Renamed stack-manipulating fragments. Added fragment-based stack clean-up. #316

src/main/fragment/mos6502-common/_stackpullbyte_1.asm

@@ -0,0 +1 @@
+pla

src/main/fragment/mos6502-common/_stackpullbyte_2.asm

@@ -0,0 +1,2 @@
+pla
+pla

src/main/fragment/mos6502-common/_stackpullbyte_3.asm

@@ -0,0 +1,3 @@
+pla
+pla
+pla

src/main/fragment/mos6502-common/_stackpullbyte_4.asm

@@ -0,0 +1,4 @@
+pla
+pla
+pla
+pla

src/main/fragment/mos6502-common/_stackpullbyte_5.asm

@@ -0,0 +1,5 @@
+pla
+pla
+pla
+pla
+pla

src/main/fragment/mos6502-common/_stackpullbyte_6.asm

@@ -0,0 +1,6 @@
+pla
+pla
+pla
+pla
+pla
+pla

src/main/fragment/mos6502-common/_stackpullbyte_7.asm

@@ -0,0 +1,7 @@
+pla
+pla
+pla
+pla
+pla
+pla
+pla

src/main/fragment/mos6502-common/_stackpullbyte_8.asm

@@ -0,0 +1,8 @@
+pla
+pla
+pla
+pla
+pla
+pla
+pla
+pla

src/main/fragment/mos6502-common/_stackpushword_=vwuc1.asm

@@ -0,0 +1,4 @@
+lda #>{c1}
+pha
+lda #<{c1}
+pha

src/main/fragment/mos6502-common/vwuz1=_stackgetword_vbuc1.asm

@@ -0,0 +1,5 @@
+tsx
+lda STACK_BASE+{c1},x
+sta {z1}
+lda STACK_BASE+{c1}+1,x
+sta {z1}+1

src/main/fragment/mos6502-undoc/_stackpullbyte_3.asm

@@ -0,0 +1,4 @@
+tsx
+txa
+axs #-3
+txs

src/main/fragment/mos6502-undoc/_stackpullbyte_4.asm

@@ -0,0 +1,4 @@
+tsx
+txa
+axs #-4
+txs

src/main/fragment/mos6502-undoc/_stackpullbyte_5.asm

@@ -0,0 +1,4 @@
+tsx
+txa
+axs #-5
+txs

src/main/fragment/mos6502-undoc/_stackpullbyte_6.asm

@@ -0,0 +1,4 @@
+tsx
+txa
+axs #-6
+txs

src/main/fragment/mos6502-undoc/_stackpullbyte_7.asm

@@ -0,0 +1,4 @@
+tsx
+txa
+axs #-7
+txs

src/main/fragment/mos6502-undoc/_stackpullbyte_8.asm

@@ -0,0 +1,4 @@
+tsx
+txa
+axs #-8
+txs

src/main/java/dk/camelot64/kickc/fragment/AsmFragmentInstanceSpecFactory.java

@@ -358,9 +358,9 @@ public class AsmFragmentInstanceSpecFactory {
          return name;
       } else if(value instanceof ParamStackValue) {
          ParamStackValue paramStackValue = (ParamStackValue) value;
-         return "_stack"+paramStackValue.getValueType().getTypeName()+"_"+bind(paramStackValue.getStackOffset());
+         return "_stackget"+paramStackValue.getValueType().getTypeName()+"_"+bind(paramStackValue.getStackOffset());
       } else if(value instanceof ParamStackPush) {
-         return "_push"+((ParamStackPush) value).getType().getTypeName()+"_";
+         return "_stackpush"+((ParamStackPush) value).getType().getTypeName()+"_";
       }
       throw new RuntimeException("Binding of value type not supported " + value.toString(program));
    }

src/main/java/dk/camelot64/kickc/passes/Pass4CodeGeneration.java

@@ -699,7 +699,7 @@ public class Pass4CodeGeneration {
          StatementAssignment assignment = (StatementAssignment) statement;
          AsmFragmentInstanceSpecFactory asmFragmentInstanceSpecFactory = new AsmFragmentInstanceSpecFactory(assignment, assignmentAlu, program);
          ensureEncoding(asm, asmFragmentInstanceSpecFactory);
-         generateAsm(asm, asmFragmentInstanceSpecFactory);
+         generateAsm(asm, asmFragmentInstanceSpecFactory.getAsmFragmentInstanceSpec());
          aluState.clear();
          return;
       }
@@ -725,13 +725,13 @@ public class Pass4CodeGeneration {
                } else {
                   AsmFragmentInstanceSpecFactory asmFragmentInstanceSpecFactory = new AsmFragmentInstanceSpecFactory(assignment, program);
                   ensureEncoding(asm, asmFragmentInstanceSpecFactory);
-                  generateAsm(asm, asmFragmentInstanceSpecFactory);
+                  generateAsm(asm, asmFragmentInstanceSpecFactory.getAsmFragmentInstanceSpec());
                }
             }
          } else if(statement instanceof StatementConditionalJump) {
             AsmFragmentInstanceSpecFactory asmFragmentInstanceSpecFactory = new AsmFragmentInstanceSpecFactory((StatementConditionalJump) statement, block, program, getGraph());
             ensureEncoding(asm, asmFragmentInstanceSpecFactory);
-            generateAsm(asm, asmFragmentInstanceSpecFactory);
+            generateAsm(asm, asmFragmentInstanceSpecFactory.getAsmFragmentInstanceSpec());
          } else if(statement instanceof StatementCall) {
             StatementCall call = (StatementCall) statement;
 
