| author | Christian Urban <christian.urban@kcl.ac.uk> | 
| Sat, 04 Sep 2021 14:08:00 +0100 | |
| changeset 833 | 7c3b8bb4a174 | 
| parent 819 | c9859fa5178f | 
| child 869 | 16247acc4b0e | 
| permissions | -rw-r--r-- | 
| 654 | 1  | 
// A Small LLVM Compiler for a Simple Functional Language  | 
| 644 | 2  | 
// (includes an external lexer and parser)  | 
| 645 | 3  | 
//  | 
| 813 | 4  | 
//  | 
| 645 | 5  | 
// call with  | 
6  | 
//  | 
|
| 813 | 7  | 
// amm fun_llvm.sc main fact.fun  | 
8  | 
// amm fun_llvm.sc main defs.fun  | 
|
9  | 
//  | 
|
10  | 
// or  | 
|
11  | 
//  | 
|
| 789 | 12  | 
// amm fun_llvm.sc write fact.fun  | 
13  | 
// amm fun_llvm.sc write defs.fun  | 
|
| 645 | 14  | 
//  | 
| 813 | 15  | 
// this will generate an .ll file.  | 
16  | 
//  | 
|
17  | 
// or  | 
|
| 654 | 18  | 
//  | 
| 813 | 19  | 
// amm fun_llvm.sc run fact.fun  | 
20  | 
// amm fun_llvm.sc run defs.fun  | 
|
21  | 
//  | 
|
22  | 
//  | 
|
23  | 
// You can interpret an .ll file using lli, for example  | 
|
24  | 
//  | 
|
25  | 
// lli fact.ll  | 
|
| 655 | 26  | 
//  | 
27  | 
// The optimiser can be invoked as  | 
|
28  | 
//  | 
|
29  | 
// opt -O1 -S in_file.ll > out_file.ll  | 
|
30  | 
// opt -O3 -S in_file.ll > out_file.ll  | 
|
31  | 
//  | 
|
| 813 | 32  | 
// The code produced for the various architectures can be obtain with  | 
| 655 | 33  | 
//  | 
34  | 
// llc -march=x86 -filetype=asm in_file.ll -o -  | 
|
35  | 
// llc -march=arm -filetype=asm in_file.ll -o -  | 
|
36  | 
//  | 
|
37  | 
// Producing an executable can be achieved by  | 
|
38  | 
//  | 
|
39  | 
// llc -filetype=obj in_file.ll  | 
|
40  | 
// gcc in_file.o -o a.out  | 
|
41  | 
// ./a.out  | 
|
42  | 
||
| 645 | 43  | 
|
| 789 | 44  | 
import $file.fun_tokens, fun_tokens._  | 
45  | 
import $file.fun_parser, fun_parser._  | 
|
| 626 | 46  | 
|
47  | 
||
| 
220
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
48  | 
// for generating new labels  | 
| 
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
49  | 
var counter = -1  | 
| 
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
50  | 
|
| 
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
51  | 
def Fresh(x: String) = {
 | 
| 
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
52  | 
counter += 1  | 
| 
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
53  | 
x ++ "_" ++ counter.toString()  | 
| 
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
54  | 
}  | 
| 
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
55  | 
|
| 678 | 56  | 
// Internal CPS language for FUN  | 
| 648 | 57  | 
abstract class KExp  | 
| 653 | 58  | 
abstract class KVal  | 
| 648 | 59  | 
|
| 653 | 60  | 
case class KVar(s: String) extends KVal  | 
61  | 
case class KNum(i: Int) extends KVal  | 
|
| 656 | 62  | 
case class Kop(o: String, v1: KVal, v2: KVal) extends KVal  | 
