diff --git a/src/compiler/llvm_codegen_expr.c b/src/compiler/llvm_codegen_expr.c
index 8a3bf686d..8ce882e55 100644
--- a/src/compiler/llvm_codegen_expr.c
+++ b/src/compiler/llvm_codegen_expr.c
@@ -4509,6 +4509,14 @@ static inline void gencontext_emit_expression_list_expr(GenContext *context, BEV
 
 static inline void llvm_emit_return_block(GenContext *context, BEValue *be_value, Type *type, Ast **stmts)
 {
+	// First case - an empty block
+	unsigned count = vec_size(stmts);
+	if (!count)
+	{
+		llvm_value_set(be_value, NULL, type_void);
+		return;
+	}
+
 	Type *type_lowered = type_lowering(type);
 	LLVMValueRef old_ret_out = context->return_out;
 	LLVMBasicBlockRef saved_block_return_exit = context->block_return_exit;
@@ -4533,23 +4541,65 @@ static inline void llvm_emit_return_block(GenContext *context, BEValue *be_value
 	context->error_var = NULL;
 	context->catch_block = NULL;
 
-	VECEACH(stmts, i)
+	// Process all but the last statement.
+	for (unsigned i = 0; i < count - 1; i++)
 	{
 		llvm_emit_stmt(context, stmts[i]);
 	}
+
+	Ast *last_stmt = stmts[count - 1];
+
+	do
+	{
+		// Do we have more than one exit?
+		// Then follow the normal path.
+		if (!llvm_basic_block_is_unused(expr_block)) break;
+
+		// Do we have a void function? That's the only
+		// possible case if the last statement isn't return.
+		if (last_stmt->ast_kind != AST_RETURN_STMT) break;
+
+		// Defers? In that case we also use the default behaviour.
+		// We might optimize this later.
+		if (last_stmt->return_stmt.defer) break;
+
+		Expr *ret_expr = last_stmt->return_stmt.expr;
+
+		// If this is a void return, we can just skip here!
+		if (!ret_expr)
+		{
+			llvm_value_set(be_value, NULL, type_void);
+			goto DONE;
+		}
+
+		// Failable? Then we use the normal path
+		if (IS_FAILABLE(ret_expr)) break;
+
+		// Optimization, emit directly to value
+		llvm_emit_expr(context, be_value, ret_expr);
+		// And remove the alloca
+		LLVMInstructionRemoveFromParent(context->return_out);
+		goto DONE;
+
+	} while (0);
+
+	// Emit the last statement
+	llvm_emit_stmt(context, last_stmt);
+
+	// In the case of a void with no return, then this may be true.
+	if (llvm_basic_block_is_unused(expr_block))
+	{
+		// Skip the expr block.
+		assert(!context->return_out);
+		llvm_value_set(be_value, NULL, type_void);
+		goto DONE;
+	}
+
 	llvm_emit_br(context, expr_block);
 
 	// Emit the exit block.
 	llvm_emit_block(context, expr_block);
 
-	context->return_out = old_ret_out;
-	context->catch_block = error_block;
-	context->error_var = error_out;
-	context->block_return_exit = saved_block_return_exit;
-	context->block_failable_exit = saved_block_failable_exit;
-	context->block_error_var = saved_block_error;
-	context->in_block--;
-
 	if (return_out)
 	{
 		llvm_value_set_address(be_value, return_out, type_lowered);
@@ -4558,6 +4608,16 @@ static inline void llvm_emit_return_block(GenContext *context, BEValue *be_value
 	{
 		llvm_value_set(be_value, NULL, type_void);
 	}
+
+DONE:
+	context->return_out = old_ret_out;
+	context->catch_block = error_block;
+	context->error_var = error_out;
+	context->block_return_exit = saved_block_return_exit;
+	context->block_failable_exit = saved_block_failable_exit;
+	context->block_error_var = saved_block_error;
+	context->in_block--;
+
 }
 
 static inline void llvm_emit_expr_block(GenContext *context, BEValue *be_value, Expr *expr)
diff --git a/src/version.h b/src/version.h
index 15a1c727d..fdca69187 100644
--- a/src/version.h
+++ b/src/version.h
@@ -1 +1 @@
-#define COMPILER_VERSION "PRE.3"
\ No newline at end of file
+#define COMPILER_VERSION "PRE.4"
\ No newline at end of file
diff --git a/test/test_suite/compile_time/typeof_example.c3t b/test/test_suite/compile_time/typeof_example.c3t
index a8d8ea5a3..f00a26baf 100644
--- a/test/test_suite/compile_time/typeof_example.c3t
+++ b/test/test_suite/compile_time/typeof_example.c3t
@@ -17,50 +17,38 @@ fn void main()
 
