3 files changed, 96 insertions, 4 deletions
diff --git a/atk16_asm/builtin_asm/bootstrap.atk16 b/atk16_asm/builtin_asm/bootstrap.atk16
index 94eed4c..e0fddfa 100644
--- a/atk16_asm/builtin_asm/bootstrap.atk16
+++ b/atk16_asm/builtin_asm/bootstrap.atk16
@@ -2,10 +2,10 @@
 
 ; memory segments
 @let stack_segment   0x8000
-@let mmio_segment    0xE000
-@let terminal_addr   0xE000
-@let keyboard_addr   0xE001
-@let gr_mode_addr    0xE002
+@let mmio_segment    0xE7F0
+@let terminal_addr   0xE7F0
+@let keyboard_addr   0xE7F1
+@let gr_mode_addr    0xE7F2
 @let sprite_mem      0xE800
 @let text_mem        0xF800 ; to 0xFFFF
 ; note: sprite memory & text memory can use the same space
diff --git a/atk16_asm/builtin_asm/std_alloc.atk16 b/atk16_asm/builtin_asm/std_alloc.atk16
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/atk16_asm/builtin_asm/std_alloc.atk16
diff --git a/atk16_asm/builtin_asm/std_math.atk16 b/atk16_asm/builtin_asm/std_math.atk16
new file mode 100644
index 0000000..58b9cdb
--- /dev/null
+++ b/atk16_asm/builtin_asm/std_math.atk16
@@ -0,0 +1,92 @@
+@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