@@ -753,9 +753,12 @@ public class Pass4CodeGeneration {
                for(RValue parameter : call.getParameters()) {
                   SymbolType parameterType = SymbolTypeInference.inferType(program.getScope(), parameter);
                   AsmFragmentInstanceSpecFactory asmFragmentInstanceSpecFactory = new AsmFragmentInstanceSpecFactory(new ParamStackPush(parameterType), parameter, program, block.getScope());
+                  asm.startChunk(block.getScope(), statement.getIndex(), statement.toString(program, verboseAliveInfo));
                   ensureEncoding(asm, asmFragmentInstanceSpecFactory);
-                  generateAsm(asm, asmFragmentInstanceSpecFactory);
+                  generateAsm(asm, asmFragmentInstanceSpecFactory.getAsmFragmentInstanceSpec());
                }
+               asm.startChunk(block.getScope(), statement.getIndex(), statement.toString(program, verboseAliveInfo));
+               asm.getCurrentChunk().setFragment("jsr");
             }
             asm.addInstruction("jsr", AsmAddressingMode.ABS, call.getProcedure().getFullName(), false);
             // Clean up the stack
@@ -765,10 +768,10 @@ public class Pass4CodeGeneration {
                   SymbolType parameterType = SymbolTypeInference.inferType(program.getScope(), parameter);
                   parameterBytes += parameterType.getSizeBytes();
                }
-               // TODO: Replace with fragment - to allow hand-coded handling of the stack pointer modifications - eg. using TSX, TXA, AXS #{}, TXS
-               for(int i = 0; i < parameterBytes; i++) {
-                  asm.addInstruction("pla", AsmAddressingMode.NON, null, false);
-               }
+               String pullSignature = "_stackpullbyte_" + Integer.toString(parameterBytes);
+               AsmFragmentInstanceSpec pullFragmentInstanceSpec = new AsmFragmentInstanceSpec(program, pullSignature, new LinkedHashMap<>(), block.getScope());
+               asm.startChunk(block.getScope(), statement.getIndex(), statement.toString(program, verboseAliveInfo));
+               generateAsm(asm, pullFragmentInstanceSpec);
             }
          } else if(statement instanceof StatementReturn) {
             Procedure.InterruptType interruptType = null;
@@ -866,24 +869,23 @@ public class Pass4CodeGeneration {
     * Generate ASM code for an ASM fragment instance
     *
     * @param asm The ASM program to generate into
-    * @param asmFragmentInstanceSpecFactory The ASM fragment instance specification factory
+    * @param fragmentInstanceSpec The ASM fragment instance specification
     */
-   private void generateAsm(AsmProgram asm, AsmFragmentInstanceSpecFactory asmFragmentInstanceSpecFactory) {
-      String initialSignature = asmFragmentInstanceSpecFactory.getAsmFragmentInstanceSpec().getSignature();
-      AsmFragmentInstanceSpec asmFragmentInstanceSpec = asmFragmentInstanceSpecFactory.getAsmFragmentInstanceSpec();
-      AsmFragmentInstance asmFragmentInstance = null;
-      StringBuffer fragmentVariationsTried = new StringBuffer();
-      while(asmFragmentInstance == null) {
+   private void generateAsm(AsmProgram asm, AsmFragmentInstanceSpec fragmentInstanceSpec) {
+      String initialSignature = fragmentInstanceSpec.getSignature();
+      AsmFragmentInstance fragmentInstance = null;
+      StringBuilder fragmentVariationsTried = new StringBuilder();
+      while(fragmentInstance == null) {
          try {
-            asmFragmentInstance = program.getAsmFragmentSynthesizer().getFragmentInstance(asmFragmentInstanceSpec, program.getLog());
+            fragmentInstance = program.getAsmFragmentSynthesizer().getFragmentInstance(fragmentInstanceSpec, program.getLog());
          } catch(AsmFragmentTemplateSynthesizer.UnknownFragmentException e) {
             // Unknown fragment - keep looking through alternative ASM fragment instance specs until we have tried them all
-            String signature = asmFragmentInstanceSpec.getSignature();
+            String signature = fragmentInstanceSpec.getSignature();
             fragmentVariationsTried.append(signature).append(" ");
-            if(asmFragmentInstanceSpec.hasNextVariation()) {
-               asmFragmentInstanceSpec.nextVariation();
+            if(fragmentInstanceSpec.hasNextVariation()) {
+               fragmentInstanceSpec.nextVariation();
                if(program.getLog().isVerboseFragmentLog()) {
-                  program.getLog().append("Fragment not found " + signature + ". Attempting another variation " + asmFragmentInstanceSpec.getSignature());
+                  program.getLog().append("Fragment not found " + signature + ". Attempting another variation " + fragmentInstanceSpec.getSignature());
                }
             } else {
                // No more variations available - fail with an error
@@ -891,8 +893,8 @@ public class Pass4CodeGeneration {
             }
          }
       }
-      asm.getCurrentChunk().setFragment(asmFragmentInstance.getFragmentName());
-      asmFragmentInstance.generate(asm);
+      asm.getCurrentChunk().setFragment(fragmentInstance.getFragmentName());
+      fragmentInstance.generate(asm);
    }
 
    /**
@@ -1047,7 +1049,7 @@ public class Pass4CodeGeneration {
             } else {
                AsmFragmentInstanceSpecFactory asmFragmentInstanceSpecFactory = new AsmFragmentInstanceSpecFactory(lValue, rValue, program, scope);
                ensureEncoding(asm, asmFragmentInstanceSpecFactory);
-               generateAsm(asm, asmFragmentInstanceSpecFactory);
+               generateAsm(asm, asmFragmentInstanceSpecFactory.getAsmFragmentInstanceSpec());
             }
          }
          transitionSetGenerated(transition);

src/test/java/dk/camelot64/kickc/test/TestPrograms.java

@@ -35,12 +35,15 @@ public class TestPrograms {
    public TestPrograms() {
    }
 
-   /*
+   @Test
+   public void testProcedureCallingConventionStack2() throws IOException, URISyntaxException {
+      compileAndCompare("procedure-callingconvention-stack-2"); //, log().verboseCreateSsa().verboseParse().verboseStatementSequence());
+   }
+
    @Test
    public void testProcedureCallingConventionStack1() throws IOException, URISyntaxException {
-      compileAndCompare("procedure-callingconvention-stack-1", log());
+      compileAndCompare("procedure-callingconvention-stack-1");
    }
-   */
 
    @Test
    public void testProcedureCallingConventionStack0() throws IOException, URISyntaxException {

src/test/kc/procedure-callingconvention-stack-2.kc

@@ -0,0 +1,11 @@
+// Test a procedure with calling convention stack - and enough parameters to use fast ASM for cleaning stack
+
+const word* SCREEN = 0x0400;
+
+void main(void) {
+    SCREEN[0] = plus(0x1234, 0x2345);
+}
+
+word __stackcall plus(word a, word b) {
+    return a+b;
+}

src/test/ref/procedure-callingconvention-stack-0.log

@@ -181,12 +181,16 @@ bend:
   // main
 main: {
     .label _0 = 2
-    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _pushbyte_=vbuc1 
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7 
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpushbyte_=vbuc1 
     lda #'0'
     pha
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpushbyte_=vbuc1 
     lda #7
     pha
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- jsr 
     jsr plus
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpullbyte_2 
     pla
     pla
     // [6] *((const byte*) SCREEN#0) ← (byte~) main::$0 -- _deref_pbuc1=vbuz1 
@@ -206,11 +210,11 @@ plus: {
     .label a = 3
     .label b = 4
     .label return = 5
-    // [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A) -- vbuz1=_stackbyte_vbuc1 
+    // [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A) -- vbuz1=_stackgetbyte_vbuc1 
     tsx
     lda STACK_BASE+OFFSET_STACK_A,x
     sta.z a
-    // [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B) -- vbuz1=_stackbyte_vbuc1 
+    // [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B) -- vbuz1=_stackgetbyte_vbuc1 
     tsx
     lda STACK_BASE+OFFSET_STACK_B,x
     sta.z b
@@ -285,12 +289,16 @@ bend_from_b1:
 bend:
   // main
 main: {
-    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _pushbyte_=vbuc1 
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7 
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpushbyte_=vbuc1 
     lda #'0'
     pha
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpushbyte_=vbuc1 
     lda #7
     pha
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- jsr 
     jsr plus
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpullbyte_2 
     pla
     pla
     // [6] *((const byte*) SCREEN#0) ← (byte~) main::$0 -- _deref_pbuc1=vbuyy 
@@ -307,11 +315,11 @@ plus: {
     .const OFFSET_STACK_A = 0
     .const OFFSET_STACK_B = 1
     .label a = 2
-    // [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A) -- vbuz1=_stackbyte_vbuc1 
+    // [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A) -- vbuz1=_stackgetbyte_vbuc1 
     tsx
     lda STACK_BASE+OFFSET_STACK_A,x
     sta.z a
-    // [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B) -- vbuaa=_stackbyte_vbuc1 
+    // [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B) -- vbuaa=_stackgetbyte_vbuc1 
     tsx
     lda STACK_BASE+OFFSET_STACK_B,x
     // [10] (byte) plus::return#0 ← (byte) plus::a#0 + (byte) plus::b#0 -- vbuaa=vbuz1_plus_vbuaa 
@@ -395,12 +403,16 @@ Score: 60
   // main
 main: {
     // plus('0', 7)
-    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _pushbyte_=vbuc1 
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7 
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpushbyte_=vbuc1 
     lda #'0'
     pha
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpushbyte_=vbuc1 
     lda #7
     pha
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- jsr 
     jsr plus
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpullbyte_2 
     pla
     pla
     // SCREEN[0] = plus('0', 7)
@@ -417,11 +429,11 @@ plus: {
     .const OFFSET_STACK_A = 0
     .const OFFSET_STACK_B = 1
     .label a = 2
-    // [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A) -- vbuz1=_stackbyte_vbuc1 
+    // [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A) -- vbuz1=_stackgetbyte_vbuc1 
     tsx
     lda STACK_BASE+OFFSET_STACK_A,x
     sta.z a
-    // [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B) -- vbuaa=_stackbyte_vbuc1 
+    // [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B) -- vbuaa=_stackgetbyte_vbuc1 
     tsx
     lda STACK_BASE+OFFSET_STACK_B,x
     // return a+b;

src/test/ref/procedure-callingconvention-stack-1.asm

@@ -0,0 +1,31 @@
+// Test a procedure with calling convention stack
+.pc = $801 "Basic"
+:BasicUpstart(main)
+.pc = $80d "Program"
+  .label SCREEN = $400
+  .const STACK_BASE = $103
+main: {
+    lda #'0'
+    pha
+    lda #7
+    pha
+    jsr plus
+    pla
+    pla
+    sty SCREEN
+    rts
+}
+// plus(byte zeropage(2) a, byte register(A) b)
+plus: {
+    .const OFFSET_STACK_A = 0
+    .const OFFSET_STACK_B = 1
+    .label a = 2
+    tsx
+    lda STACK_BASE+OFFSET_STACK_A,x
+    sta.z a
+    tsx
+    lda STACK_BASE+OFFSET_STACK_B,x
+    clc
+    adc.z a
+    rts
+}

src/test/ref/procedure-callingconvention-stack-1.cfg

@@ -0,0 +1,29 @@
+@begin: scope:[]  from
+  [0] phi()
+  to:@1
+@1: scope:[]  from @begin
+  [1] phi()
+  [2] call main 
+  to:@end
+@end: scope:[]  from @1
+  [3] phi()
+
+(void()) main()
+main: scope:[main]  from @1
+  [4] phi()
+  [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7 
+  [6] *((const byte*) SCREEN#0) ← (byte~) main::$0
+  to:main::@return
+main::@return: scope:[main]  from main
+  [7] return 
+  to:@return
+
+__stackcall (byte()) plus((byte) plus::a , (byte) plus::b)
+plus: scope:[plus]  from
+  [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A)
+  [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B)
+  [10] (byte) plus::return#0 ← (byte) plus::a#0 + (byte) plus::b#0
+  to:plus::@return
+plus::@return: scope:[plus]  from plus
+  [11] return (byte) plus::return#0
+  to:@return