 // #expect: typeof_example.ll
 
-  %f = alloca float, align 4
-  %i = alloca i32, align 4
-  %value = alloca float, align 4
-  %blockret = alloca i32, align 4
-  %temp = alloca float, align 4
-  %result = alloca i32*, align 8
-  %f2 = alloca float, align 4
-  %value1 = alloca i32, align 4
-  %blockret2 = alloca float, align 4
-  %temp3 = alloca i32, align 4
-  %result4 = alloca float*, align 8
-  store float 1.000000e+02, float* %f, align 4
-  %0 = load float, float* %f, align 4
-  store float %0, float* %value, align 4
-  %1 = load float, float* %value, align 4
-  store float %1, float* %temp, align 4
-  %ptrptr = bitcast float* %temp to i32*
-  store i32* %ptrptr, i32** %result, align 8
-  %2 = load i32*, i32** %result, align 8
-  %3 = load i32, i32* %2, align 8
-  store i32 %3, i32* %blockret, align 4
-  br label %expr_block.exit
-
-expr_block.exit:
-  %4 = load i32, i32* %blockret, align 4
-  store i32 %4, i32* %i, align 4
-  %5 = load i32, i32* %i, align 4
-  store i32 %5, i32* %value1, align 4
-  %6 = load i32, i32* %value1, align 4
-  store i32 %6, i32* %temp3, align 4
-  %ptrptr5 = bitcast i32* %temp3 to float*
-  store float* %ptrptr5, float** %result4, align 8
-  %7 = load float*, float** %result4, align 8
-  %8 = load float, float* %7, align 8
-  store float %8, float* %blockret2, align 4
-  br label %expr_block.exit6
-
-expr_block.exit6:
-  %9 = load float, float* %blockret2, align 4
-  store float %9, float* %f2, align 4
-  %10 = load float, float* %f, align 4
-  %fpfpext = fpext float %10 to double
-  %11 = load i32, i32* %i, align 4
-  %12 = load float, float* %f2, align 4
-  %fpfpext7 = fpext float %12 to double
-  call void (i8*, ...) @printf(i8* getelementptr inbounds ([23 x i8], [23 x i8]* @.str, i32 0, i32 0), double %fpfpext, i32 %11, double %fpfpext7)
-  ret void
\ No newline at end of file
+   %f = alloca float, align 4
+   %i = alloca i32, align 4
+   %value = alloca float, align 4
+   %temp = alloca float, align 4
+   %result = alloca i32*, align 8
+   %f2 = alloca float, align 4
+   %value1 = alloca i32, align 4
+   %temp2 = alloca i32, align 4
+   %result3 = alloca float*, align 8
+   store float 1.000000e+02, float* %f, align 4
+   %0 = load float, float* %f, align 4
+   store float %0, float* %value, align 4
+   %1 = load float, float* %value, align 4
+   store float %1, float* %temp, align 4
+   %ptrptr = bitcast float* %temp to i32*
+   store i32* %ptrptr, i32** %result, align 8
+   %2 = load i32*, i32** %result, align 8
+   %3 = load i32, i32* %2, align 8
+   store i32 %3, i32* %i, align 4
+   %4 = load i32, i32* %i, align 4
+   store i32 %4, i32* %value1, align 4
+   %5 = load i32, i32* %value1, align 4
+   store i32 %5, i32* %temp2, align 4
+   %ptrptr4 = bitcast i32* %temp2 to float*
+   store float* %ptrptr4, float** %result3, align 8
+   %6 = load float*, float** %result3, align 8
+   %7 = load float, float* %6, align 8
+   store float %7, float* %f2, align 4
+   %8 = load float, float* %f, align 4
+   %fpfpext = fpext float %8 to double
+   %9 = load i32, i32* %i, align 4
+   %10 = load float, float* %f2, align 4
+   %fpfpext5 = fpext float %10 to double
+   call void (i8*, ...) @printf(i8* getelementptr inbounds ([23 x i8], [23 x i8]* @.str, i32 0, i32 0), double %fpfpext, i32 %9, double %fpfpext5)
+   ret void
\ No newline at end of file
diff --git a/test/test_suite/from_docs/examples_macro_function.c3t b/test/test_suite/from_docs/examples_macro_function.c3t
index d1a73e8ce..b5096cb80 100644
--- a/test/test_suite/from_docs/examples_macro_function.c3t
+++ b/test/test_suite/from_docs/examples_macro_function.c3t
@@ -39,7 +39,6 @@ entry:
   %b = alloca i32, align 4
   %a1 = alloca i32 (i32)*, align 8
   %b2 = alloca i32, align 4
-  %blockret = alloca i32, align 4
   store i32 2, i32* %a, align 4
   store i32 3, i32* %b, align 4
   store i32 (i32)* @example.square, i32 (i32)** %a1, align 8
@@ -47,14 +46,9 @@ entry:
   %0 = load i32 (i32)*, i32 (i32)** %a1, align 8
   %1 = load i32, i32* %b2, align 4
   %2 = call i32 %0(i32 %1)
-  store i32 %2, i32* %blockret, align 4
-  br label %expr_block.exit
-
-expr_block.exit:                                  ; preds = %entry
-  %3 = load i32, i32* %blockret, align 4
-  %4 = load i32, i32* %a, align 4
-  %add = add i32 %3, %4
-  %5 = load i32, i32* %b, align 4
-  %add3 = add i32 %add, %5
+  %3 = load i32, i32* %a, align 4
+  %add = add i32 %2, %3
+  %4 = load i32, i32* %b, align 4
+  %add3 = add i32 %add, %4
   ret i32 %add3
 }
\ No newline at end of file
diff --git a/test/test_suite/macros/macro_with_body.c3t b/test/test_suite/macros/macro_with_body.c3t
index 7d59e3f20..76af6d816 100644
--- a/test/test_suite/macros/macro_with_body.c3t
+++ b/test/test_suite/macros/macro_with_body.c3t
@@ -83,20 +83,17 @@ define void @main()
   %6 = load i32, i32* %x, align 4
   %7 = load i32, i32* %dy, align 4
   %8 = call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([19 x i8], [19 x i8]* @.str, i32 0, i32 0), i32 %6, i32 %7)
-  br label %expr_block.exit
-
-expr_block.exit:
   store i32 10, i32* %times, align 4
   store i32 0, i32* %i, align 4
   br label %for.cond
 
