Merge branch 'master' of lukelau.me:/srv/git/scheme

[scheme.git] / codegen.scm

diff --git a/codegen.scm b/codegen.scm
index 9b3878a2e947f3d3314c8c9c4bc96bca38a83a7f..78ebbad84bbbed8cab7cc3c013ad395d53f91eec 100644 (file)
--- a/codegen.scm
+++ b/codegen.scm
@@ -1,6 +1,8 @@
 (load "typecheck.scm")
 (load "ast.scm")
 
+(define target 'darwin)
+
 (define (emit . s)
   (begin
     (apply printf s)
@@ -47,6 +49,8 @@
   (emit "not %rax")
   (emit "andq $1, %rax"))
 
+; 'write file handle addr-string num-bytes
+     
 (define (codegen-print x si env)
   (codegen-expr x si env) ; x should be a static-string, producing a label
 
@@ -63,10 +67,17 @@
   (emit "not %rcx")      ; -%rcx = strlen + 1
   (emit "dec %rcx")
 
-  (emit "mov %rcx, %rdx") ; number of bytes
-  (emit "mov %rbx, %rsi") ; addr of string
-  (emit "mov $1, %rax") ; file handle 1 (stdout)
-  (emit "mov $1, %rdi") ; syscall 1 (write)
+  (case target
+    ('darwin
+     (emit "movq %rbx, %rsi") ; string addr
+     (emit "movq %rcx, %rdx") ; num bytes
+     (emit "movq $1, %rdi")   ; file handle (stdout)
+     (emit "movq $0x2000004, %rax")) ; syscall 4 (write)
+    ('linux
+     (emit "mov %rbx, %rsi")  ; string addr
+     (emit "mov %rcx, %rdx")  ; num bytes
+     (emit "mov $1, %rax")    ; file handle (stdout)
+     (emit "mov $1, %rdi"))) ; syscall 1 (write)
   (emit "syscall"))
 
 (define (range s n)
@@ -105,8 +116,26 @@
   (set! cur-lambda (+ 1 cur-lambda))
   (format "_lambda~a" (- cur-lambda 1)))
 
+; a closure on the heap looks like:
+; 0-x    x+0   x+4       x+12      x+20
+; label #vars var1....  var2....  var3....
+
+(define (codegen-closure label captured si env)
+  (let* ((heap-offsets (range 4 (length captured))) ; 4, 12, 20, etc.
+        (inner-si (- si (* (length captured) wordsize))))
+    (emit "movl $~a, (heap_start)")
+    (emit "add $4, (heap_start)")
+    (for-each (lambda (var-name new-offset)
+               (emit "movq ~a(%rbp), ~a(heap_start)" ; todo: do we need to copy this?
+                     (cdr (assoc var-name env))
+                     new-offset)
+               (emit "add $8, (heap_start)")
+             captured
+             stack-offsets)
+)))
                                        ; for now we can only call closures
 (define (codegen-call closure args si env)
+;  (codegen-expr closure si env)
   (when (not (eq? (ast-type closure) 'closure))
     (error #f (format "~a is not a closure" closure)))
   (let* ((captured (caddr closure))
@@ -170,6 +199,11 @@
     (emit "pop %rbp") ; restore caller's base pointer
     (emit "ret")))
 
+(define (codegen-string label)
+  (case target
+    ('darwin (emit "movq ~a@GOTPCREL(%rip), %rax" label))
+    ('linux  (emit "lea $~a, %rax" label))))
+
 (define cur-label 0)
 (define (fresh-label)
   (set! cur-label (+ 1 cur-label))
@@ -190,7 +224,7 @@
 (define (codegen-expr e si env)
   (case (ast-type e)
     ('builtin e)
-    ('closure e)
+    ('closure (codegen-closure (cadr e) (caddr e) si env))
     ('app
      (let ((callee (codegen-expr (car e) si env)))
        (case callee
@@ -215,7 +249,7 @@
     ('bool-literal (emit "movq $~a, %rax" (if e 1 0)))
     ('int-literal (emit "movq $~a, %rax" e))
     
-    ('static-string (emit "lea ~a, %rax" (cadr e))) ; move label
+    ('static-string (codegen-string (cadr e)))
 
     (else (error #f "don't know how to codegen this"))))
 
@@ -362,7 +396,9 @@
 
                                        ; exit syscall
     (emit "mov %rax, %rdi")
-    (emit "mov $60, %rax")
+    (case target
+      ('darwin (emit "movq $0x2000001, %rax"))
+      ('linux (emit "mov $60, %rax")))
     (emit "syscall")
 
     (emit ".data")
@@ -372,13 +408,21 @@
 
     (for-each emit-string-data strings)))
 
-(define (compile-to-binary program output)
+(define (compile-to-binary program output t)
+  (set! target t)
   (when (not (eq? (typecheck program) 'int)) (error #f "not an int"))
   (let ([tmp-path "/tmp/a.s"])
     (when (file-exists? tmp-path) (delete-file tmp-path))
     (with-output-to-file tmp-path
       (lambda () (codegen program)))
-    (system (format "clang -nostdlib /tmp/a.s -o ~a" output))))
+
+    (case target
+      ('darwin
+       (system "as /tmp/a.s -o /tmp/a.o")
+       (system (format "ld /tmp/a.o -e _start -macosx_version_min 10.14 -static -o ~a" output)))
+      ('linux
+       (system "as /tmp/a.s -o /tmp/a.o")
+       (system (format "ld /tmp/a.o -o ~a" output))))))
 
 ; NOTES
 ; syscalls in linux use the following arguments for syscall instruction: