New target: C64 with SuperCPU in native mode

doc/api/target-platforms.md

@@ -9,7 +9,9 @@ The following platforms are currently supported:
 
 * `c64` – Commodore 64
 
-* `c64_scpu` – Commodore 64 with SuperCPU (very buggy)
+* `c64_scpu` – Commodore 64 with SuperCPU in emulation mode
+
+* `c64_scpu16` – Commodore 64 with SuperCPU in native, 16-bit mode (very buggy)
 
 * `c16` – Commodore 16
 

include/c64_scpu16.ini

@@ -0,0 +1,22 @@
+; Commodore 64 with a SuperCPU, running in native mode
+; assuming a program loaded from disk or tape
+; This has lower compatibility with inline assembly, but it's clearly faster.
+
+[compilation]
+arch=65816
+modules=c64_hardware,loader_0801_16bit,c64_kernal,c64_panic,stdlib
+emit_65816=native
+
+[allocation]
+main_org=$811
+zp_pointers=$C1,$C3,$FB,$FD,$39,$3B,$3D,$43,$45,$47,$4B
+himem_style=per_bank
+himem_start=after_code
+himem_end=$9FFF
+
+[output]
+style=per_bank
+format=startaddr,allocated
+extension=prg
+
+

include/loader_0801_16bit.mfk

@@ -0,0 +1,21 @@
+array _basic_loader @$801 = [
+    $0b,
+    $08,
+    10,
+    0,
+    $9e,
+    $32,
+    $30,
+    $36,
+    $31,
+    0,
+    0,
+    0
+    ]
+
+
+asm void __init_16bit() @$80D {
+    clc
+    xce
+    sep #$30
+}

src/main/scala/millfork/parser/MfParser.scala

@@ -486,7 +486,7 @@ case class MfParser(filename: String, input: String, currentDirectory: String, o
             case _ => false
           }) ErrorReporting.warn("Assembly non-interrupt function `$name` contains RTI, did you mean RTS?", options, Some(p))
         }
-        if (!flags("macro")) {
+        if (!name.startsWith("__") && !flags("macro")) {
           xs.last match {
             case AssemblyStatement(Opcode.RTS, _, _, _) => () // OK
             case AssemblyStatement(Opcode.RTI, _, _, _) => () // OK