src/test/ref/procedure-callingconvention-stack-1.log

@@ -0,0 +1,449 @@
+Culled Empty Block (label) @1
+Culled Empty Block (label) plus::@1
+
+CONTROL FLOW GRAPH SSA
+@begin: scope:[]  from
+  (byte*) SCREEN#0 ← ((byte*)) (number) $400
+  to:@2
+
+(void()) main()
+main: scope:[main]  from @2
+  (byte~) main::$0 ← call plus (byte) '0' (number) 7 
+  *((byte*) SCREEN#0 + (number) 0) ← (byte~) main::$0
+  to:main::@return
+main::@return: scope:[main]  from main
+  return 
+  to:@return
+
+__stackcall (byte()) plus((byte) plus::a , (byte) plus::b)
+plus: scope:[plus]  from
+  (byte) plus::a#0 ← param((byte) plus::a)
+  (byte) plus::b#0 ← param((byte) plus::b)
+  (byte~) plus::$0 ← (byte) plus::a#0 + (byte) plus::b#0
+  (byte) plus::return#0 ← (byte~) plus::$0
+  to:plus::@return
+plus::@return: scope:[plus]  from plus
+  (byte) plus::return#1 ← phi( plus/(byte) plus::return#0 )
+  return (byte) plus::return#1
+  to:@return
+@2: scope:[]  from @begin
+  call main 
+  to:@3
+@3: scope:[]  from @2
+  to:@end
+@end: scope:[]  from @3
+
+SYMBOL TABLE SSA
+(label) @2
+(label) @3
+(label) @begin
+(label) @end
+(byte*) SCREEN
+(byte*) SCREEN#0
+(void()) main()
+(byte~) main::$0
+(label) main::@return
+__stackcall (byte()) plus((byte) plus::a , (byte) plus::b)
+(byte~) plus::$0
+(label) plus::@return
+(byte) plus::a
+(byte) plus::a#0
+(byte) plus::b
+(byte) plus::b#0
+(byte) plus::return
+(byte) plus::return#0
+(byte) plus::return#1
+
+Adding number conversion cast (unumber) 7 in (byte~) main::$0 ← call plus (byte) '0' (number) 7 
+Adding number conversion cast (unumber) 0 in *((byte*) SCREEN#0 + (number) 0) ← (byte~) main::$0
+Successful SSA optimization PassNAddNumberTypeConversions
+Inlining cast (byte*) SCREEN#0 ← (byte*)(number) $400
+Successful SSA optimization Pass2InlineCast
+Simplifying constant pointer cast (byte*) 1024
+Simplifying constant integer cast 7
+Simplifying constant integer cast 0
+Successful SSA optimization PassNCastSimplification
+Finalized unsigned number type (byte) 7
+Finalized unsigned number type (byte) 0
+Successful SSA optimization PassNFinalizeNumberTypeConversions
+Alias (byte) plus::return#0 = (byte~) plus::$0 (byte) plus::return#1 
+Successful SSA optimization Pass2AliasElimination
+Constant (const byte*) SCREEN#0 = (byte*) 1024
+Successful SSA optimization Pass2ConstantIdentification
+Simplifying expression containing zero SCREEN#0 in [2] *((const byte*) SCREEN#0 + (byte) 0) ← (byte~) main::$0
+Successful SSA optimization PassNSimplifyExpressionWithZero
+Adding NOP phi() at start of @begin
+Adding NOP phi() at start of @2
+Adding NOP phi() at start of @3
+Adding NOP phi() at start of @end
+Adding NOP phi() at start of main
+CALL GRAPH
+Calls in [] to main:2 
+Calls in [main] to plus:6 
+
+Created 0 initial phi equivalence classes
+Coalesced down to 0 phi equivalence classes
+Culled Empty Block (label) @3
+Renumbering block @2 to @1
+Adding NOP phi() at start of @begin
+Adding NOP phi() at start of @1
+Adding NOP phi() at start of @end
+Adding NOP phi() at start of main
+Calling convention STACK_CALL replacing param((byte) plus::a) with paramstack(byte,plus::OFFSET_STACK_A)
+Calling convention STACK_CALL replacing param((byte) plus::b) with paramstack(byte,plus::OFFSET_STACK_B)
+
+FINAL CONTROL FLOW GRAPH
+@begin: scope:[]  from
+  [0] phi()
+  to:@1
+@1: scope:[]  from @begin
+  [1] phi()
+  [2] call main 
+  to:@end
+@end: scope:[]  from @1
+  [3] phi()
+
+(void()) main()
+main: scope:[main]  from @1
+  [4] phi()
+  [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7 
+  [6] *((const byte*) SCREEN#0) ← (byte~) main::$0
+  to:main::@return
+main::@return: scope:[main]  from main
+  [7] return 
+  to:@return
+
+__stackcall (byte()) plus((byte) plus::a , (byte) plus::b)
+plus: scope:[plus]  from
+  [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A)
+  [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B)
+  [10] (byte) plus::return#0 ← (byte) plus::a#0 + (byte) plus::b#0
+  to:plus::@return
+plus::@return: scope:[plus]  from plus
+  [11] return (byte) plus::return#0
+  to:@return
+
+
+VARIABLE REGISTER WEIGHTS
+(byte*) SCREEN
+(void()) main()
+(byte~) main::$0 0.5
+__stackcall (byte()) plus((byte) plus::a , (byte) plus::b)
+(byte) plus::a
+(byte) plus::a#0 2.0
+(byte) plus::b
+(byte) plus::b#0 4.0
+(byte) plus::return
+(byte) plus::return#0 2.0
+
+Initial phi equivalence classes
+Added variable main::$0 to zero page equivalence class [ main::$0 ]
+Added variable plus::a#0 to zero page equivalence class [ plus::a#0 ]
+Added variable plus::b#0 to zero page equivalence class [ plus::b#0 ]
+Added variable plus::return#0 to zero page equivalence class [ plus::return#0 ]
+Complete equivalence classes
+[ main::$0 ]
+[ plus::a#0 ]
+[ plus::b#0 ]
+[ plus::return#0 ]
+Allocated zp ZP_BYTE:2 [ main::$0 ]
+Allocated zp ZP_BYTE:3 [ plus::a#0 ]
+Allocated zp ZP_BYTE:4 [ plus::b#0 ]
+Allocated zp ZP_BYTE:5 [ plus::return#0 ]
+
+INITIAL ASM
+Target platform is c64basic / MOS6502X
+  // File Comments
+// Test a procedure with calling convention stack
+  // Upstart
+.pc = $801 "Basic"
+:BasicUpstart(bbegin)
+.pc = $80d "Program"
+  // Global Constants & labels
+  .label SCREEN = $400
+  .const STACK_BASE = $103
+  // @begin
+bbegin:
+  // [1] phi from @begin to @1 [phi:@begin->@1]
+b1_from_bbegin:
+  jmp b1
+  // @1
+b1:
+  // [2] call main 
+  // [4] phi from @1 to main [phi:@1->main]
+main_from_b1:
+  jsr main
+  // [3] phi from @1 to @end [phi:@1->@end]
+bend_from_b1:
+  jmp bend
+  // @end
+bend:
+  // main
+main: {
+    .label _0 = 2
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7 
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpushbyte_=vbuc1 
+    lda #'0'
+    pha
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpushbyte_=vbuc1 
+    lda #7
+    pha
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- jsr 
+    jsr plus
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpullbyte_2 
+    pla
+    pla
+    // [6] *((const byte*) SCREEN#0) ← (byte~) main::$0 -- _deref_pbuc1=vbuz1 
+    lda.z _0
+    sta SCREEN
+    jmp breturn
+    // main::@return
+  breturn:
+    // [7] return 
+    rts
+}
+  // plus
+// plus(byte zeropage(3) a, byte zeropage(4) b)
+plus: {
+    .const OFFSET_STACK_A = 0
+    .const OFFSET_STACK_B = 1
+    .label a = 3
+    .label b = 4
+    .label return = 5
+    // [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A) -- vbuz1=_stackgetbyte_vbuc1 
+    tsx
+    lda STACK_BASE+OFFSET_STACK_A,x
+    sta.z a
+    // [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B) -- vbuz1=_stackgetbyte_vbuc1 
+    tsx
+    lda STACK_BASE+OFFSET_STACK_B,x
+    sta.z b
+    // [10] (byte) plus::return#0 ← (byte) plus::a#0 + (byte) plus::b#0 -- vbuz1=vbuz2_plus_vbuz3 
+    lda.z a
+    clc
+    adc.z b
+    sta.z return
+    jmp breturn
+    // plus::@return
+  breturn:
+    // [11] return (byte) plus::return#0
+    rts
+}
+  // File Data
+
+REGISTER UPLIFT POTENTIAL REGISTERS
+Statement [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  [ main::$0 ] ( main:2 [ main::$0 ] ) always clobbers reg byte a 
+Statement [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A) [ plus::a#0 ] ( main:2::plus:5 [ main::$0 plus::a#0 ] ) always clobbers reg byte a reg byte x 
+Removing always clobbered register reg byte a as potential for zp ZP_BYTE:2 [ main::$0 ]
