X-Git-Url: https://git.lukelau.me/?p=scheme.git;a=blobdiff_plain;f=codegen.scm;h=dededbec01e0856b19703d8945e096e49b316aa5;hp=10e005f8ad90e840aad396da4a0cdb5032163f6c;hb=8e106ca13666680051f91ab3f49ce2bd7e19ead7;hpb=8894a56619450d998be2cbbfd45c70f16d642936

diff --git a/codegen.scm b/codegen.scm
index 10e005f..dededbe 100644
--- a/codegen.scm
+++ b/codegen.scm
@@ -9,6 +9,40 @@
     (apply printf s)
     (display "\n")))
 
+(define wordsize 8)
+
+(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)))]))
+
+(define (on-stack? expr)
+  (case (ast-type expr)
+    ['stack (cadr expr)]
+    [else #f]))
+
+					; 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)
@@ -76,8 +110,6 @@
     ('linux  (emit "mov $1, %rax"))) ; syscall 1 (write)
   (emit "syscall"))
 
-(define wordsize 8)
-
 (define (codegen-let bindings body si env)
 
 					; is this a closure that captures itself?
@@ -86,24 +118,54 @@
     (and (eqv? (ast-type expr) 'closure)
 	 (memv name (caddr expr))))
 
-  (let* ((stack-offsets (map (lambda (name x) ; assoc map of binding name to offset
+
+  ;; (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))))
+  
+  (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)))
+			     (range 0 (length bindings)))]
+	 [inner-si (- si (* (length bindings) wordsize))]
 
-	 (get-offset (lambda (n) (cdr (assoc n stack-offsets))))
+	 [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))])
@@ -113,23 +175,26 @@
 					; 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))))
 		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))
-    (error #f (format "Variable ~a is not bound" name)))
-  (let ((offset (cdr (assoc name env))))
-    (emit "movq ~a(%rbp), %rax" offset)))
+  (let ([binding (assoc name (env-bindings env))])
+    (if (not binding)
+	(error #f (format "Variable ~a is not bound" name))
+	(emit "movq ~a(%rbp), %rax" (cdr binding)))))
 
 (define cur-lambda 0)
 (define (fresh-lambda)
@@ -163,7 +228,7 @@
 					; 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:
@@ -238,7 +303,8 @@
 			       (* (- 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: ")
@@ -296,6 +362,82 @@
     (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 (codegen-destructor tor)
+    (let* ([res (codegen-expr (cadr e) si env)]
+	   [info (cadr tor)]
+	   [index (caddr info)]
+	   [type (car info)]
+	   [sum (cadr info)])
+      (when (not (stack-expr? res))
+	(error #f "codegened something that wasn't a stack expression"))
+					;TODO handle stack types
+      (emit "movq ~a(%rbp), %rax"
+	    (- si (data-product-offset (env-data-layouts env) type sum index)))))				      
+
+  (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)]
+	   
+	   [insert-product
+	    (lambda (expr i)
+	      (let ([res (codegen-expr expr si env)]
+		    [stack-offset (- si (data-product-offset (env-data-layouts env)
+							     type sum
+							     i))])
+		(if (stack-expr? res)
+		    (error #f "todo: handle stack-exprs in stack exprs")
+		    (emit "movq %rax, ~a(%rbp)" stack-offset))))])
+
+					; emit the tag
+      (emit "movq $~a, ~a(%rbp)" tag si)
+      
+      (for-each insert-product args (range 0 (length args)))
+      (type-size (env-data-layouts env) type)))
+  
+  (let* ([tor (data-tor env e)]
+	 [constructor (eqv? 'constructor (caddr (cadr tor)))])
+    (if constructor
+	(codegen-constructor tor)
+	(codegen-destructor tor))))
+
+(define stack-expr? number?)
+
+					; returns a number if result was stored on stack 
 (define (codegen-expr e si env)
   (emit "# ~a" e)
   (case (ast-type e)
@@ -309,10 +451,14 @@
        ('= (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)
@@ -329,9 +475,24 @@
     ('static-string (emit "movq ~a@GOTPCREL(%rip), %rax"
 			  (cadr e)))
 
-    (else (error #f "don't know how to codegen this"))))
+    ('stack (codegen-expr (caddr e) si env))
 
+    (else (error #f "don't know how to codegen this"))))
 
+					; takes in a expr annotated with types and returns a type-less AST
+					; with stack values wrapped
+(define (annotate-stack-values data-layout ann-e)
+  (define (stack-type? type)
+    (assoc type data-layout))
+  (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 ,(ast-traverse strip e))
+	(ast-traverse (lambda (x)
+			(annotate-stack-values data-layout x))
+		      e))))
 
 (define (free-vars prog)
   (define bound '())
@@ -449,29 +610,6 @@
   (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
 					; -----------------------
@@ -518,13 +656,18 @@
 (define (codegen program)
   (set! cur-label 0)
   (set! cur-lambda 0)
-  (let* ([body (program-body program)]	
+  (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)]
 	 
-	 (extract-res-0 (extract-strings body))
-	 (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)))
+	 (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")
@@ -538,7 +681,7 @@
 
     (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")
@@ -580,6 +723,6 @@
 ; %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