| 653 | 63  | 
case class KCall(o: String, vrs: List[KVal]) extends KVal  | 
| 655 | 64  | 
case class KWrite(v: KVal) extends KVal  | 
| 649 | 65  | 
|
| 819 | 66  | 
case class KLet(x: String, e1: KVal, e2: KExp) extends KExp {
 | 
67  | 
override def toString = s"LET $x = $e1 in \n$e2"  | 
|
| 649 | 68  | 
}  | 
| 819 | 69  | 
case class KIf(x1: String, e1: KExp, e2: KExp) extends KExp {
 | 
70  | 
  def pad(e: KExp) = e.toString.replaceAll("(?m)^", "  ")
 | 
|
71  | 
||
72  | 
override def toString =  | 
|
73  | 
     s"IF $x1\nTHEN\n${pad(e1)}\nELSE\n${pad(e2)}"
 | 
|
| 648 | 74  | 
}  | 
| 653 | 75  | 
case class KReturn(v: KVal) extends KExp  | 
| 648 | 76  | 
|
| 654 | 77  | 
|
| 655 | 78  | 
// CPS translation from Exps to KExps using a  | 
| 654 | 79  | 
// continuation k.  | 
| 653 | 80  | 
def CPS(e: Exp)(k: KVal => KExp) : KExp = e match {
 | 
81  | 
case Var(s) => k(KVar(s))  | 
|
82  | 
case Num(i) => k(KNum(i))  | 
|
83  | 
  case Aop(o, e1, e2) => {
 | 
|
84  | 
    val z = Fresh("tmp")
 | 
|
85  | 
CPS(e1)(y1 =>  | 
|
| 656 | 86  | 
CPS(e2)(y2 => KLet(z, Kop(o, y1, y2), k(KVar(z)))))  | 
| 653 | 87  | 
}  | 
88  | 
  case If(Bop(o, b1, b2), e1, e2) => {
 | 
|
89  | 
    val z = Fresh("tmp")
 | 
|
90  | 
CPS(b1)(y1 =>  | 
|
| 655 | 91  | 
CPS(b2)(y2 =>  | 
| 656 | 92  | 
KLet(z, Kop(o, y1, y2), KIf(z, CPS(e1)(k), CPS(e2)(k)))))  | 
| 653 | 93  | 
}  | 
94  | 
  case Call(name, args) => {
 | 
|
95  | 
    def aux(args: List[Exp], vs: List[KVal]) : KExp = args match {
 | 
|
96  | 
      case Nil => {
 | 
|
97  | 
          val z = Fresh("tmp")
 | 
|
98  | 
KLet(z, KCall(name, vs), k(KVar(z)))  | 
|
99  | 
}  | 
|
100  | 
case e::es => CPS(e)(y => aux(es, vs ::: List(y)))  | 
|
| 648 | 101  | 
}  | 
| 653 | 102  | 
aux(args, Nil)  | 
103  | 
}  | 
|
| 656 | 104  | 
case Sequence(e1, e2) =>  | 
| 679 | 105  | 
CPS(e1)(_ => CPS(e2)(y2 => k(y2)))  | 
| 655 | 106  | 
  case Write(e) => {
 | 
107  | 
    val z = Fresh("tmp")
 | 
|
108  | 
CPS(e)(y => KLet(z, KWrite(y), k(KVar(z))))  | 
|
109  | 
}  | 
|
| 653 | 110  | 
}  | 
111  | 
||
| 679 | 112  | 
//initial continuation  | 
| 653 | 113  | 
def CPSi(e: Exp) = CPS(e)(KReturn)  | 
114  | 
||
| 819 | 115  | 
//some testcases:  | 
116  | 
// numbers and vars  | 
|
117  | 
println(CPSi(Num(1)).toString)  | 
|
118  | 
println(CPSi(Var("z")).toString)
 | 
|
119  | 
||
120  | 
// a * 3  | 
|
| 653 | 121  | 
val e1 = Aop("*", Var("a"), Num(3))
 | 
| 819 | 122  | 
println(CPSi(e1).toString)  | 
| 653 | 123  | 
|
| 819 | 124  | 
// (a * 3) + 4  | 
| 653 | 125  | 
val e2 = Aop("+", Aop("*", Var("a"), Num(3)), Num(4))
 | 
| 819 | 126  | 
println(CPSi(e2).toString)  | 
| 653 | 127  | 
|
| 819 | 128  | 
// 2 + (a * 3)  | 
| 653 | 129  | 
val e3 = Aop("+", Num(2), Aop("*", Var("a"), Num(3)))
 | 
| 819 | 130  | 
println(CPSi(e3).toString)  | 
| 648 | 131  | 