+Removing always clobbered register reg byte x as potential for zp ZP_BYTE:2 [ main::$0 ]
+Statement [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B) [ plus::a#0 plus::b#0 ] ( main:2::plus:5 [ main::$0 plus::a#0 plus::b#0 ] ) always clobbers reg byte a reg byte x 
+Removing always clobbered register reg byte a as potential for zp ZP_BYTE:3 [ plus::a#0 ]
+Removing always clobbered register reg byte x as potential for zp ZP_BYTE:3 [ plus::a#0 ]
+Statement [10] (byte) plus::return#0 ← (byte) plus::a#0 + (byte) plus::b#0 [ plus::return#0 ] ( main:2::plus:5 [ main::$0 plus::return#0 ] ) always clobbers reg byte a 
+Statement [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  [ main::$0 ] ( main:2 [ main::$0 ] ) always clobbers reg byte a 
+Statement [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A) [ plus::a#0 ] ( main:2::plus:5 [ main::$0 plus::a#0 ] ) always clobbers reg byte a reg byte x 
+Statement [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B) [ plus::a#0 plus::b#0 ] ( main:2::plus:5 [ main::$0 plus::a#0 plus::b#0 ] ) always clobbers reg byte a reg byte x 
+Statement [10] (byte) plus::return#0 ← (byte) plus::a#0 + (byte) plus::b#0 [ plus::return#0 ] ( main:2::plus:5 [ main::$0 plus::return#0 ] ) always clobbers reg byte a 
+Potential registers zp ZP_BYTE:2 [ main::$0 ] : zp ZP_BYTE:2 , reg byte y , 
+Potential registers zp ZP_BYTE:3 [ plus::a#0 ] : zp ZP_BYTE:3 , reg byte y , 
+Potential registers zp ZP_BYTE:4 [ plus::b#0 ] : zp ZP_BYTE:4 , reg byte a , reg byte x , reg byte y , 
+Potential registers zp ZP_BYTE:5 [ plus::return#0 ] : zp ZP_BYTE:5 , reg byte a , reg byte x , reg byte y , 
+
+REGISTER UPLIFT SCOPES
+Uplift Scope [plus] 4: zp ZP_BYTE:4 [ plus::b#0 ] 2: zp ZP_BYTE:3 [ plus::a#0 ] 2: zp ZP_BYTE:5 [ plus::return#0 ] 
+Uplift Scope [main] 0.5: zp ZP_BYTE:2 [ main::$0 ] 
+Uplift Scope [] 
+
+Uplifting [plus] best 81 combination reg byte a [ plus::b#0 ] zp ZP_BYTE:3 [ plus::a#0 ] reg byte a [ plus::return#0 ] 
+Uplifting [main] best 78 combination reg byte y [ main::$0 ] 
+Uplifting [] best 78 combination 
+Attempting to uplift remaining variables inzp ZP_BYTE:3 [ plus::a#0 ]
+Uplifting [plus] best 78 combination zp ZP_BYTE:3 [ plus::a#0 ] 
+Allocated (was zp ZP_BYTE:3) zp ZP_BYTE:2 [ plus::a#0 ]
+
+ASSEMBLER BEFORE OPTIMIZATION
+  // File Comments
+// Test a procedure with calling convention stack
+  // Upstart
+.pc = $801 "Basic"
+:BasicUpstart(bbegin)
+.pc = $80d "Program"
+  // Global Constants & labels
+  .label SCREEN = $400
+  .const STACK_BASE = $103
+  // @begin
+bbegin:
+  // [1] phi from @begin to @1 [phi:@begin->@1]
+b1_from_bbegin:
+  jmp b1
+  // @1
+b1:
+  // [2] call main 
+  // [4] phi from @1 to main [phi:@1->main]
+main_from_b1:
+  jsr main
+  // [3] phi from @1 to @end [phi:@1->@end]
+bend_from_b1:
+  jmp bend
+  // @end
+bend:
+  // main
+main: {
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7 
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpushbyte_=vbuc1 
+    lda #'0'
+    pha
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpushbyte_=vbuc1 
+    lda #7
+    pha
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- jsr 
+    jsr plus
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpullbyte_2 
+    pla
+    pla
+    // [6] *((const byte*) SCREEN#0) ← (byte~) main::$0 -- _deref_pbuc1=vbuyy 
+    sty SCREEN
+    jmp breturn
+    // main::@return
+  breturn:
+    // [7] return 
+    rts
+}
+  // plus
+// plus(byte zeropage(2) a, byte register(A) b)
+plus: {
+    .const OFFSET_STACK_A = 0
+    .const OFFSET_STACK_B = 1
+    .label a = 2
+    // [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A) -- vbuz1=_stackgetbyte_vbuc1 
+    tsx
+    lda STACK_BASE+OFFSET_STACK_A,x
+    sta.z a
+    // [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B) -- vbuaa=_stackgetbyte_vbuc1 
+    tsx
+    lda STACK_BASE+OFFSET_STACK_B,x
+    // [10] (byte) plus::return#0 ← (byte) plus::a#0 + (byte) plus::b#0 -- vbuaa=vbuz1_plus_vbuaa 
+    clc
+    adc.z a
+    jmp breturn
+    // plus::@return
+  breturn:
+    // [11] return (byte) plus::return#0
+    rts
+}
+  // File Data
+
+ASSEMBLER OPTIMIZATIONS
+Removing instruction jmp b1
+Removing instruction jmp bend
+Removing instruction jmp breturn
+Removing instruction jmp breturn
+Succesful ASM optimization Pass5NextJumpElimination
+Removing instruction b1_from_bbegin:
+Removing instruction b1:
+Removing instruction main_from_b1:
+Removing instruction bend_from_b1:
+Succesful ASM optimization Pass5RedundantLabelElimination
+Removing instruction bend:
+Removing instruction breturn:
+Removing instruction breturn:
+Succesful ASM optimization Pass5UnusedLabelElimination
+Updating BasicUpstart to call main directly
+Removing instruction jsr main
+Succesful ASM optimization Pass5SkipBegin
+Removing instruction bbegin:
+Succesful ASM optimization Pass5UnusedLabelElimination
+
+FINAL SYMBOL TABLE
+(label) @1
+(label) @begin
+(label) @end
+(byte*) SCREEN
+(const byte*) SCREEN#0 SCREEN = (byte*) 1024
+(const word) STACK_BASE STACK_BASE = (word) $103
+(void()) main()
+(byte~) main::$0 reg byte y 0.5
+(label) main::@return
+__stackcall (byte()) plus((byte) plus::a , (byte) plus::b)
+(label) plus::@return
+(const byte) plus::OFFSET_STACK_A OFFSET_STACK_A = (byte) 0
+(const byte) plus::OFFSET_STACK_B OFFSET_STACK_B = (byte) 1
+(byte) plus::a
+(byte) plus::a#0 a zp ZP_BYTE:2 2.0
+(byte) plus::b
+(byte) plus::b#0 reg byte a 4.0
+(byte) plus::return
+(byte) plus::return#0 reg byte a 2.0
+
+reg byte y [ main::$0 ]
+zp ZP_BYTE:2 [ plus::a#0 ]
+reg byte a [ plus::b#0 ]
+reg byte a [ plus::return#0 ]
+
+
+FINAL ASSEMBLER
+Score: 60
+
+  // File Comments
+// Test a procedure with calling convention stack
+  // Upstart
+.pc = $801 "Basic"
+:BasicUpstart(main)
+.pc = $80d "Program"
+  // Global Constants & labels
+  .label SCREEN = $400
+  .const STACK_BASE = $103
+  // @begin
+  // [1] phi from @begin to @1 [phi:@begin->@1]
+  // @1
+  // [2] call main 
+  // [4] phi from @1 to main [phi:@1->main]
+  // [3] phi from @1 to @end [phi:@1->@end]
+  // @end
+  // main
+main: {
+    // plus('0', 7)
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7 
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpushbyte_=vbuc1 
+    lda #'0'
+    pha
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpushbyte_=vbuc1 
+    lda #7
+    pha
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- jsr 
+    jsr plus
+    // [5] (byte~) main::$0 ← call plus (byte) '0' (byte) 7  -- _stackpullbyte_2 
+    pla
+    pla
+    // SCREEN[0] = plus('0', 7)
+    // [6] *((const byte*) SCREEN#0) ← (byte~) main::$0 -- _deref_pbuc1=vbuyy 
+    sty SCREEN
+    // main::@return
+    // }
+    // [7] return 
+    rts
+}
+  // plus
+// plus(byte zeropage(2) a, byte register(A) b)
+plus: {
+    .const OFFSET_STACK_A = 0
+    .const OFFSET_STACK_B = 1
+    .label a = 2
+    // [8] (byte) plus::a#0 ← paramstack(byte,plus::OFFSET_STACK_A) -- vbuz1=_stackgetbyte_vbuc1 
+    tsx
+    lda STACK_BASE+OFFSET_STACK_A,x
+    sta.z a
+    // [9] (byte) plus::b#0 ← paramstack(byte,plus::OFFSET_STACK_B) -- vbuaa=_stackgetbyte_vbuc1 
+    tsx
+    lda STACK_BASE+OFFSET_STACK_B,x
+    // return a+b;
+    // [10] (byte) plus::return#0 ← (byte) plus::a#0 + (byte) plus::b#0 -- vbuaa=vbuz1_plus_vbuaa 
+    clc
+    adc.z a
+    // plus::@return
+    // }
+    // [11] return (byte) plus::return#0
+    rts
+}
+  // File Data
+

