From: Luke Lau <luke_lau@icloud.com>
Date: Wed, 24 Jul 2019 23:59:30 +0000 (+0100)
Subject: Merge branch 'master' of lukelau.me:/srv/git/scheme
X-Git-Url: https://git.lukelau.me/?p=scheme.git;a=commitdiff_plain;h=5d4aafc1235538212989893f15006acc5d7d8f03

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

5d4aafc1235538212989893f15006acc5d7d8f03
diff --cc .gitignore
index e4e5f6c,e4e5f6c..02cfc43
--- a/.gitignore
+++ b/.gitignore
@@@ -1,1 -1,1 +1,2 @@@
  *~
++TAGS
diff --cc codegen.scm
index 04a03fa,571f1ef..78ebbad
--- a/codegen.scm
+++ b/codegen.scm
@@@ -64,17 -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 %rax, %rsi") ; string addr
++     (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 %rax, %rsi")  ; string addr
++     (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)
@@@ -113,8 -105,25 +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))
@@@ -179,11 -187,6 +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 "movq $~a, %rax" label))))
++    ('linux  (emit "lea $~a, %rax" label))))
 +
  (define cur-label 0)
  (define (fresh-label)
    (set! cur-label (+ 1 cur-label))
@@@ -361,27 -379,30 +396,40 @@@
  
  					; 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 "\t.data")
+     (emit ".data")
  
-     (for-each codegen-string-data strings)))
+     (emit "heap_start:")
+     (emit "\t.quad 0")
+ 
+     (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:
+ ; %rax = syscall #
+ ; %rdi = 1st arg
+ ; %rsi = 2nd arg
+ ; %rdx = 3rd arg
+ ; %r10 = 4th arg
+ ; %r8  = 5th arg
+ ; %r9  = 6th arg