|
| 819 | 132  | 
//(1 - 2) + (a * 3)  | 
| 653 | 133  | 
val e4 = Aop("+", Aop("-", Num(1), Num(2)), Aop("*", Var("a"), Num(3)))
 | 
| 819 | 134  | 
println(CPSi(e4).toString)  | 
| 653 | 135  | 
|
| 819 | 136  | 
// 3 + 4 ; 1 * 7  | 
137  | 
val es = Sequence(Aop("+", Num(3), Num(4)),
 | 
|
138  | 
                  Aop("*", Num(1), Num(7)))
 | 
|
139  | 
println(CPSi(es).toString)  | 
|
140  | 
||
141  | 
// if (1 == 1) then 3 else 4  | 
|
| 653 | 142  | 
val e5 = If(Bop("==", Num(1), Num(1)), Num(3), Num(4))
 | 
| 819 | 143  | 
println(CPSi(e5).toString)  | 
| 653 | 144  | 
|
| 819 | 145  | 
// if (1 == 1) then 3 + 7 else 4 * 2  | 
146  | 
val ei = If(Bop("==", Num(1), Num(1)), 
 | 
|
147  | 
                Aop("+", Num(3), Num(7)),
 | 
|
148  | 
                Aop("*", Num(4), Num(2)))
 | 
|
149  | 
println(CPSi(ei).toString)  | 
|
150  | 
||
151  | 
||
152  | 
// if (10 != 10) then e5 else 40  | 
|
| 653 | 153  | 
val e6 = If(Bop("!=", Num(10), Num(10)), e5, Num(40))
 | 
| 819 | 154  | 
println(CPSi(e6).toString)  | 
| 648 | 155  | 
|
| 653 | 156  | 
|
| 819 | 157  | 
// foo(3)  | 
158  | 
val e7 = Call("foo", List(Num(3)))
 | 
|
159  | 
println(CPSi(e7).toString)  | 
|
160  | 
||
161  | 
// foo(3 * 1, 4, 5 + 6)  | 
|
162  | 
val e8 = Call("foo", List(Aop("*", Num(3), Num(1)), 
 | 
|
163  | 
Num(4),  | 
|
164  | 
                          Aop("+", Num(5), Num(6))))
 | 
|
165  | 
println(CPSi(e8).toString)  | 
|
| 653 | 166  | 
|
| 819 | 167  | 
// a * 3 ; b + 6  | 
168  | 
val e9 = Sequence(Aop("*", Var("a"), Num(3)), 
 | 
|
169  | 
                  Aop("+", Var("b"), Num(6)))
 | 
|
170  | 
println(CPSi(e9).toString)  | 
|
| 649 | 171  | 
|
| 819 | 172  | 
|
173  | 
val e10 = Aop("*", Aop("+", Num(1), Call("foo", List(Var("a"), Num(3)))), Num(4))
 | 
|
174  | 
println(CPSi(e10).toString)  | 
|
175  | 
||
| 653 | 176  | 
|
| 648 | 177  | 
|
178  | 
||
179  | 
||
| 625 | 180  | 
// convenient string interpolations  | 
181  | 
// for instructions, labels and methods  | 
|
182  | 
import scala.language.implicitConversions  | 
|
183  | 
import scala.language.reflectiveCalls  | 
|
184  | 
||
185  | 
implicit def sring_inters(sc: StringContext) = new {
 | 
|
186  | 
def i(args: Any*): String = " " ++ sc.s(args:_*) ++ "\n"  | 
|
187  | 
def l(args: Any*): String = sc.s(args:_*) ++ ":\n"  | 
|
188  | 
def m(args: Any*): String = sc.s(args:_*) ++ "\n"  | 
|
189  | 
}  | 
|
190  | 
||
| 656 | 191  | 
// mathematical and boolean operations  | 
| 653 | 192  | 
def compile_op(op: String) = op match {
 | 
193  | 
case "+" => "add i32 "  | 
|
194  | 
case "*" => "mul i32 "  | 
|
195  | 
case "-" => "sub i32 "  | 
|
| 656 | 196  | 
case "/" => "sdiv i32 "  | 
197  | 
case "%" => "srem i32 "  | 
|
| 653 | 198  | 
case "==" => "icmp eq i32 "  | 
| 813 | 199  | 
case "<=" => "icmp sle i32 " // signed less or equal  | 