src/test/ref/procedure-callingconvention-stack-1.sym

@@ -0,0 +1,24 @@
+(label) @1
+(label) @begin
+(label) @end
+(byte*) SCREEN
+(const byte*) SCREEN#0 SCREEN = (byte*) 1024
+(const word) STACK_BASE STACK_BASE = (word) $103
+(void()) main()
+(byte~) main::$0 reg byte y 0.5
+(label) main::@return
+__stackcall (byte()) plus((byte) plus::a , (byte) plus::b)
+(label) plus::@return
+(const byte) plus::OFFSET_STACK_A OFFSET_STACK_A = (byte) 0
+(const byte) plus::OFFSET_STACK_B OFFSET_STACK_B = (byte) 1
+(byte) plus::a
+(byte) plus::a#0 a zp ZP_BYTE:2 2.0
+(byte) plus::b
+(byte) plus::b#0 reg byte a 4.0
+(byte) plus::return
+(byte) plus::return#0 reg byte a 2.0
+
+reg byte y [ main::$0 ]
+zp ZP_BYTE:2 [ plus::a#0 ]
+reg byte a [ plus::b#0 ]
+reg byte a [ plus::return#0 ]

src/test/ref/procedure-callingconvention-stack-2.asm

@@ -0,0 +1,53 @@
+// Test a procedure with calling convention stack - and enough parameters to use fast ASM for cleaning stack
+.pc = $801 "Basic"
+:BasicUpstart(main)
+.pc = $80d "Program"
+  .label SCREEN = $400
+  .const STACK_BASE = $103
+main: {
+    .label _0 = 2
+    lda #>$1234
+    pha
+    lda #<$1234
+    pha
+    lda #>$2345
+    pha
+    lda #<$2345
+    pha
+    jsr plus
+    tsx
+    txa
+    axs #-4
+    txs
+    lda.z _0
+    sta SCREEN
+    lda.z _0+1
+    sta SCREEN+1
+    rts
+}
+// plus(word zeropage(4) a, word zeropage(6) b)
+plus: {
+    .const OFFSET_STACK_A = 0
+    .const OFFSET_STACK_B = 2
+    .label a = 4
+    .label b = 6
+    .label return = 4
+    tsx
+    lda STACK_BASE+OFFSET_STACK_A,x
+    sta.z a
+    lda STACK_BASE+OFFSET_STACK_A+1,x
+    sta.z a+1
+    tsx
+    lda STACK_BASE+OFFSET_STACK_B,x
+    sta.z b
+    lda STACK_BASE+OFFSET_STACK_B+1,x
+    sta.z b+1
+    lda.z return
+    clc
+    adc.z b
+    sta.z return
+    lda.z return+1
+    adc.z b+1
+    sta.z return+1
+    rts
+}

