Arrays

Makefile

@@ -10,7 +10,7 @@ clean:
 test: flow tests/bash/runtests.sh
 	(cd tests/bash; chmod +x runtests.sh; ./runtests.sh)
 
-test2: flow tests/input19.flow lib/printint.c
-	./flow tests/input19.flow 
+test2: flow tests/input20.flow lib/printint.c
+	./flow tests/input20.flow 
 	gcc -o out out.s lib/printint.c
 	./out

grammar.txt

@@ -77,7 +77,8 @@ global_declaration: function_declaration | var_declaration
 function_declaration: type identifier '(' ')' compound_statement
     ;
 
-var_declaration: type identifier_list ';'
+variable_declaration: type identifier_List ';'
+    | type identifier '[' P_INTLIT ']' ';'
     ;
 
 type: type_keyword opt_pointer
@@ -108,3 +109,7 @@ primary_expression: IDENTIFIER
     | STRING_LITERAL
     | '(' expression ')'
     ;
+
+postfix_expression : primary_expression
+    | postfix_expression '[' expression ']'
+

out.s

@@ -4,51 +4,50 @@
 a:	.long	0
 	.data
 	.globl	b
-b:	.long	0
-	.data
-	.globl	c
-c:	.long	0
-	.data
-	.globl	d
-d:	.long	0
-	.data
-	.globl	e
-e:	.long	0
+b:	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
+	.quad	0
 	.text
 	.globl	main
 	.type	main, @function
 main:
 	pushq	%rbp
 	movq	%rsp, %rbp
-	movq	$2, %r8
-	movl	%r8d, a(%rip)
-	movq	$4, %r8
-	movl	%r8d, b(%rip)
-	movq	$3, %r8
-	movl	%r8d, c(%rip)
-	movq	$2, %r8
-	movl	%r8d, d(%rip)
-	movzbl	a(%rip), %r8
-	movzbl	b(%rip), %r9
+	movq	$12, %r8
+	leaq	b(%rip), %r9
+	movq	$3, %r10
+	salq	$2, %r10
+	addq	%r9, %r10
+	movq	%r8, (%r10)
+	leaq	b(%rip), %r8
+	movq	$3, %r9
+	salq	$2, %r9
 	addq	%r8, %r9
-	movq	%r9, %rdi
-	call	printint
-	movq	%rax, %r8
-	movzbl	c(%rip), %r8
-	movzbl	d(%rip), %r9
-	addq	%r8, %r9
-	movq	%r9, %rdi
-	call	printint
-	movq	%rax, %r8
+	movq	(%r9), %r9
+	movl	%r9d, a(%rip)
 	movzbl	a(%rip), %r8
-	movzbl	b(%rip), %r9
-	addq	%r8, %r9
-	movzbl	c(%rip), %r8
-	movzbl	d(%rip), %r10
-	addq	%r8, %r10
-	imulq	%r9, %r10
-	movl	%r10d, e(%rip)
-	movzbl	e(%rip), %r8
 	movq	%r8, %rdi
 	call	printint
 	movq	%rax, %r9

src/cg.c

@@ -4,7 +4,6 @@
 
 // Code generator for x86-64
 
-
 // List of available registers and their names.
 // We need a list of byte and doubleword registers, too
 static int freereg[4];
@@ -139,13 +138,6 @@ int cgdiv(int r1, int r2) {
   return (r1);
 }
 
-// Call printint() with the given register
-void cgprintint(int r) {
-  fprintf(Outfile, "\tmovq\t%s, %%rdi\n", reglist[r]);
-  fprintf(Outfile, "\tcall\tprintint\n");
-  free_register(r);
-}
-
 // Call a function with one argument from the given register
 // Return the register with the result
 int cgcall(int r, int id) {
@@ -206,12 +198,18 @@ void cgglobsym(int id) {
   // Get the size of the type
   typesize = cgprimsize(Gsym[id].type);
 