-for.cond:
+for.cond:                                         ; preds = %for.inc, %entry
   %9 = load i32, i32* %i, align 4
   %10 = load i32, i32* %times, align 4
   %lt = icmp slt i32 %9, %10
   br i1 %lt, label %for.body, label %for.exit
 
-for.body:
+for.body:                                         ; preds = %for.cond
   %11 = load i32, i32* %i, align 4
   %add = add i32 %11, 1
   store i32 %add, i32* %loop, align 4
@@ -104,14 +101,12 @@ for.body:
   %13 = call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([11 x i8], [11 x i8]* @.str.1, i32 0, i32 0), i32 %12)
   br label %for.inc
 
-for.inc:
+for.inc:                                          ; preds = %for.body
   %14 = load i32, i32* %i, align 4
   %add1 = add i32 %14, 1
   store i32 %add1, i32* %i, align 4
   br label %for.cond
 
-for.exit:
-  br label %expr_block.exit2
-
-expr_block.exit2:
+for.exit:                                         ; preds = %for.cond
   ret void
+}
diff --git a/test/test_suite/macros/type_params.c3t b/test/test_suite/macros/type_params.c3t
index 4354361a5..2af84ec15 100644
--- a/test/test_suite/macros/type_params.c3t
+++ b/test/test_suite/macros/type_params.c3t
@@ -19,48 +19,24 @@ fn void test()
 /* #expect: type_params.ll
 
   %x = alloca i32, align 4
-  %blockret = alloca i32, align 4
   %a = alloca i32, align 4
   %d = alloca double, align 8
-  %blockret1 = alloca double, align 8
-  %a2 = alloca double, align 8
+  %a1 = alloca double, align 8
   %d2 = alloca double, align 8
-  %blockret4 = alloca double, align 8
-  %a5 = alloca double, align 8
+  %a2 = alloca double, align 8
   %z = alloca i16, align 2
-  %blockret7 = alloca i16, align 2
-  %a8 = alloca i16, align 2
+  %a3 = alloca i16, align 2
   store i32 0, i32* %a, align 4
   %0 = load i32, i32* %a, align 4
-  store i32 %0, i32* %blockret, align 4
-  br label %expr_block.exit
-
-expr_block.exit:                                  ; preds = %entry
-  %1 = load i32, i32* %blockret, align 4
-  store i32 %1, i32* %x, align 4
+  store i32 %0, i32* %x, align 4
+  store double 0.000000e+00, double* %a1, align 8
+  %1 = load double, double* %a1, align 8
+  store double %1, double* %d, align 8
   store double 0.000000e+00, double* %a2, align 8
   %2 = load double, double* %a2, align 8
-  store double %2, double* %blockret1, align 8
-  br label %expr_block.exit3
-
-expr_block.exit3:                                 ; preds = %expr_block.exit
-  %3 = load double, double* %blockret1, align 8
-  store double %3, double* %d, align 8
-  store double 0.000000e+00, double* %a5, align 8
-  %4 = load double, double* %a5, align 8
-  store double %4, double* %blockret4, align 8
-  br label %expr_block.exit6
-
-expr_block.exit6:                                 ; preds = %expr_block.exit3
-  %5 = load double, double* %blockret4, align 8
-  store double %5, double* %d2, align 8
-  store i16 0, i16* %a8, align 2
-  %6 = load i16, i16* %a8, align 2
-  store i16 %6, i16* %blockret7, align 2
-  br label %expr_block.exit9
-
-expr_block.exit9:                                 ; preds = %expr_block.exit6
-  %7 = load i16, i16* %blockret7, align 2
-  store i16 %7, i16* %z, align 2
+  store double %2, double* %d2, align 8
+  store i16 0, i16* %a3, align 2
+  %3 = load i16, i16* %a3, align 2
+  store i16 %3, i16* %z, align 2
   ret void
 }
diff --git a/test/test_suite/macros/userland_bitcast.c3t b/test/test_suite/macros/userland_bitcast.c3t
index 36e230cf1..6cc4c1830 100644
--- a/test/test_suite/macros/userland_bitcast.c3t
+++ b/test/test_suite/macros/userland_bitcast.c3t
@@ -87,7 +87,6 @@ entry:
   %x = alloca i16, align 2
   %z = alloca %Foo, align 2
   %expr = alloca %Foo, align 2
-  %blockret = alloca i64, align 8
   %x1 = alloca i64, align 8
   %b = alloca i16*, align 8
   %to = alloca i16*, align 8
@@ -131,12 +130,7 @@ for.inc:                                          ; preds = %for.body
 
