5b9030b 100644
--- a/test/e2e/fibo/fibo.atk16
+++ b/test/e2e/fibo/fibo.atk16
@@ -1,7 +1,6 @@
 @include %bootstrap
 
-@label iters
-  51
+@data iters 51
 
 @label main
   ldi iters RA
diff --git a/test/e2e/std_bump_alloc/run_bump_alloc_overlap.atk16 b/test/e2e/std_bump_alloc/run_bump_alloc_overlap.atk16
index 8dcc6fa..08ced49 100644
--- a/test/e2e/std_bump_alloc/run_bump_alloc_overlap.atk16
+++ b/test/e2e/std_bump_alloc/run_bump_alloc_overlap.atk16
@@ -1,8 +1,7 @@
 @include %bootstrap
 @include %std_bump_alloc
 
-@label n
-  0x8000
+@data n 0x8000
 
 @label main
   calli bump_reset
diff --git a/test/e2e/std_mem/run_memcopy.atk16 b/test/e2e/std_mem/run_memcopy.atk16
index 66f3c1d..a011cb9 100644
--- a/test/e2e/std_mem/run_memcopy.atk16
+++ b/test/e2e/std_mem/run_memcopy.atk16
@@ -1,10 +1,8 @@
 @include %bootstrap
 @include %std_mem
 
-@label mem_offset_from
-  0x9000
-@label mem_offset_to
-  0x9100
+@data mem_offset_from 0x9000
+@data mem_offset_to 0x9100
 
 @label main
   ; set values 0x0000..0x0004 to 3
diff --git a/test/e2e/std_mem/run_memset.atk16 b/test/e2e/std_mem/run_memset.atk16
index 07e7666..b6a012c 100644
--- a/test/e2e/std_mem/run_memset.atk16
+++ b/test/e2e/std_mem/run_memset.atk16
@@ -1,8 +1,7 @@
 @include %bootstrap
 @include %std_mem
 
-@label mem_offset
-  0x9000
+@data mem_offset 0x9000
 
 @label main
   ; set 0x9005 to value 4