+  // Generate the global identity and the label
   fprintf(Outfile, "\t.data\n" "\t.globl\t%s\n", Gsym[id].name);
-  switch(typesize) {
-    case 1: fprintf(Outfile, "%s:\t.byte\t0\n", Gsym[id].name); break;
-    case 4: fprintf(Outfile, "%s:\t.long\t0\n", Gsym[id].name); break;
-    case 8: fprintf(Outfile, "%s:\t.quad\t0\n", Gsym[id].name); break;
-    default: fatald("Unknown typesize in cgglobsym: ", typesize);
+  fprintf(Outfile, "%s:", Gsym[id].name);
+
+  // Generate the space
+  for (int i=0; i < Gsym[id].size; i++) {
+    switch(typesize) {
+      case 1: fprintf(Outfile, "\t.byte\t0\n"); break;
+      case 4: fprintf(Outfile, "\t.long\t0\n"); break;
+      case 8: fprintf(Outfile, "\t.quad\t0\n"); break;
+      default: fatald("Unknown typesize in cgglobsym: ", typesize);
+    }
   }
 }
 

src/decl.c

@@ -11,18 +11,10 @@
 int parse_type(void) {
   int type;
   switch (Token.token) {
-    case T_VOID:
-      type = P_VOID;
-      break;
-    case T_CHAR:
-      type = P_CHAR;
-      break;
-    case T_INT:
-      type = P_INT;
-      break;
-    case T_LONG:
-      type = P_LONG;
-      break;
+    case T_VOID: type = P_VOID; break;
+    case T_CHAR: type = P_CHAR; break;
+    case T_INT:  type = P_INT;  break;
+    case T_LONG: type = P_LONG; break;
     default:
       fatald("Illegal type, token", Token.token);
   }
@@ -40,25 +32,42 @@ int parse_type(void) {
   return (type);
 }
 
-// variable_declaration: type identifier ';'  ;
-//
 // Parse the declaration of a variable.
 // The identifier has been scanned & we have the type
 void var_declaration(int type) {
   int id;
-
+ 
   // Text now has the identifier's name.
-  // Add it as a known identifier
-  // and generate its space in assembly
-  id = addglob(Text, type, S_VARIABLE, 0);
-  genglobsym(id);
+  // If the enxt token is a '['
+  if (Token.token == T_LBRACKET)
+  {
+    // Skip past the '['
+    scan(&Token);
+
+    // Check we have an array size
+    if (Token.token == T_INTLIT)
+    {
+      // Add this as a known array and generate its spaces in assembly.
+      // We treat the array as a pointer to its element's type
+      id = addglob(Text, pointer_to(type), S_ARRAY, 0, Token.intvalue);
+      genglobsym(id);
+    }
+
+    // Ensure we have a following ']'
+    scan(&Token);
+    match(T_RBRACKET, "]");
+  } 
+  else 
+  {
+    // Add it as a known identifier
+    // and generate its space in assembly
+    id = addglob(Text, type, S_VARIABLE, 0, 1);
+    genglobsym(id);
+  }
   // Get the trailing semicolon
   semi();
 }
 
-//
-// function_declaration: type identifier '(' ')' compound_statement   ;
-//
 // Parse the declaration of a simplistic function.
 // The identifier has been scanned & we have the type
 AST_T *function_declaration(int type) {
@@ -70,7 +79,7 @@ AST_T *function_declaration(int type) {
   // to the symbol table, and set the Functionid global
   // to the function's symbol-id
   endlabel = genlabel();
-  nameslot = addglob(Text, type, S_FUNCTION, endlabel);
+  nameslot = addglob(Text, type, S_FUNCTION, endlabel, 0);
   Functionid = nameslot;
 
   // Scan in the parentheses

src/gen.c

@@ -186,21 +186,23 @@ int genAST(AST_T *n, int label, int parentASTop) {
   return (NOREG);		// Keep -Wall happy
 }
 