 for.exit:                                         ; preds = %for.cond
   %12 = load i64, i64* %x1, align 8
-  store i64 %12, i64* %blockret, align 8
-  br label %expr_block.exit
-
-expr_block.exit:                                  ; preds = %for.exit
-  %13 = load i64, i64* %blockret, align 8
-  ret i64 %13
+  ret i64 %12
 }
 
 ; Function Attrs: nounwind
@@ -144,7 +138,6 @@ define i32 @userland_bitcast.test(float %0) #0 {
 entry:
   %x = alloca float, align 4
   %expr = alloca float, align 4
-  %blockret = alloca [4 x i8], align 1
   %x1 = alloca [4 x i8], align 1
   %b = alloca i8*, align 8
   %to = alloca i8*, align 8
@@ -183,17 +176,11 @@ for.inc:                                          ; preds = %for.body
   br label %for.cond
 
 for.exit:                                         ; preds = %for.cond
-  %9 = bitcast [4 x i8]* %blockret to i8*
+  %9 = bitcast i32* %tempcoerce to i8*
   %10 = bitcast [4 x i8]* %x1 to i8*
-  call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %9, i8* align 1 %10, i32 4, i1 false)
-  br label %expr_block.exit
-
-expr_block.exit:                                  ; preds = %for.exit
-  %11 = bitcast i32* %tempcoerce to i8*
-  %12 = bitcast [4 x i8]* %blockret to i8*
-  call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %11, i8* align 1 %12, i32 4, i1 false)
-  %13 = load i32, i32* %tempcoerce, align 4
-  ret i32 %13
+  call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %9, i8* align 1 %10, i32 4, i1 false)
+  %11 = load i32, i32* %tempcoerce, align 4
+  ret i32 %11
 }
 
 ; Function Attrs: nounwind
@@ -202,33 +189,29 @@ entry:
   %f = alloca float, align 4
   %i = alloca i32, align 4
   %expr = alloca float, align 4
-  %blockret = alloca i32, align 4
   %x = alloca i32, align 4
   %b = alloca i32*, align 8
   %to = alloca i32*, align 8
   %i1 = alloca i64, align 8
   %f2 = alloca float, align 4
   %expr3 = alloca i32, align 4
-  %blockret4 = alloca float, align 4
-  %x5 = alloca float, align 4
-  %b6 = alloca i32*, align 8
-  %to7 = alloca i32*, align 8
-  %i9 = alloca i64, align 8
+  %x4 = alloca float, align 4
+  %b5 = alloca i32*, align 8
+  %to6 = alloca i32*, align 8
+  %i8 = alloca i64, align 8
   %d = alloca double, align 8
   %l = alloca i64, align 8
-  %expr20 = alloca double, align 8
-  %blockret21 = alloca i64, align 8
-  %x22 = alloca i64, align 8
-  %b23 = alloca i64*, align 8
-  %to25 = alloca i64*, align 8
-  %i26 = alloca i64, align 8
+  %expr18 = alloca double, align 8
+  %x19 = alloca i64, align 8
+  %b20 = alloca i64*, align 8
+  %to22 = alloca i64*, align 8
+  %i23 = alloca i64, align 8
   %d2 = alloca double, align 8
-  %expr36 = alloca double, align 8
-  %blockret37 = alloca double, align 8
-  %x38 = alloca double, align 8
-  %b39 = alloca i64*, align 8
-  %to41 = alloca i64*, align 8
-  %i43 = alloca i64, align 8
+  %expr32 = alloca double, align 8
+  %x33 = alloca double, align 8
+  %b34 = alloca i64*, align 8
+  %to36 = alloca i64*, align 8
+  %i38 = alloca i64, align 8
   store float 0x4028B4BC60000000, float* %f, align 4
   %0 = load float, float* %f, align 4
   store float %0, float* %expr, align 4
@@ -262,137 +245,117 @@ for.inc:                                          ; preds = %for.body
 
 for.exit:                                         ; preds = %for.cond
   %8 = load i32, i32* %x, align 4
-  store i32 %8, i32* %blockret, align 4
-  br label %expr_block.exit
+  store i32 %8, i32* %i, align 4
+  %9 = load i32, i32* %i, align 4
+  store i32 %9, i32* %expr3, align 4
+  store i32* %expr3, i32** %b5, align 8
+  %ptrptr7 = bitcast float* %x4 to i32*
+  store i32* %ptrptr7, i32** %to6, align 8
+  store i64 0, i64* %i8, align 8
+  br label %for.cond9
 
