X-Git-Url: https://git.lukelau.me/?p=scheme.git;a=blobdiff_plain;f=codegen.scm;h=dd7950765c9ef375d443a110788a9146079fad20;hp=52df26ece76527068fd89a7c397c3ca632aeaef4;hb=d6b60d54bc90e5ebdc643d05d0b806ae7fd8aa7c;hpb=6b60f721c4be97e0a79e1b16028a5d180eb0ba1e

diff --git a/codegen.scm b/codegen.scm
index 52df26e..dd79507 100644
--- a/codegen.scm
+++ b/codegen.scm
@@ -9,6 +9,58 @@
     (apply printf s)
     (display "\n")))
 
+(define wordsize 8)
+
+(define (stack-type? data-layouts type)
+  (if (assoc type data-layouts) #t #f))
+
+
+(define (type-size data-layouts type)
+
+  (define (adt-size adt)
+    (let ([sizes
+	   (map (lambda (sum)
+		  (fold-left (lambda (acc x) (+ acc (type-size data-layouts x)))
+			     wordsize ; one word needed to store tag
+			     (cdr sum)))
+		(cdr adt))])
+      (apply max sizes)))
+  
+  (case type
+    ['Int wordsize]
+    ['Bool wordsize]
+    [else
+     (let ([adt (assoc type data-layouts)])
+       (if adt
+	   (adt-size adt)
+	   (error #f "unknown size" type)))]))
+
+					; returns the size of an expression's result in bytes
+(define (expr-size e)
+  (if (eqv? (ast-type e) 'stack)
+      (cadr e)
+      wordsize))
+
+(define (on-stack? expr)
+  (case (ast-type expr)
+    ['stack (cadr expr)]
+    [else #f]))
+
+; does a movsq for something on the stack
+(define (emit-stack-copy src dst size)
+  (emit "leaq ~a(%rbp), %rsi" (- src size))
+  (emit "leaq ~a(%rbp), %rdi" (- dst size))
+  (emit "movq $~a, %rcx" (/ size wordsize))
+  (emit "rep movsq"))	  
+  
+
+					; an environment consists of adt layouts in scope,
+					; and any bound variables.
+					; bound variables are an assoc list with their stack offset
+(define make-env list)
+(define env-data-layouts car)
+(define env-bindings cadr)
+
 (define (codegen-add xs si env)
   (define (go ys)
     (if (null? ys)
@@ -76,13 +128,6 @@
     ('linux  (emit "mov $1, %rax"))) ; syscall 1 (write)
   (emit "syscall"))
 
-(define (range s n)
-  (if (= 0 n) '()
-      (append (range s (- n 1))
-	      (list (+ s (- n 1))))))
-
-(define wordsize 8)
-
 (define (codegen-let bindings body si env)
 
 					; is this a closure that captures itself?
@@ -91,50 +136,111 @@
     (and (eqv? (ast-type expr) 'closure)
 	 (memv name (caddr expr))))
 
-  (let* ((stack-offsets (map (lambda (name x) ; assoc map of binding name to offset
-			       (cons name (- si (* x wordsize))))
-			     (map car bindings)
-			     (range 0 (length bindings))))
-	 (inner-si (- si (* (length bindings) wordsize)))
-
-	 (get-offset (lambda (n) (cdr (assoc n stack-offsets))))
+  ;; (define (emit-scc scc env)
+  ;;   ; acc is a pair of the env and list of touchups
+  ;;   (define (emit-binding acc binding)
+  ;;     (let ([binding-name (car binding)]
+  ;; 	    [binding-body (cadr binding)]
+
+  ;; 	    [other-bindings (filter
+  ;; 			     (lambda (x) (not (eqv? binding-name x)))
+  ;; 			     scc)]
+  ;; 	    [mutually-recursives
+  ;; 	     (filter
+  ;; 	      (lambda (other-binding)
+  ;; 		(memv other-binding (references binding-body)))
+  ;; 	      other-bindings)]
+
+  ;; 	    [new-touchups (append touchups (cdr acc))])
+
+  ;; 					; TODO: assert that the only mutually recursives are closures
+  ;; 	(for-each
+  ;; 	 (lambda (binding)
+  ;; 	   (when (not (eqv? (ast-type (cadr binding))
+  
+  ;; 	(emit "asdf")
+  ;; 	(cons new-env new-touchups)
+  ;; 	))
+
+  ;;   (fold-left emit-binding (cons env '()) scc))))
+					; assoc map of binding name to size
+  (define stack-sizes
+    (map (lambda (binding) (cons (car binding) (expr-size (cadr binding))))
+	 bindings))
+
+					; assoc map of binding name to offset
+  (define stack-offsets
+					; 2  4  2  8  6
+    (let* ([totals                      ; 2  6  8  16 22
+	    (reverse (fold-left (lambda (acc x)
+				  (if (null? acc)
+				      (list x)
+				      (cons (+ x (car acc)) acc)))
+				'()
+				(map cdr stack-sizes)))]
+					; 0  2  6  8  16
+	   [relative-offsets (map - totals (map cdr stack-sizes))]
+	   [absolute-offsets (map (lambda (x) (- si x)) relative-offsets)])
+      (map cons (map car stack-sizes) absolute-offsets)))
+  
+  (let* (
+					; the stack index used when codegening binding body and main body
+					; -> stack ->
+					; [stack-offsets | inner-si]
+	 [inner-si (- si (fold-left + 0 (map cdr stack-sizes)))]
+
+	 [get-offset (lambda (n) (cdr (assoc n stack-offsets)))]
 	 