src/test/ref/procedure-callingconvention-stack-2.cfg

@@ -0,0 +1,29 @@
+@begin: scope:[]  from
+  [0] phi()
+  to:@1
+@1: scope:[]  from @begin
+  [1] phi()
+  [2] call main 
+  to:@end
+@end: scope:[]  from @1
+  [3] phi()
+
+(void()) main()
+main: scope:[main]  from @1
+  [4] phi()
+  [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345 
+  [6] *((const word*) SCREEN#0) ← (word~) main::$0
+  to:main::@return
+main::@return: scope:[main]  from main
+  [7] return 
+  to:@return
+
+__stackcall (word()) plus((word) plus::a , (word) plus::b)
+plus: scope:[plus]  from
+  [8] (word) plus::a#0 ← paramstack(word,plus::OFFSET_STACK_A)
+  [9] (word) plus::b#0 ← paramstack(word,plus::OFFSET_STACK_B)
+  [10] (word) plus::return#0 ← (word) plus::a#0 + (word) plus::b#0
+  to:plus::@return
+plus::@return: scope:[plus]  from plus
+  [11] return (word) plus::return#0
+  to:@return

src/test/ref/procedure-callingconvention-stack-2.log

@@ -0,0 +1,515 @@
+Fixing pointer array-indexing *((word*) SCREEN + (number) 0)
+Culled Empty Block (label) @1
+Culled Empty Block (label) plus::@1
+
+CONTROL FLOW GRAPH SSA
+@begin: scope:[]  from
+  (word*) SCREEN#0 ← ((word*)) (number) $400
+  to:@2
+
+(void()) main()
+main: scope:[main]  from @2
+  (word~) main::$0 ← call plus (number) $1234 (number) $2345 
+  (number~) main::$1 ← (number) 0 * (const byte) SIZEOF_WORD
+  *((word*) SCREEN#0 + (number~) main::$1) ← (word~) main::$0
+  to:main::@return
+main::@return: scope:[main]  from main
+  return 
+  to:@return
+
+__stackcall (word()) plus((word) plus::a , (word) plus::b)
+plus: scope:[plus]  from
+  (word) plus::a#0 ← param((word) plus::a)
+  (word) plus::b#0 ← param((word) plus::b)
+  (word~) plus::$0 ← (word) plus::a#0 + (word) plus::b#0
+  (word) plus::return#0 ← (word~) plus::$0
+  to:plus::@return
+plus::@return: scope:[plus]  from plus
+  (word) plus::return#1 ← phi( plus/(word) plus::return#0 )
+  return (word) plus::return#1
+  to:@return
+@2: scope:[]  from @begin
+  call main 
+  to:@3
+@3: scope:[]  from @2
+  to:@end
+@end: scope:[]  from @3
+
+SYMBOL TABLE SSA
+(label) @2
+(label) @3
+(label) @begin
+(label) @end
+(word*) SCREEN
+(word*) SCREEN#0
+(const byte) SIZEOF_WORD = (byte) 2
+(void()) main()
+(word~) main::$0
+(number~) main::$1
+(label) main::@return
+__stackcall (word()) plus((word) plus::a , (word) plus::b)
+(word~) plus::$0
+(label) plus::@return
+(word) plus::a
+(word) plus::a#0
+(word) plus::b
+(word) plus::b#0
+(word) plus::return
+(word) plus::return#0
+(word) plus::return#1
+
+Adding number conversion cast (unumber) $1234 in (word~) main::$0 ← call plus (number) $1234 (number) $2345 
+Adding number conversion cast (unumber) $2345 in (word~) main::$0 ← call plus (unumber)(number) $1234 (number) $2345 
+Adding number conversion cast (unumber) 0 in (number~) main::$1 ← (number) 0 * (const byte) SIZEOF_WORD
+Adding number conversion cast (unumber) main::$1 in (number~) main::$1 ← (unumber)(number) 0 * (const byte) SIZEOF_WORD
+Successful SSA optimization PassNAddNumberTypeConversions
+Inlining cast (word*) SCREEN#0 ← (word*)(number) $400
+Successful SSA optimization Pass2InlineCast
+Simplifying constant pointer cast (word*) 1024
+Simplifying constant integer cast $1234
+Simplifying constant integer cast $2345
+Simplifying constant integer cast 0
+Successful SSA optimization PassNCastSimplification
+Finalized unsigned number type (word) $1234
+Finalized unsigned number type (word) $2345
+Finalized unsigned number type (byte) 0
+Successful SSA optimization PassNFinalizeNumberTypeConversions
+Inferred type updated to byte in (unumber~) main::$1 ← (byte) 0 * (const byte) SIZEOF_WORD
+Alias (word) plus::return#0 = (word~) plus::$0 (word) plus::return#1 
+Successful SSA optimization Pass2AliasElimination
+Constant right-side identified [2] (byte~) main::$1 ← (byte) 0 * (const byte) SIZEOF_WORD
+Successful SSA optimization Pass2ConstantRValueConsolidation
+Constant (const word*) SCREEN#0 = (word*) 1024
+Constant (const byte) main::$1 = 0*SIZEOF_WORD
+Successful SSA optimization Pass2ConstantIdentification
+Simplifying constant evaluating to zero (byte) 0*(const byte) SIZEOF_WORD in 
+Successful SSA optimization PassNSimplifyConstantZero
+Simplifying expression containing zero SCREEN#0 in [3] *((const word*) SCREEN#0 + (const byte) main::$1) ← (word~) main::$0
+Successful SSA optimization PassNSimplifyExpressionWithZero
+Eliminating unused constant (const byte) main::$1
+Eliminating unused constant (const byte) SIZEOF_WORD
+Successful SSA optimization PassNEliminateUnusedVars
+Adding NOP phi() at start of @begin
+Adding NOP phi() at start of @2
+Adding NOP phi() at start of @3
+Adding NOP phi() at start of @end
+Adding NOP phi() at start of main
+CALL GRAPH
+Calls in [] to main:2 
+Calls in [main] to plus:6 
+
+Created 0 initial phi equivalence classes
+Coalesced down to 0 phi equivalence classes
+Culled Empty Block (label) @3
+Renumbering block @2 to @1
+Adding NOP phi() at start of @begin
+Adding NOP phi() at start of @1
+Adding NOP phi() at start of @end
+Adding NOP phi() at start of main
+Calling convention STACK_CALL replacing param((word) plus::a) with paramstack(word,plus::OFFSET_STACK_A)
+Calling convention STACK_CALL replacing param((word) plus::b) with paramstack(word,plus::OFFSET_STACK_B)
+
+FINAL CONTROL FLOW GRAPH
+@begin: scope:[]  from
+  [0] phi()
+  to:@1
+@1: scope:[]  from @begin
+  [1] phi()
+  [2] call main 
+  to:@end
+@end: scope:[]  from @1
+  [3] phi()
+
+(void()) main()
+main: scope:[main]  from @1
+  [4] phi()
+  [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345 
+  [6] *((const word*) SCREEN#0) ← (word~) main::$0
+  to:main::@return
+main::@return: scope:[main]  from main
+  [7] return 
+  to:@return
+
+__stackcall (word()) plus((word) plus::a , (word) plus::b)
+plus: scope:[plus]  from
+  [8] (word) plus::a#0 ← paramstack(word,plus::OFFSET_STACK_A)
+  [9] (word) plus::b#0 ← paramstack(word,plus::OFFSET_STACK_B)
+  [10] (word) plus::return#0 ← (word) plus::a#0 + (word) plus::b#0
+  to:plus::@return
+plus::@return: scope:[plus]  from plus
+  [11] return (word) plus::return#0
+  to:@return
+
+
+VARIABLE REGISTER WEIGHTS
+(word*) SCREEN
+(void()) main()
+(word~) main::$0 0.5
+__stackcall (word()) plus((word) plus::a , (word) plus::b)
+(word) plus::a
+(word) plus::a#0 2.0
+(word) plus::b
+(word) plus::b#0 4.0
+(word) plus::return
+(word) plus::return#0 2.0
+
+Initial phi equivalence classes
+Added variable main::$0 to zero page equivalence class [ main::$0 ]
+Added variable plus::a#0 to zero page equivalence class [ plus::a#0 ]
+Added variable plus::b#0 to zero page equivalence class [ plus::b#0 ]
+Added variable plus::return#0 to zero page equivalence class [ plus::return#0 ]
+Complete equivalence classes
+[ main::$0 ]
+[ plus::a#0 ]
+[ plus::b#0 ]
+[ plus::return#0 ]
+Allocated zp ZP_WORD:2 [ main::$0 ]
+Allocated zp ZP_WORD:4 [ plus::a#0 ]
+Allocated zp ZP_WORD:6 [ plus::b#0 ]
+Allocated zp ZP_WORD:8 [ plus::return#0 ]
+
+INITIAL ASM
+Target platform is c64basic / MOS6502X
+  // File Comments
+// Test a procedure with calling convention stack - and enough parameters to use fast ASM for cleaning stack
+  // Upstart
+.pc = $801 "Basic"
+:BasicUpstart(bbegin)
+.pc = $80d "Program"
+  // Global Constants & labels
+  .label SCREEN = $400
+  .const STACK_BASE = $103
+  // @begin
+bbegin:
+  // [1] phi from @begin to @1 [phi:@begin->@1]
+b1_from_bbegin:
+  jmp b1
+  // @1
+b1:
+  // [2] call main 
+  // [4] phi from @1 to main [phi:@1->main]
+main_from_b1:
+  jsr main
+  // [3] phi from @1 to @end [phi:@1->@end]
+bend_from_b1:
+  jmp bend
+  // @end
+bend:
+  // main
+main: {
+    .label _0 = 2
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345 
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  -- _stackpushword_=vwuc1 
+    lda #>$1234
+    pha
+    lda #<$1234
+    pha
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  -- _stackpushword_=vwuc1 
+    lda #>$2345
+    pha
+    lda #<$2345
+    pha
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  -- jsr 
+    jsr plus
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  -- _stackpullbyte_4 
+    tsx
+    txa
+    axs #-4
+    txs
+    // [6] *((const word*) SCREEN#0) ← (word~) main::$0 -- _deref_pwuc1=vwuz1 
+    lda.z _0
+    sta SCREEN
+    lda.z _0+1
+    sta SCREEN+1
+    jmp breturn
+    // main::@return
+  breturn:
+    // [7] return 
+    rts
+}
+  // plus
+// plus(word zeropage(4) a, word zeropage(6) b)
+plus: {
+    .const OFFSET_STACK_A = 0
+    .const OFFSET_STACK_B = 2
+    .label a = 4
+    .label b = 6
+    .label return = 8
+    // [8] (word) plus::a#0 ← paramstack(word,plus::OFFSET_STACK_A) -- vwuz1=_stackgetword_vbuc1 
+    tsx
+    lda STACK_BASE+OFFSET_STACK_A,x
+    sta.z a
+    lda STACK_BASE+OFFSET_STACK_A+1,x
+    sta.z a+1
+    // [9] (word) plus::b#0 ← paramstack(word,plus::OFFSET_STACK_B) -- vwuz1=_stackgetword_vbuc1 
+    tsx
+    lda STACK_BASE+OFFSET_STACK_B,x
+    sta.z b