200  | 
case "<" => "icmp slt i32 " // signed less than  | 
|
| 653 | 201  | 
}  | 
| 
220
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
202  | 
|
| 813 | 203  | 
// compile K values  | 
| 653 | 204  | 
def compile_val(v: KVal) : String = v match {
 | 
205  | 
case KNum(i) => s"$i"  | 
|
206  | 
case KVar(s) => s"%$s"  | 
|
| 656 | 207  | 
case Kop(op, x1, x2) =>  | 
| 653 | 208  | 
    s"${compile_op(op)} ${compile_val(x1)}, ${compile_val(x2)}"
 | 
209  | 
case KCall(x1, args) =>  | 
|
210  | 
    s"call i32 @$x1 (${args.map(compile_val).mkString("i32 ", ", i32 ", "")})"
 | 
|
| 655 | 211  | 
case KWrite(x1) =>  | 
212  | 
    s"call i32 @printInt (i32 ${compile_val(x1)})"
 | 
|
| 653 | 213  | 
}  | 
| 648 | 214  | 
|
| 649 | 215  | 
// compile K expressions  | 
216  | 
def compile_exp(a: KExp) : String = a match {
 | 
|
| 653 | 217  | 
case KReturn(v) =>  | 
218  | 
    i"ret i32 ${compile_val(v)}"
 | 
|
219  | 
case KLet(x: String, v: KVal, e: KExp) =>  | 
|
220  | 
    i"%$x = ${compile_val(v)}" ++ compile_exp(e)
 | 
|
221  | 
  case KIf(x, e1, e2) => {
 | 
|
| 679 | 222  | 
    val if_br = Fresh("if_branch")
 | 
223  | 
    val else_br = Fresh("else_branch")
 | 
|
| 649 | 224  | 
i"br i1 %$x, label %$if_br, label %$else_br" ++  | 
225  | 
l"\n$if_br" ++  | 
|
| 653 | 226  | 
compile_exp(e1) ++  | 
| 649 | 227  | 
l"\n$else_br" ++  | 
| 653 | 228  | 
compile_exp(e2)  | 
| 
220
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
229  | 
}  | 
| 653 | 230  | 
}  | 
231  | 
||
| 655 | 232  | 
|
233  | 
val prelude = """  | 
|
234  | 
@.str = private constant [4 x i8] c"%d\0A\00"  | 
|
235  | 
||
236  | 
declare i32 @printf(i8*, ...)  | 
|
| 
220
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
237  | 
|
| 655 | 238  | 
define i32 @printInt(i32 %x) {
 | 
239  | 
%t0 = getelementptr [4 x i8], [4 x i8]* @.str, i32 0, i32 0  | 
|
| 679 | 240  | 
call i32 (i8*, ...) @printf(i8* %t0, i32 %x)  | 
| 655 | 241  | 
ret i32 %x  | 
242  | 
}  | 
|
243  | 
||
244  | 
"""  | 
|
| 653 | 245  | 
|
| 
220
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
246  | 
|
| 625 | 247  | 
// compile function for declarations and main  | 
248  | 
def compile_decl(d: Decl) : String = d match {
 | 
|
| 649 | 249  | 
  case Def(name, args, body) => { 
 | 
250  | 
    m"define i32 @$name (${args.mkString("i32 %", ", i32 %", "")}) {" ++
 | 
|
| 653 | 251  | 
compile_exp(CPSi(body)) ++  | 
| 649 | 252  | 
m"}\n"  | 
| 
221
 
824ffbf66ab4
added fun tail
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents: 
220 
diff
changeset
 | 
253  | 
}  | 
| 649 | 254  | 
  case Main(body) => {
 | 
255  | 
    m"define i32 @main() {" ++
 | 
|
| 789 | 256  | 
compile_exp(CPS(body)(_ => KReturn(KNum(0)))) ++  | 
| 649 | 257  | 
m"}\n"  | 
| 
221
 
824ffbf66ab4
added fun tail
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents: 