 	 [inner-env
 	  (fold-left
 	   (lambda (env comps)
-	     (let ([scc-env
+	     (let* ([scc-binding-offsets
 		     (fold-left
 		      (lambda (acc name)
 			(cons (cons name (get-offset name))
 			      acc))
-		     env
-		     comps)])
+		      (env-bindings env)
+		      comps)]
+		    [scc-env (make-env (env-data-layouts env) scc-binding-offsets)])
 	       (for-each 
 		(lambda (name)
-		  (let ([expr (cadr (assoc name bindings))])
+		  (let* ([expr (cadr (assoc name bindings))]
+			 [size (expr-size expr)])
 		    (emit "## generating ~a with scc-env ~a" name scc-env)
 		    (if (self-captive-closure? name expr)
 					; if self-captive, insert a flag into the environment to let
 					; codegen-closure realise this!
 			(codegen-expr expr
 				      inner-si
+				      (make-env
+				       (env-data-layouts scc-env)
 				       (cons (cons name 'self-captive)
-					    scc-env))
+					     (env-bindings scc-env))))
 			(codegen-expr expr inner-si scc-env))
-		    (emit "movq %rax, ~a(%rbp)" (get-offset name))))
+
+		    (if (on-stack? expr)
+						  ; copy over whatevers on the stack
+			(emit-stack-copy inner-si (get-offset name) size)
+			(emit "movq %rax, ~a(%rbp)" (get-offset name)))))
 		comps)
 	       scc-env))
-	   env (reverse (sccs (graph bindings))))])
+	   env
+	   (reverse (sccs (graph bindings))))])
     
     (for-each (lambda (form)
 		(codegen-expr form inner-si inner-env))
 	      body)))
 
-(define (codegen-var name si env)
-  (when (not (assoc name env))
+(define (codegen-var e si env)
+  (let* ([stack-size (on-stack? e)]
+	 [name (if (on-stack? e) (caddr e) e)]
+	 [stack-offset (cdr (assoc name (env-bindings env)))])
+    (when (not stack-offset)
       (error #f (format "Variable ~a is not bound" name)))
-  (let ((offset (cdr (assoc name env))))
-    (emit "movq ~a(%rbp), %rax" offset)))
+
+    (if (on-stack? e)
+	(emit-stack-copy stack-offset si stack-size)
+	(emit "movq ~a(%rbp), %rax" stack-offset))))
 
 (define cur-lambda 0)
 (define (fresh-lambda)
@@ -168,7 +274,7 @@
 					; store the captured vars
     (for-each
      (lambda (var-name heap-offset)
-       (let ([stack-offset (cdr (assoc var-name env))])
+       (let ([stack-offset (cdr (assoc var-name (env-bindings env)))])
 	 (emit "### captive ~a" var-name)
 	 (if (eqv? stack-offset 'self-captive)
 					; captive refers to this closure:
@@ -243,7 +349,8 @@
 			       (* (- wordsize) (+ 1 i)))
 			     (range 0 (length params))))
 
-	 (env (map cons params stack-offsets)))
+	 [bindings (map cons params stack-offsets)]
+	 [env (make-env '() bindings)])
     (emit "~a:" label)
 
     (display "## lambda captives: ")
@@ -301,6 +408,99 @@
     (codegen-expr else si env)
     (emit "~a:" exit-label)))
 
+(define (data-tor env e)
+  (if (not (list? e)) #f    
+      (assoc (car e) (flat-map data-tors (env-data-layouts env)))))
+
+					; returns the internal offset in bytes of a product within an ADT
+					; given the constructor layout
+					; constructor-layout: (foo (Int Bool))
+(define (data-product-offset data-layouts type sum index)
+  (let* ([products (cdr (assoc sum (cdr (assoc type data-layouts))))]
+	 [to-traverse (list-head products index)])
+    (fold-left
+     (lambda (acc t) (+ acc (type-size data-layouts t)))
+     wordsize ; skip the tag in the first word
+     to-traverse)))
+
+(define (data-sum-tag data-layouts type sum)
+
+  (define (go acc sums)
+    (when (null? sums) (error #f "data-sum-tag no sum for type" sum type))
+    (if (eqv? sum (car sums))
+	acc
+	(go (+ 1 acc) (cdr sums))))
+  (let* ([type-sums (cdr (assoc type data-layouts))])
+    (go 0 (map car type-sums))))
+
+(define (codegen-data-tor e si env)
+
+  (define dls (env-data-layouts env))
+
+  (define (codegen-destructor tor)
+    (let* ([res (codegen-expr (cadr e) si env)]
+	   [info (cadr tor)]
+	   [type (car info)]
+	   [sum (cadr info)]
+   	   [index (caddr info)]
+   	   [product-type (cadddr info)]
+	   [product-type-size (type-size dls product-type)]
+
+	   [safe-space-offset (- si (type-size dls type))]
+
+	   [inner-offset (- si (data-product-offset dls type sum index))])
+      
+      (when (not (on-stack? (cadr e)))
+	(error #f "trying to destruct something that isn't a stack expression"))      
+      (emit "# deconstructing")
+
+      (if (stack-type? (env-data-layouts env) product-type)
+					; if copying from the stack, need to first copy
+					; to a safe space above to avoid overwriting
+					; the original result on the stack
+					; this is bad. please remove this in the rewrite.
+	  (begin
+	    (emit-stack-copy inner-offset safe-space-offset product-type-size)
+	    (emit-stack-copy safe-space-offset si product-type-size))
+	  (emit "movq ~a(%rbp), %rax" inner-offset))))
+
+  (define (codegen-constructor tor)
+    (let* ([info (cadr tor)]
+	   [type (car info)]
+	   [sum (cadr info)]
+	   [constructor (car e)]
+
+	   [args (cdr e)]
+
+	   [tag (data-sum-tag (env-data-layouts env)
+			      type
+			      sum)]
+
+	   [inner-si (- si (type-size dls type))]
+
+	   [product-types (cdr (assoc sum (cdr (assoc type dls))))]
+	   
+	   [insert-product
+	    (lambda (expr i product-type)
+	      (let ([dest-offset
+		     (- si (data-product-offset dls type sum i))]
+		    [product-size (type-size dls product-type)])
+		(codegen-expr expr inner-si env)
+		(if (on-stack? expr)
+		    (emit-stack-copy inner-si dest-offset product-size)
+		    (emit "movq %rax, ~a(%rbp)" dest-offset))))])
+	   
+					; emit the tag
+      (emit "movq $~a, ~a(%rbp)" tag si)     
+					; generate products
+      (for-each insert-product args (range 0 (length args)) product-types)))
+  
+  (let* ([tor (data-tor env e)]
+	 [constructor (eqv? 'constructor (caddr (cadr tor)))])
+    (if constructor
+	(codegen-constructor tor)
+	(codegen-destructor tor))))
+
 (define (codegen-expr e si env)
   (emit "# ~a" e)
   (case (ast-type e)
@@ -314,10 +514,14 @@
        ('= (codegen-eq  (cadr e) (caddr e) si env))
        ('bool->int (codegen-expr (cadr e) si env))
        ('print (codegen-print (cadr e) si env))
-       (else (codegen-call (car e) (cdr e) si env))))
+       (else
+	(if (data-tor env e)
+	    (codegen-data-tor e si env)
+	    (codegen-call (car e) (cdr e) si env)))))
 
 					; this is a builtin being passed around as a variable
-    ('builtin (emit "movq $~a, %rax" (builtin-id e)))
+					; this should have been converted to a closure!
+    ('builtin (error #f "passing about a builtin!" e))
 
     ('let (codegen-let (let-bindings e)
 		       (let-body e)
@@ -334,9 +538,28 @@
     ('static-string (emit "movq ~a@GOTPCREL(%rip), %rax"
 			  (cadr e)))
 
