cx16: set rom bank to 0 at startup (for faster kernal API calls)

cx16: callfar() with constant address generates shorter asm

codeGenCpu6502/src/prog8/codegen/cpu6502/BuiltinFunctionsAsmGen.kt

@@ -284,15 +284,26 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         if(asmgen.options.compTarget.name != "cx16")
             throw AssemblyError("callfar only works on cx16 target at this time")
 
-        asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)      // bank
-        asmgen.out("  sta  (++)+0")
-        asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AY)     // jump address
-        asmgen.out("  sta  (+)+0 |  sty  (+)+1")
-        asmgen.assignExpressionToRegister(fcall.args[2], RegisterOrPair.AY)     // uword argument
-        asmgen.out("""
-            jsr  cx16.JSRFAR
-+           .word  0
-+           .byte  0""")
+        val constBank = fcall.args[0].asConstInteger()
+        val constAddress = fcall.args[1].asConstInteger()
+        if(constBank!=null && constAddress!=null) {
+            asmgen.assignExpressionToRegister(fcall.args[2], RegisterOrPair.AY)     // uword argument
+            asmgen.out("""
+                jsr  cx16.JSRFAR
+                .word  ${constAddress.toHex()}
+                .byte  $constBank""")
+        } else {
+            asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)      // bank
+            asmgen.out("  sta  (++)+0")
+            asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AY)     // jump address
+            asmgen.out("  sta  (+)+0 |  sty  (+)+1")
+            asmgen.assignExpressionToRegister(fcall.args[2], RegisterOrPair.AY)     // uword argument
+            asmgen.out("""
+                jsr  cx16.JSRFAR
++               .word  0
++               .byte  0""")
+        }
+
         // note that by convention the values in A+Y registers are now the return value of the call.
         if(resultRegister!=null)  {
             assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister, false, fcall.position, null, asmgen), RegisterOrPair.AY)

compiler/res/prog8lib/cx16/syslib.p8

@@ -483,7 +483,7 @@ romsub $C006 = x16edit_loadfile_options(ubyte firstbank @X, ubyte lastbank @Y, s
                 uword filenameLengthAndOptions @R1, uword tabstopAndWordwrap @R2,
                 uword disknumberAndColors @R3, uword headerAndStatusColors @R4) clobbers(A,X,Y)
 
-; Audio (rom bank 10 - you have to activate this bank manually first! Or use callfar/JSRFAR.)
+; Audio (rom bank 10 - you have to activate this bank manually first! Or use the stubs in the audio module. Or use callfar/JSRFAR manually.)
 romsub $C09F = audio_init() clobbers(A,X,Y) -> bool @Pc     ; (re)initialize both vera PSG and YM audio chips
 romsub $C000 = bas_fmfreq(ubyte channel @A, uword freq @XY, bool noretrigger @Pc) clobbers(A,X,Y) -> bool @Pc
 romsub $C003 = bas_fmnote(ubyte channel @A, ubyte note @X, ubyte fracsemitone @Y, bool noretrigger @Pc) clobbers(A,X,Y) -> bool @Pc
@@ -1397,6 +1397,7 @@ asmsub  init_system()  {
         lda  #PROG8_VARSHIGH_RAMBANK
         sta  $00    ; select ram bank
         lda  #0
+        sta  $01    ; set ROM bank to kernal bank to speed up kernal calls
         tax
         tay
         cli
@@ -1413,6 +1414,7 @@ asmsub  init_system_phase2()  {
         sta  restore_irq._orig_irqvec+1
         lda  #PROG8_VARSHIGH_RAMBANK
         sta  $00    ; select ram bank
+        stz  $01    ; set ROM bank to kernal bank to speed up kernal calls
         cli
         cld
         clc

docs/source/todo.rst

@@ -1,9 +1,6 @@
 TODO
 ====
 
-Commit "tweak program start initialization and fix cleanup at exit for atari and pet compiler targets #04cb684f"  causing programs to crash (assem, rockrunner)
-
-
 Improve register load order in subroutine call args assignments:
 in certain situations, the "wrong" order of evaluation of function call arguments is done which results
 in overwriting registers that already got their value, which requires a lot of stack juggling (especially on plain 6502 cpu!)

examples/test.p8

@@ -3,12 +3,12 @@
 %option no_sysinit
 
 main {
-    ubyte @shared @nozp value1 = 99
-    ubyte @shared @requirezp value2 = 42
     sub start() {
-        txt.print_ub(value1)
-        txt.nl()
-        txt.print_ub(value2)
-        txt.nl()
+        uword @shared address = $C09F
+        ubyte @shared bank = 10
+        uword @shared argument = $1234
+
+        void callfar(bank, address, argument)
+        void callfar(10, $C09F, argument)
     }
 }