220 
diff
changeset
 | 
258  | 
}  | 
| 
220
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
259  | 
}  | 
| 
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
260  | 
|
| 813 | 261  | 
|
| 626 | 262  | 
// main compiler functions  | 
| 813 | 263  | 
def compile(prog: List[Decl]) : String =  | 
| 789 | 264  | 
prelude ++ (prog.map(compile_decl).mkString)  | 
265  | 
||
| 
220
 
141041fc76b5
added
 
Christian Urban <christian dot urban at kcl dot ac dot uk> 
parents:  
diff
changeset
 | 
266  | 
|
| 813 | 267  | 
import ammonite.ops._  | 
268  | 
||
269  | 
||
| 789 | 270  | 
@main  | 
| 813 | 271  | 
def main(fname: String) = {
 | 
| 789 | 272  | 
val path = os.pwd / fname  | 
273  | 
    val file = fname.stripSuffix("." ++ path.ext)
 | 
|
274  | 
val tks = tokenise(os.read(path))  | 
|
275  | 
val ast = parse_tks(tks)  | 
|
| 813 | 276  | 
println(compile(ast))  | 
| 644 | 277  | 
}  | 
278  | 
||
| 789 | 279  | 
@main  | 
280  | 
def write(fname: String) = {
 | 
|
281  | 
val path = os.pwd / fname  | 
|
282  | 
    val file = fname.stripSuffix("." ++ path.ext)
 | 
|
283  | 
val tks = tokenise(os.read(path))  | 
|
284  | 
val ast = parse_tks(tks)  | 
|
| 813 | 285  | 
val code = compile(ast)  | 
| 789 | 286  | 
os.write.over(os.pwd / (file ++ ".ll"), code)  | 
| 626 | 287  | 
}  | 
288  | 
||
| 789 | 289  | 
@main  | 
290  | 
def run(fname: String) = {
 | 
|
291  | 
val path = os.pwd / fname  | 
|
292  | 
    val file = fname.stripSuffix("." ++ path.ext)
 | 
|
| 813 | 293  | 
write(fname)  | 
| 789 | 294  | 
    os.proc("llc", "-filetype=obj", file ++ ".ll").call()
 | 
| 813 | 295  | 
    os.proc("gcc", file ++ ".o", "-o", file ++ ".bin").call()
 | 
296  | 
os.proc(os.pwd / (file ++ ".bin")).call(stdout = os.Inherit)  | 
|
297  | 
println(s"done.")  | 
|
| 657 | 298  | 
}  | 
| 644 | 299  | 
|
| 657 | 300  | 
|
301  | 
||
302  | 
||
| 819 | 303  | 
// CPS functions  | 
304  | 
/*  | 
|
| 657 | 305  | 
|
| 819 | 306  | 
def fact(n: Int) : Int =  | 
307  | 
if (n == 0) 1 else n * fact(n - 1)  | 
|
308  | 
||
309  | 
fact(6)  | 
|
310  | 
||
311  | 
def factT(n: Int, acc: Int) : Int =  | 
|
312  | 
if (n == 0) acc else factT(n - 1, acc * n)  | 
|
313  | 
||
314  | 
factT(6, 1)  | 
|
315  | 
||
316  | 
def factC(n: Int, ret: Int => Int) : Int = {
 | 
|
317  | 
if (n == 0) ret(1)  | 
|
318  | 
else factC(n - 1, x => ret(x * n))  | 
|
319  | 
}  | 
|
320  | 
||
321  | 
factC(6, x => x)  | 
|
322  | 
factC(6, x => {println(s"The final Result is $x") ; 0})
 | 
|
323  | 
factC(6, _ + 1)  | 
|
324  | 
||
325  | 
def fibC(n: Int, ret: Int => Int) : Int = {
 | 
|
326  | 
if (n == 0 || n == 1) ret(1)  | 
|
327  | 
else fibC(n - 1, x => fibC(n - 2, y => ret(x + y)))  | 
|
328  | 
}  | 
|
329  | 
||
330  | 
fibC(10, x => {println(s"Result: $x") ; 1})
 | 
|
331  | 
||
332  | 
||
333  | 
*/  |