-expr_block.exit:                                  ; preds = %for.exit
-  %9 = load i32, i32* %blockret, align 4
-  store i32 %9, i32* %i, align 4
-  %10 = load i32, i32* %i, align 4
-  store i32 %10, i32* %expr3, align 4
-  store i32* %expr3, i32** %b6, align 8
-  %ptrptr8 = bitcast float* %x5 to i32*
-  store i32* %ptrptr8, i32** %to7, align 8
-  store i64 0, i64* %i9, align 8
-  br label %for.cond10
+for.cond9:                                        ; preds = %for.inc14, %for.exit
+  %10 = load i64, i64* %i8, align 8
+  %lt10 = icmp ult i64 %10, 4
+  br i1 %lt10, label %for.body11, label %for.exit16
 
-for.cond10:                                       ; preds = %for.inc15, %expr_block.exit
-  %11 = load i64, i64* %i9, align 8
-  %lt11 = icmp ult i64 %11, 4
-  br i1 %lt11, label %for.body12, label %for.exit17
+for.body11:                                       ; preds = %for.cond9
+  %11 = load i32*, i32** %to6, align 8
+  %12 = load i64, i64* %i8, align 8
+  %ptroffset12 = getelementptr inbounds i32, i32* %11, i64 %12
+  %13 = load i32*, i32** %b5, align 8
+  %14 = load i64, i64* %i8, align 8
+  %ptroffset13 = getelementptr inbounds i32, i32* %13, i64 %14
+  %15 = load i32, i32* %ptroffset13, align 4
+  store i32 %15, i32* %ptroffset12, align 4
+  br label %for.inc14
 
-for.body12:                                       ; preds = %for.cond10
-  %12 = load i32*, i32** %to7, align 8
-  %13 = load i64, i64* %i9, align 8
-  %ptroffset13 = getelementptr inbounds i32, i32* %12, i64 %13
-  %14 = load i32*, i32** %b6, align 8
-  %15 = load i64, i64* %i9, align 8
-  %ptroffset14 = getelementptr inbounds i32, i32* %14, i64 %15
-  %16 = load i32, i32* %ptroffset14, align 4
-  store i32 %16, i32* %ptroffset13, align 4
-  br label %for.inc15
+for.inc14:                                        ; preds = %for.body11
+  %16 = load i64, i64* %i8, align 8
+  %add15 = add i64 %16, 4
+  store i64 %add15, i64* %i8, align 8
+  br label %for.cond9
 
-for.inc15:                                        ; preds = %for.body12
-  %17 = load i64, i64* %i9, align 8
-  %add16 = add i64 %17, 4
-  store i64 %add16, i64* %i9, align 8
-  br label %for.cond10
-
-for.exit17:                                       ; preds = %for.cond10
-  %18 = load float, float* %x5, align 4
-  store float %18, float* %blockret4, align 4
-  br label %expr_block.exit18
-
-expr_block.exit18:                                ; preds = %for.exit17
-  %19 = load float, float* %blockret4, align 4
-  store float %19, float* %f2, align 4
-  %20 = load float, float* %f, align 4
-  %fpfpext = fpext float %20 to double
-  %21 = load i32, i32* %i, align 4
-  %22 = load float, float* %f2, align 4
-  %fpfpext19 = fpext float %22 to double
-  call void (i8*, ...) @printf(i8* getelementptr inbounds ([16 x i8], [16 x i8]* @.str, i32 0, i32 0), double %fpfpext, i32 %21, double %fpfpext19)
+for.exit16:                                       ; preds = %for.cond9
+  %17 = load float, float* %x4, align 4
+  store float %17, float* %f2, align 4
+  %18 = load float, float* %f, align 4
+  %fpfpext = fpext float %18 to double
+  %19 = load i32, i32* %i, align 4
+  %20 = load float, float* %f2, align 4
+  %fpfpext17 = fpext float %20 to double
+  call void (i8*, ...) @printf(i8* getelementptr inbounds ([16 x i8], [16 x i8]* @.str, i32 0, i32 0), double %fpfpext, i32 %19, double %fpfpext17)
   store double 1.235300e+268, double* %d, align 8
-  %23 = load double, double* %d, align 8
-  store double %23, double* %expr20, align 8
-  %ptrptr24 = bitcast double* %expr20 to i64*
-  store i64* %ptrptr24, i64** %b23, align 8
-  store i64* %x22, i64** %to25, align 8
-  store i64 0, i64* %i26, align 8
-  br label %for.cond27
+  %21 = load double, double* %d, align 8
+  store double %21, double* %expr18, align 8
+  %ptrptr21 = bitcast double* %expr18 to i64*
+  store i64* %ptrptr21, i64** %b20, align 8
+  store i64* %x19, i64** %to22, align 8
+  store i64 0, i64* %i23, align 8
+  br label %for.cond24
 
-for.cond27:                                       ; preds = %for.inc32, %expr_block.exit18
-  %24 = load i64, i64* %i26, align 8
-  %lt28 = icmp ult i64 %24, 8
-  br i1 %lt28, label %for.body29, label %for.exit34
+for.cond24:                                       ; preds = %for.inc29, %for.exit16
+  %22 = load i64, i64* %i23, align 8
+  %lt25 = icmp ult i64 %22, 8
+  br i1 %lt25, label %for.body26, label %for.exit31
 
-for.body29:                                       ; preds = %for.cond27
-  %25 = load i64*, i64** %to25, align 8
-  %26 = load i64, i64* %i26, align 8
-  %ptroffset30 = getelementptr inbounds i64, i64* %25, i64 %26
-  %27 = load i64*, i64** %b23, align 8
-  %28 = load i64, i64* %i26, align 8
-  %ptroffset31 = getelementptr inbounds i64, i64* %27, i64 %28
-  %29 = load i64, i64* %ptroffset31, align 8
-  store i64 %29, i64* %ptroffset30, align 8
-  br label %for.inc32
+for.body26:                                       ; preds = %for.cond24
+  %23 = load i64*, i64** %to22, align 8
+  %24 = load i64, i64* %i23, align 8
+  %ptroffset27 = getelementptr inbounds i64, i64* %23, i64 %24
+  %25 = load i64*, i64** %b20, align 8
+  %26 = load i64, i64* %i23, align 8
+  %ptroffset28 = getelementptr inbounds i64, i64* %25, i64 %26
+  %27 = load i64, i64* %ptroffset28, align 8
+  store i64 %27, i64* %ptroffset27, align 8
+  br label %for.inc29
 
-for.inc32:                                        ; preds = %for.body29
-  %30 = load i64, i64* %i26, align 8
-  %add33 = add i64 %30, 8
-  store i64 %add33, i64* %i26, align 8
-  br label %for.cond27
+for.inc29:                                        ; preds = %for.body26
+  %28 = load i64, i64* %i23, align 8
+  %add30 = add i64 %28, 8
+  store i64 %add30, i64* %i23, align 8
+  br label %for.cond24
 
-for.exit34:                                       ; preds = %for.cond27
-  %31 = load i64, i64* %x22, align 8
-  store i64 %31, i64* %blockret21, align 8
-  br label %expr_block.exit35
+for.exit31:                                       ; preds = %for.cond24
+  %29 = load i64, i64* %x19, align 8
+  store i64 %29, i64* %l, align 8
+  %30 = load double, double* %d, align 8
+  store double %30, double* %expr32, align 8
+  %ptrptr35 = bitcast double* %expr32 to i64*
+  store i64* %ptrptr35, i64** %b34, align 8
+  %ptrptr37 = bitcast double* %x33 to i64*
+  store i64* %ptrptr37, i64** %to36, align 8
+  store i64 0, i64* %i38, align 8
+  br label %for.cond39
 
-expr_block.exit35:                                ; preds = %for.exit34
-  %32 = load i64, i64* %blockret21, align 8
-  store i64 %32, i64* %l, align 8
-  %33 = load double, double* %d, align 8
-  store double %33, double* %expr36, align 8
-  %ptrptr40 = bitcast double* %expr36 to i64*
-  store i64* %ptrptr40, i64** %b39, align 8
-  %ptrptr42 = bitcast double* %x38 to i64*
-  store i64* %ptrptr42, i64** %to41, align 8
-  store i64 0, i64* %i43, align 8
-  br label %for.cond44
+for.cond39:                                       ; preds = %for.inc44, %for.exit31
+  %31 = load i64, i64* %i38, align 8
+  %lt40 = icmp ult i64 %31, 8
+  br i1 %lt40, label %for.body41, label %for.exit46
 
-for.cond44:                                       ; preds = %for.inc49, %expr_block.exit35
-  %34 = load i64, i64* %i43, align 8
-  %lt45 = icmp ult i64 %34, 8
-  br i1 %lt45, label %for.body46, label %for.exit51
+for.body41:                                       ; preds = %for.cond39
+  %32 = load i64*, i64** %to36, align 8
+  %33 = load i64, i64* %i38, align 8
+  %ptroffset42 = getelementptr inbounds i64, i64* %32, i64 %33
+  %34 = load i64*, i64** %b34, align 8
+  %35 = load i64, i64* %i38, align 8
+  %ptroffset43 = getelementptr inbounds i64, i64* %34, i64 %35
+  %36 = load i64, i64* %ptroffset43, align 8
+  store i64 %36, i64* %ptroffset42, align 8
+  br label %for.inc44
 