-    (else (error #f "don't know how to codegen this"))))
-
-
+    ('stack (case (ast-type (caddr e))
+	      ['var (codegen-var e si env)]
+	      [else (codegen-expr (caddr e) si env)]))
+
+    (else (error #f "don't know how to codegen this")))
+  (emit "# done ~a" e))
+
+					; takes in a expr annotated with types and returns a type-less AST
+					; with stack values wrapped
+(define (annotate-stack-values data-layouts ann-e)
+  (define (stack-type? type)
+    (assoc type data-layouts))
+  (define (strip e)
+    (ast-traverse strip (ann-expr e)))
+  (let* ([e (ann-expr ann-e)]
+	 [type (ann-type ann-e)])
+    (if (stack-type? type)
+	`(stack ,(type-size data-layouts type)
+		,(ast-traverse (lambda (x) (annotate-stack-values data-layouts x)) e))
+	(ast-traverse (lambda (x)
+			(annotate-stack-values data-layouts x))
+		      e))))
 
 (define (free-vars prog)
   (define bound '())
@@ -349,15 +572,15 @@
 	    (set! bound (append (lambda-args e) bound))
 	    (collect (lambda-body e))))
 
-      ('app (fold-map collect e))
-      ('if (fold-map collect (cdr e)))
+      ('app (flat-map collect e))
+      ('if (flat-map collect (cdr e)))
       ('let
-	  (let ([bind-fvs (fold-map (lambda (a)
+	  (let ([bind-fvs (flat-map (lambda (a)
 				      (begin
 					(set! bound (cons (car a) bound))
 					(collect (cdr a))))
 				    (let-bindings e))])
-	    (append bind-fvs (fold-map collect (let-body e)))))
+	    (append bind-fvs (flat-map collect (let-body e)))))
       (else '())))
   (collect prog))
 
@@ -454,29 +677,6 @@
   (emit "~a:" (car s))
   (emit "\t.string \"~a\"" (cdr s)))
 
-;; (define (amd64-abi f)
-;; 					; preserve registers
-;;   (emit "push %rbp")
-;;   ;; (emit "push %rbx")
-;;   ;; (for-each (lambda (i)
-;;   ;; 	      (emit (string-append
-;;   ;; 		     "push %r"
-;;   ;; 		     (number->string i))))
-;;   ;; 	    '(12 13 14 15))
-
-;;   (emit "movq %rsp, %rbp")              ; set up the base pointer
-
-;;   (f) ; call stuff
-;; 					; restore preserved registers
-;;   ;; (for-each (lambda (i)
-;;   ;; 	      (emit (string-append
-;;   ;; 		     "pop %r"
-;;   ;; 		     (number->string i))))
-;;   ;; 	    '(15 14 13 12))
-;;   ;; (emit "pop %rbx")
-;;   (emit "pop %rbp")
-;;   (emit "ret"))
-
 					; 24(%rbp) mem arg 1
 					; 16(%rbp) mem arg 0          prev frame
 					; -----------------------
@@ -523,11 +723,18 @@
 (define (codegen program)
   (set! cur-label 0)
   (set! cur-lambda 0)
-  (let* ((extract-res-0 (extract-strings program))
-	 (strings (car extract-res-0))
-	 (extract-res-1 (extract-lambdas (cdr extract-res-0)))
-	 (lambdas (car extract-res-1))
-	 (xform-prog (cdr extract-res-1)))
+  (let* ([data-layouts (program-data-layouts program)]
+
+	 [pattern-matched (expand-pattern-matches program)]
+	 [type-annotated (annotate-types pattern-matched)]
+	 [stack-annotated (annotate-stack-values data-layouts
+						 type-annotated)]
+	 
+	 (strings-res (extract-strings stack-annotated))
+	 (strings (car strings-res))
+	 (lambdas-res (extract-lambdas (cdr strings-res)))
+	 (lambdas (car lambdas-res))
+	 (xform-prog (cdr lambdas-res)))
 
     (emit "\t.global _start")
     (emit "\t.text")
@@ -541,7 +748,7 @@
 
     (emit "movq %rsp, %rbp")            ; set up the base pointer
     
-    (codegen-expr xform-prog wordsize '())
+    (codegen-expr xform-prog (- wordsize) (make-env data-layouts '()))
 
 					; exit syscall
     (emit "mov %rax, %rdi")
@@ -559,7 +766,7 @@
 
 (define (compile-to-binary program output t)
   (set! target t)
-  (when (not (eq? (typecheck program) 'int)) (error #f "not an int"))
+  (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
@@ -583,6 +790,6 @@
 ; %r8  = 5th arg
 ; %r9  = 6th arg
 
-; on darwin, the syscall is offset by 0x2000000
+; on darwin, unix/posix syscalls are offset by 0x2000000 (syscall classes)
 ; https://opensource.apple.com/source/xnu/xnu-2782.20.48/bsd/kern/syscalls.master
 ; documentation for most syscalls: /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk/usr/include/sys