+
 void genpreamble() {
   cgpreamble();
 }
+
 void genpostamble() {
   cgpostamble();
 }
+
 void genfreeregs() {
   freeall_registers();
 }
-void genprintint(int reg) {
-  cgprintint(reg);
-}
+
 void genglobsym(int id) {
   cgglobsym(id);
 }
+
 int genprimsize(int type) {
   return (cgprimsize(type));
 }

src/include/decl.h

@@ -21,7 +21,6 @@ int genAST(struct ASTnode *n, int reg, int parentASTop);
 void genpreamble();
 void genpostamble();
 void genfreeregs();
-void genprintint(int reg);
 void genglobsym(int id);
 int genprimsize(int type);
 void genreturn(int reg, int id);
@@ -39,7 +38,6 @@ int cgsub(int r1, int r2);
 int cgmul(int r1, int r2);
 int cgdiv(int r1, int r2);
 int cgshlconst(int r, int val);
-void cgprintint(int r);
 int cgcall(int r, int id);
 int cgstorglob(int r, int id);
 void cgglobsym(int id);
@@ -76,7 +74,7 @@ void fatalc(char *s, int c);
 
 // sym.c
 int findglob(char *s);
-int addglob(char *name, int type, int stype, int endlabel);
+int addglob(char *name, int type, int stype, int endlabel, int size);
 
 // decl.c
 void var_declaration(int type);

src/include/defs.h

@@ -81,4 +81,5 @@ typedef struct symtable {
   int type;			          // Primitive type for the symbol
   int stype;			        // Structural type for the symbol
   int endlabel;			      // For S_FUNCTIONs, the end label
+  int size;               // Number of elements in the symbol
 }SymTable_T;

src/main.c

@@ -58,7 +58,7 @@ int main(int argc, char *argv[]) {
     exit(1);
   }
   // For now, ensure that void printint() is defined
-  addglob("printint", P_CHAR, S_FUNCTION, 0);
+  addglob("printint", P_CHAR, S_FUNCTION, 0, 0);
 
   scan(&Token);			// Get the first token from the input
   genpreamble();		// Output the preamble

src/scan.c

@@ -169,6 +169,8 @@ int scan(token_T *t) {
     case '}': t->token = T_RBRACE; break;
     case '(': t->token = T_LPAREN; break;
     case ')': t->token = T_RPAREN; break;
+    case '[': t->token = T_LBRACKET; break;
+    case ']': t->token = T_RBRACKET; break; 
     case '=': if ((c = next()) == '=') { t->token = T_EQ; } else { putback(c); t->token = T_ASSIGN; } break;
     case '!': if ((c = next()) == '=') { t->token = T_NE; } else { fatalc("Unrecognised character", c); } break;
     case '<': if ((c = next()) == '=') { t->token = T_LE; } else { putback(c); t->token = T_LT; } break;

src/sym.c

@@ -27,10 +27,13 @@ static int newglob(void) {
   return (p);
 }
 
-// Add a global symbol to the symbol table.
-// Also set up its type and structural type.
+// Add a global symbol to the symbol table. Set up its:
+// + type: char, int etc.
+// + structural typ: var, function, array etc.
+// + size: number of elements
+// + endlabel: if this is a function
 // Return the slot number in the symbol table
-int addglob(char *name, int type, int stype, int endlabel) {
+int addglob(char *name, int type, int stype, int endlabel, int size) {
   int y;
 
   // If this is already in the symbol table, return the existing slot
@@ -44,5 +47,6 @@ int addglob(char *name, int type, int stype, int endlabel) {
   Gsym[y].type = type;
   Gsym[y].stype = stype;
   Gsym[y].endlabel = endlabel;
+  Gsym[y].size = size;
   return (y);
 }

tests/input20.flow

@@ -0,0 +1,9 @@
+int a;
+int b[25];
+
+int main() {
+  b[3] = 12;
+  a = b[3];
+  printint(a);
+  return(0);
+}

tests/results/out.input20.flow

@@ -0,0 +1 @@
+12