+    lda STACK_BASE+OFFSET_STACK_B+1,x
+    sta.z b+1
+    // [10] (word) plus::return#0 ← (word) plus::a#0 + (word) plus::b#0 -- vwuz1=vwuz2_plus_vwuz3 
+    lda.z a
+    clc
+    adc.z b
+    sta.z return
+    lda.z a+1
+    adc.z b+1
+    sta.z return+1
+    jmp breturn
+    // plus::@return
+  breturn:
+    // [11] return (word) plus::return#0
+    rts
+}
+  // File Data
+
+REGISTER UPLIFT POTENTIAL REGISTERS
+Statement [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  [ main::$0 ] ( main:2 [ main::$0 ] ) always clobbers reg byte a reg byte x 
+Statement [6] *((const word*) SCREEN#0) ← (word~) main::$0 [ ] ( main:2 [ ] ) always clobbers reg byte a 
+Statement [8] (word) plus::a#0 ← paramstack(word,plus::OFFSET_STACK_A) [ plus::a#0 ] ( main:2::plus:5 [ main::$0 plus::a#0 ] ) always clobbers reg byte a reg byte x 
+Statement [9] (word) plus::b#0 ← paramstack(word,plus::OFFSET_STACK_B) [ plus::a#0 plus::b#0 ] ( main:2::plus:5 [ main::$0 plus::a#0 plus::b#0 ] ) always clobbers reg byte a reg byte x 
+Statement [10] (word) plus::return#0 ← (word) plus::a#0 + (word) plus::b#0 [ plus::return#0 ] ( main:2::plus:5 [ main::$0 plus::return#0 ] ) always clobbers reg byte a 
+Potential registers zp ZP_WORD:2 [ main::$0 ] : zp ZP_WORD:2 , 
+Potential registers zp ZP_WORD:4 [ plus::a#0 ] : zp ZP_WORD:4 , 
+Potential registers zp ZP_WORD:6 [ plus::b#0 ] : zp ZP_WORD:6 , 
+Potential registers zp ZP_WORD:8 [ plus::return#0 ] : zp ZP_WORD:8 , 
+
+REGISTER UPLIFT SCOPES
+Uplift Scope [plus] 4: zp ZP_WORD:6 [ plus::b#0 ] 2: zp ZP_WORD:4 [ plus::a#0 ] 2: zp ZP_WORD:8 [ plus::return#0 ] 
+Uplift Scope [main] 0.5: zp ZP_WORD:2 [ main::$0 ] 
+Uplift Scope [] 
+
+Uplifting [plus] best 132 combination zp ZP_WORD:6 [ plus::b#0 ] zp ZP_WORD:4 [ plus::a#0 ] zp ZP_WORD:8 [ plus::return#0 ] 
+Uplifting [main] best 132 combination zp ZP_WORD:2 [ main::$0 ] 
+Uplifting [] best 132 combination 
+Coalescing zero page register [ zp ZP_WORD:4 [ plus::a#0 ] ] with [ zp ZP_WORD:8 [ plus::return#0 ] ] - score: 1
+
+ASSEMBLER BEFORE OPTIMIZATION
+  // File Comments
+// Test a procedure with calling convention stack - and enough parameters to use fast ASM for cleaning stack
+  // Upstart
+.pc = $801 "Basic"
+:BasicUpstart(bbegin)
+.pc = $80d "Program"
+  // Global Constants & labels
+  .label SCREEN = $400
+  .const STACK_BASE = $103
+  // @begin
+bbegin:
+  // [1] phi from @begin to @1 [phi:@begin->@1]
+b1_from_bbegin:
+  jmp b1
+  // @1
+b1:
+  // [2] call main 
+  // [4] phi from @1 to main [phi:@1->main]
+main_from_b1:
+  jsr main
+  // [3] phi from @1 to @end [phi:@1->@end]
+bend_from_b1:
+  jmp bend
+  // @end
+bend:
+  // main
+main: {
+    .label _0 = 2
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345 
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  -- _stackpushword_=vwuc1 
+    lda #>$1234
+    pha
+    lda #<$1234
+    pha
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  -- _stackpushword_=vwuc1 
+    lda #>$2345
+    pha
+    lda #<$2345
+    pha
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  -- jsr 
+    jsr plus
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  -- _stackpullbyte_4 
+    tsx
+    txa
+    axs #-4
+    txs
+    // [6] *((const word*) SCREEN#0) ← (word~) main::$0 -- _deref_pwuc1=vwuz1 
+    lda.z _0
+    sta SCREEN
+    lda.z _0+1
+    sta SCREEN+1
+    jmp breturn
+    // main::@return
+  breturn:
+    // [7] return 
+    rts
+}
+  // plus
+// plus(word zeropage(4) a, word zeropage(6) b)
+plus: {
+    .const OFFSET_STACK_A = 0
+    .const OFFSET_STACK_B = 2
+    .label a = 4
+    .label b = 6
+    .label return = 4
+    // [8] (word) plus::a#0 ← paramstack(word,plus::OFFSET_STACK_A) -- vwuz1=_stackgetword_vbuc1 
+    tsx
+    lda STACK_BASE+OFFSET_STACK_A,x
+    sta.z a
+    lda STACK_BASE+OFFSET_STACK_A+1,x
+    sta.z a+1
+    // [9] (word) plus::b#0 ← paramstack(word,plus::OFFSET_STACK_B) -- vwuz1=_stackgetword_vbuc1 
+    tsx
+    lda STACK_BASE+OFFSET_STACK_B,x
+    sta.z b
+    lda STACK_BASE+OFFSET_STACK_B+1,x
+    sta.z b+1
+    // [10] (word) plus::return#0 ← (word) plus::a#0 + (word) plus::b#0 -- vwuz1=vwuz1_plus_vwuz2 
+    lda.z return
+    clc
+    adc.z b
+    sta.z return
+    lda.z return+1
+    adc.z b+1
+    sta.z return+1
+    jmp breturn
+    // plus::@return
+  breturn:
+    // [11] return (word) plus::return#0
+    rts
+}
+  // File Data
+
+ASSEMBLER OPTIMIZATIONS
+Removing instruction jmp b1
+Removing instruction jmp bend
+Removing instruction jmp breturn
+Removing instruction jmp breturn
+Succesful ASM optimization Pass5NextJumpElimination
+Removing instruction b1_from_bbegin:
+Removing instruction b1:
+Removing instruction main_from_b1:
+Removing instruction bend_from_b1:
+Succesful ASM optimization Pass5RedundantLabelElimination
+Removing instruction bend:
+Removing instruction breturn:
+Removing instruction breturn:
+Succesful ASM optimization Pass5UnusedLabelElimination
+Updating BasicUpstart to call main directly
+Removing instruction jsr main
+Succesful ASM optimization Pass5SkipBegin
+Removing instruction bbegin:
+Succesful ASM optimization Pass5UnusedLabelElimination
+
+FINAL SYMBOL TABLE
+(label) @1
+(label) @begin
+(label) @end
+(word*) SCREEN
+(const word*) SCREEN#0 SCREEN = (word*) 1024
+(const word) STACK_BASE STACK_BASE = (word) $103
+(void()) main()
+(word~) main::$0 $0 zp ZP_WORD:2 0.5
+(label) main::@return
+__stackcall (word()) plus((word) plus::a , (word) plus::b)
+(label) plus::@return
+(const byte) plus::OFFSET_STACK_A OFFSET_STACK_A = (byte) 0
+(const byte) plus::OFFSET_STACK_B OFFSET_STACK_B = (byte) 2
+(word) plus::a
+(word) plus::a#0 a zp ZP_WORD:4 2.0
+(word) plus::b
+(word) plus::b#0 b zp ZP_WORD:6 4.0
+(word) plus::return
+(word) plus::return#0 return zp ZP_WORD:4 2.0
+
+zp ZP_WORD:2 [ main::$0 ]
+zp ZP_WORD:4 [ plus::a#0 plus::return#0 ]
+zp ZP_WORD:6 [ plus::b#0 ]
+
+
+FINAL ASSEMBLER
+Score: 114
+
+  // File Comments
+// Test a procedure with calling convention stack - and enough parameters to use fast ASM for cleaning stack
+  // Upstart
+.pc = $801 "Basic"
+:BasicUpstart(main)
+.pc = $80d "Program"
+  // Global Constants & labels
+  .label SCREEN = $400
+  .const STACK_BASE = $103
+  // @begin
+  // [1] phi from @begin to @1 [phi:@begin->@1]
+  // @1
+  // [2] call main 
+  // [4] phi from @1 to main [phi:@1->main]
+  // [3] phi from @1 to @end [phi:@1->@end]
+  // @end
+  // main
+main: {
+    .label _0 = 2
+    // plus(0x1234, 0x2345)
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345 
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  -- _stackpushword_=vwuc1 
+    lda #>$1234
+    pha
+    lda #<$1234
+    pha
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  -- _stackpushword_=vwuc1 
+    lda #>$2345
+    pha
+    lda #<$2345
+    pha
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  -- jsr 
+    jsr plus
+    // [5] (word~) main::$0 ← call plus (word) $1234 (word) $2345  -- _stackpullbyte_4 
+    tsx
+    txa
+    axs #-4
+    txs
+    // SCREEN[0] = plus(0x1234, 0x2345)
+    // [6] *((const word*) SCREEN#0) ← (word~) main::$0 -- _deref_pwuc1=vwuz1 
+    lda.z _0
+    sta SCREEN
+    lda.z _0+1
+    sta SCREEN+1
+    // main::@return
+    // }
+    // [7] return 
+    rts
+}
+  // plus
+// plus(word zeropage(4) a, word zeropage(6) b)
+plus: {
+    .const OFFSET_STACK_A = 0
+    .const OFFSET_STACK_B = 2
+    .label a = 4
+    .label b = 6
+    .label return = 4
+    // [8] (word) plus::a#0 ← paramstack(word,plus::OFFSET_STACK_A) -- vwuz1=_stackgetword_vbuc1 
+    tsx
+    lda STACK_BASE+OFFSET_STACK_A,x
+    sta.z a
+    lda STACK_BASE+OFFSET_STACK_A+1,x
+    sta.z a+1
+    // [9] (word) plus::b#0 ← paramstack(word,plus::OFFSET_STACK_B) -- vwuz1=_stackgetword_vbuc1 
+    tsx
+    lda STACK_BASE+OFFSET_STACK_B,x
+    sta.z b
+    lda STACK_BASE+OFFSET_STACK_B+1,x
+    sta.z b+1
+    // return a+b;
+    // [10] (word) plus::return#0 ← (word) plus::a#0 + (word) plus::b#0 -- vwuz1=vwuz1_plus_vwuz2 
+    lda.z return
+    clc
+    adc.z b
+    sta.z return
+    lda.z return+1
+    adc.z b+1
+    sta.z return+1
+    // plus::@return
+    // }
+    // [11] return (word) plus::return#0
+    rts
+}
+  // File Data
+

src/test/ref/procedure-callingconvention-stack-2.sym

@@ -0,0 +1,23 @@
+(label) @1
+(label) @begin
+(label) @end
+(word*) SCREEN
+(const word*) SCREEN#0 SCREEN = (word*) 1024
+(const word) STACK_BASE STACK_BASE = (word) $103
+(void()) main()
+(word~) main::$0 $0 zp ZP_WORD:2 0.5
+(label) main::@return
+__stackcall (word()) plus((word) plus::a , (word) plus::b)
+(label) plus::@return
+(const byte) plus::OFFSET_STACK_A OFFSET_STACK_A = (byte) 0
+(const byte) plus::OFFSET_STACK_B OFFSET_STACK_B = (byte) 2
+(word) plus::a
+(word) plus::a#0 a zp ZP_WORD:4 2.0
+(word) plus::b
+(word) plus::b#0 b zp ZP_WORD:6 4.0
+(word) plus::return
+(word) plus::return#0 return zp ZP_WORD:4 2.0
+
+zp ZP_WORD:2 [ main::$0 ]
+zp ZP_WORD:4 [ plus::a#0 plus::return#0 ]
+zp ZP_WORD:6 [ plus::b#0 ]