-for.body46:                                       ; preds = %for.cond44
-  %35 = load i64*, i64** %to41, align 8
-  %36 = load i64, i64* %i43, align 8
-  %ptroffset47 = getelementptr inbounds i64, i64* %35, i64 %36
-  %37 = load i64*, i64** %b39, align 8
-  %38 = load i64, i64* %i43, align 8
-  %ptroffset48 = getelementptr inbounds i64, i64* %37, i64 %38
-  %39 = load i64, i64* %ptroffset48, align 8
-  store i64 %39, i64* %ptroffset47, align 8
-  br label %for.inc49
+for.inc44:                                        ; preds = %for.body41
+  %37 = load i64, i64* %i38, align 8
+  %add45 = add i64 %37, 8
+  store i64 %add45, i64* %i38, align 8
+  br label %for.cond39
 
-for.inc49:                                        ; preds = %for.body46
-  %40 = load i64, i64* %i43, align 8
-  %add50 = add i64 %40, 8
-  store i64 %add50, i64* %i43, align 8
-  br label %for.cond44
-
-for.exit51:                                       ; preds = %for.cond44
-  %41 = load double, double* %x38, align 8
-  store double %41, double* %blockret37, align 8
-  br label %expr_block.exit52
-
-expr_block.exit52:                                ; preds = %for.exit51
-  %42 = load double, double* %blockret37, align 8
-  store double %42, double* %d2, align 8
-  %43 = load double, double* %d, align 8
-  %44 = load i64, i64* %l, align 8
-  %45 = load double, double* %d2, align 8
-  call void (i8*, ...) @printf(i8* getelementptr inbounds ([18 x i8], [18 x i8]* @.str.1, i32 0, i32 0), double %43, i64 %44, double %45)
+for.exit46:                                       ; preds = %for.cond39
+  %38 = load double, double* %x33, align 8
+  store double %38, double* %d2, align 8
+  %39 = load double, double* %d, align 8
+  %40 = load i64, i64* %l, align 8
+  %41 = load double, double* %d2, align 8
+  call void (i8*, ...) @printf(i8* getelementptr inbounds ([18 x i8], [18 x i8]* @.str.1, i32 0, i32 0), double %39, i64 %40, double %41)
   ret void
 }
\ No newline at end of file
diff --git a/test/test_suite/statements/foreach_custom_macro.c3t b/test/test_suite/statements/foreach_custom_macro.c3t
index e28d6c0fc..99b3f9eb5 100644
--- a/test/test_suite/statements/foreach_custom_macro.c3t
+++ b/test/test_suite/statements/foreach_custom_macro.c3t
@@ -45,11 +45,10 @@ entry:
   %f = alloca i32, align 4
   %.iterator = alloca %FooIterator, align 8
   %f1 = alloca %Foo*, align 8
-  %blockret = alloca %FooIterator, align 8
   %literal = alloca %FooIterator, align 8
   %it = alloca %FooIterator*, align 8
   %value = alloca i32*, align 8
-  %blockret2 = alloca i8, align 1
+  %blockret = alloca i8, align 1
   %0 = bitcast [3 x i32]* %i to i8*
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %0, i8* align 4 bitcast ([3 x i32]* @.__const to i8*), i32 12, i1 false)
   %1 = getelementptr inbounds %Foo, %Foo* %x, i32 0, i32 0
@@ -64,68 +63,62 @@ entry:
   %6 = getelementptr inbounds %FooIterator, %FooIterator* %literal, i32 0, i32 1
   %7 = load %Foo*, %Foo** %f1, align 8
   store %Foo* %7, %Foo** %6, align 8
-  %8 = bitcast %FooIterator* %blockret to i8*
+  %8 = bitcast %FooIterator* %.iterator to i8*
   %9 = bitcast %FooIterator* %literal to i8*
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 8 %8, i8* align 8 %9, i32 16, i1 false)
-  br label %expr_block.exit
-
-expr_block.exit:                                  ; preds = %entry
-  %10 = bitcast %FooIterator* %.iterator to i8*
-  %11 = bitcast %FooIterator* %blockret to i8*
-  call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 8 %10, i8* align 8 %11, i32 16, i1 false)
   br label %for.cond
 
-for.cond:                                         ; preds = %expr_block.exit
+for.cond:                                         ; preds = %entry
   store %FooIterator* %.iterator, %FooIterator** %it, align 8
   store i32* %f, i32** %value, align 8
-  %12 = load %FooIterator*, %FooIterator** %it, align 8
-  %13 = getelementptr inbounds %FooIterator, %FooIterator* %12, i32 0, i32 0
-  %14 = load i64, i64* %13, align 8
-  %15 = load %FooIterator*, %FooIterator** %it, align 8
-  %16 = getelementptr inbounds %FooIterator, %FooIterator* %15, i32 0, i32 1
-  %17 = load %Foo*, %Foo** %16, align 8
-  %18 = getelementptr inbounds %Foo, %Foo* %17, i32 0, i32 0
-  %19 = getelementptr inbounds %"int[]", %"int[]"* %18, i32 0, i32 1
-  %20 = load i64, i64* %19, align 8
-  %eq = icmp eq i64 %14, %20
+  %10 = load %FooIterator*, %FooIterator** %it, align 8
+  %11 = getelementptr inbounds %FooIterator, %FooIterator* %10, i32 0, i32 0
+  %12 = load i64, i64* %11, align 8
+  %13 = load %FooIterator*, %FooIterator** %it, align 8
+  %14 = getelementptr inbounds %FooIterator, %FooIterator* %13, i32 0, i32 1
+  %15 = load %Foo*, %Foo** %14, align 8
+  %16 = getelementptr inbounds %Foo, %Foo* %15, i32 0, i32 0
+  %17 = getelementptr inbounds %"int[]", %"int[]"* %16, i32 0, i32 1
+  %18 = load i64, i64* %17, align 8
+  %eq = icmp eq i64 %12, %18
   br i1 %eq, label %if.then, label %if.exit
 
 if.then:                                          ; preds = %for.cond
-  store i8 0, i8* %blockret2, align 1
-  br label %expr_block.exit3
+  store i8 0, i8* %blockret, align 1
+  br label %expr_block.exit
 
 if.exit:                                          ; preds = %for.cond
-  %21 = load i32*, i32** %value, align 8
-  %22 = load %FooIterator*, %FooIterator** %it, align 8
-  %23 = getelementptr inbounds %FooIterator, %FooIterator* %22, i32 0, i32 1
-  %24 = load %Foo*, %Foo** %23, align 8
-  %25 = getelementptr inbounds %Foo, %Foo* %24, i32 0, i32 0
-  %26 = getelementptr inbounds %"int[]", %"int[]"* %25, i32 0, i32 0
-  %27 = load i32*, i32** %26, align 8
-  %28 = load %FooIterator*, %FooIterator** %it, align 8
-  %29 = getelementptr inbounds %FooIterator, %FooIterator* %28, i32 0, i32 0
-  %30 = load i64, i64* %29, align 8
-  %add = add i64 %30, 1
-  store i64 %add, i64* %29, align 8
-  %ptroffset = getelementptr inbounds i32, i32* %27, i64 %30
-  %31 = load i32, i32* %ptroffset, align 4
-  store i32 %31, i32* %21, align 8
-  store i8 1, i8* %blockret2, align 1
-  br label %expr_block.exit3
+  %19 = load i32*, i32** %value, align 8
+  %20 = load %FooIterator*, %FooIterator** %it, align 8
+  %21 = getelementptr inbounds %FooIterator, %FooIterator* %20, i32 0, i32 1
+  %22 = load %Foo*, %Foo** %21, align 8
+  %23 = getelementptr inbounds %Foo, %Foo* %22, i32 0, i32 0
+  %24 = getelementptr inbounds %"int[]", %"int[]"* %23, i32 0, i32 0
+  %25 = load i32*, i32** %24, align 8
+  %26 = load %FooIterator*, %FooIterator** %it, align 8
+  %27 = getelementptr inbounds %FooIterator, %FooIterator* %26, i32 0, i32 0
+  %28 = load i64, i64* %27, align 8
+  %add = add i64 %28, 1
+  store i64 %add, i64* %27, align 8
+  %ptroffset = getelementptr inbounds i32, i32* %25, i64 %28
+  %29 = load i32, i32* %ptroffset, align 4
+  store i32 %29, i32* %19, align 8
+  store i8 1, i8* %blockret, align 1
+  br label %expr_block.exit
 
-expr_block.exit3:                                 ; preds = %if.exit, %if.then
-  %32 = load i8, i8* %blockret2, align 1
-  %33 = trunc i8 %32 to i1
-  br i1 %33, label %for.body, label %for.exit
+expr_block.exit:                                  ; preds = %if.exit, %if.then
+  %30 = load i8, i8* %blockret, align 1
+  %31 = trunc i8 %30 to i1
+  br i1 %31, label %for.body, label %for.exit
 
-for.body:                                         ; preds = %expr_block.exit3
-  %34 = load i32, i32* %f, align 4
-  %35 = call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str, i32 0, i32 0), i32 %34)
+for.body:                                         ; preds = %expr_block.exit
+  %32 = load i32, i32* %f, align 4
+  %33 = call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str, i32 0, i32 0), i32 %32)
   br label %while.begin
 
 while.begin:                                      ; preds = %for.body
   br label %for.exit
 
-for.exit:                                         ; preds = %while.begin, %expr_block.exit3
+for.exit:                                         ; preds = %while.begin, %expr_block.exit
   ret void
 }
\ No newline at end of file