rosella/gc.c

#include <assert.h>
#include <inttypes.h>
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>

#include "gc.h"

gc_stats_t gc_stats;

/* Helper macros to reduce duplication */
#define VECTOR_BYTES(nelem) (sizeof(vector_t) + (sizeof(value_t) * (nelem)))
#define BYTESTR_BYTES(size) (sizeof(byte_string_t) + (size))

/* Alignment must ensure each object has enough room to hold a pair (BH . new_addr) */
#define GC_ALIGNMENT  ((size_t)(sizeof(pair_t)))

object_t *get_object(value_t v)
{
  if (is_object(v))
    return _get_object(v);
  else
    abort();
}

value_t cons(value_t car, value_t cdr)
{
  gc_root_t car_root, cdr_root;
  pair_t *p;

  register_gc_root(&car_root, car);
  register_gc_root(&cdr_root, cdr);

  p = gc_alloc(sizeof(pair_t));
  p->car = car_root.value;
  p->cdr = cdr_root.value;

  unregister_gc_root(&car_root);
  unregister_gc_root(&cdr_root);

  return pair_value(p);
}

pair_t *get_pair(value_t v)
{
  if (is_pair(v))
    return _get_pair(v);
  else
    abort();
}

value_t make_box(value_t initial_value)
{
  gc_root_t iv_root;
  box_t *box;

  register_gc_root(&iv_root, initial_value);

  box = (box_t*)gc_alloc(sizeof(box_t));
  box->tag = TYPE_TAG_BOX;
  box->value = iv_root.value;

  unregister_gc_root(&iv_root);

  return object_value(box);
}

box_t *get_box(value_t v)
{
  if (is_box(v))
    return _get_box(v);
  else
    abort();
}

value_t make_vector(size_t nelem, value_t initial_value)
{
  gc_root_t iv_root;
  vector_t *vec;

  register_gc_root(&iv_root, initial_value);

  vec = (vector_t*)gc_alloc(VECTOR_BYTES(nelem));
  vec->tag = TYPE_TAG_VECTOR;
  vec->size = nelem;

  for (int i = 0; i < nelem; ++i)
    vec->elements[i] = iv_root.value;

  unregister_gc_root(&iv_root);

  return object_value(vec);
}

vector_t *get_vector(value_t v)
{
  if (is_vector(v))
    return _get_vector(v);
  else
    abort();
}

value_t make_byte_string(size_t size, int default_value)
{
  const size_t nbytes = BYTESTR_BYTES(size);
  byte_string_t *str;
  
  str = (byte_string_t*)gc_alloc(nbytes);
  str->tag = TYPE_TAG_BYTESTR;
  str->size = size;

  memset(str->bytes, default_value, size);

  return object_value(str);
}

byte_string_t *get_byte_string(value_t v)
{
  if (is_byte_string(v))
    return _get_byte_string(v);
  else
    abort();
}

intptr_t get_fixnum(value_t v)
{
  if (is_fixnum(v))
    return _get_fixnum(v);
  else
    abort();
}

/****************************************************************************/

static char      *gc_ranges[2];
static size_t     gc_min_size;
static size_t     gc_max_size;
static size_t     gc_soft_limit;

static int        gc_current_range;
static size_t     gc_free_space;
static char      *gc_free_ptr;

static gc_root_t  gc_root_list = {
  .value = NIL,
  .prev  = &gc_root_list,
  .next  = &gc_root_list
};

static inline size_t gc_align(size_t nbytes) __attribute__ ((const));
static int gc_range_of(void *object) __attribute__ ((const));
static void transfer_object(value_t *value);
static size_t transfer_children(object_t *object);

static inline size_t gc_align(size_t nbytes)
{
  return ((nbytes + GC_ALIGNMENT - 1) & ~(GC_ALIGNMENT - 1));
}

static int gc_range_of(void *object)
{
  if (((uintptr_t)object >= (uintptr_t)gc_ranges[0]) &&
      ((uintptr_t)object <  (uintptr_t)gc_ranges[1]))
    return 0;
  if (((uintptr_t)object >= (uintptr_t)gc_ranges[1]) &&
      ((uintptr_t)object <  (uintptr_t)gc_ranges[2]))
    return 1;
  return -1;
}

void gc_init(size_t min_size, size_t max_size)
{
  assert(min_size <= max_size);

  gc_ranges[0] = (char*)malloc(max_size);
  gc_ranges[1] = (char*)malloc(max_size);

  assert(gc_ranges[0] && gc_ranges[1]);

  gc_current_range = 0;
  gc_free_ptr = gc_ranges[gc_current_range];

  gc_min_size   = min_size;
  gc_max_size   = max_size;
  gc_soft_limit = gc_min_size;
  gc_free_space = gc_soft_limit;

  gc_stats.collections = 0;
  gc_stats.total_ticks = 0;
  gc_stats.high_water  = 0;
}

void register_gc_root(gc_root_t *root, value_t v)
{
  root->value = v;
  root->prev  = &gc_root_list;
  root->next  = gc_root_list.next;
  root->next->prev  = root;
  gc_root_list.next = root;
}

void unregister_gc_root(gc_root_t *root)
{
  assert(root && root->prev && root->next);             /* Uninitialized */
  assert((root->prev != root) && (root->next != root)); /* Already removed */

  /* Cut the given root out of the list */
  root->prev->next = root->next;
  root->next->prev = root->prev;

  /* Remove dead references to root list; protects against double-removal */
  root->prev = root->next = root;
}

void *gc_alloc(size_t nbytes)
{
  nbytes = gc_align(nbytes);

  if (nbytes > gc_free_space)
    collect_garbage(nbytes);

  void *p = gc_free_ptr;
  gc_free_ptr   += nbytes;
  gc_free_space -= nbytes;
  //debug(("Found %d bytes at %#.8p.\n", nbytes, p));
  return p;
}

static void transfer_object(value_t *value)
{
  if (is_object(*value))
  {
    object_t *obj = get_object(*value);
    value_t new_value;

    assert(gc_range_of(obj) != gc_current_range);

    if (obj->tag == BROKEN_HEART)
    {
      /* Object has already been moved; just update the reference */
      new_value = obj->payload.values[0];
    }
    else
    {
      size_t nbytes;

      switch (obj->tag)
      {
      case TYPE_TAG_VECTOR:
        nbytes = VECTOR_BYTES(((const vector_t*)obj)->size);
        break;
      case TYPE_TAG_BYTESTR:
        nbytes = BYTESTR_BYTES(((const byte_string_t*)obj)->size);
        break;
      case TYPE_TAG_BOX:
      default: /* pair */
        nbytes = sizeof(pair_t);
        break;
      }

      {
        void *newobj = gc_alloc(nbytes);
        memcpy(newobj, obj, nbytes);

        /* Keep the original tag bits (pair or object) */
        new_value = object_value(newobj) | (*value & 2);
      }

      obj->tag = BROKEN_HEART;
      obj->payload.values[0] = new_value;
    }

    *value = new_value;
  }
}

static inline size_t transfer_children(object_t *obj)
{
  switch (obj->tag)
  {
  case TYPE_TAG_VECTOR:
    {
      vector_t *vec = (vector_t*)obj;

      for (size_t i = 0; i < vec->size; ++i)
      {
        transfer_object(&vec->elements[i]);
      }

      return VECTOR_BYTES(vec->size);
    }
  case TYPE_TAG_BYTESTR:
    {
      return BYTESTR_BYTES(((const byte_string_t*)obj)->size);
    }
  case TYPE_TAG_BOX:
  default: /* pair */
    {
      pair_t *p = (pair_t*)obj;
      transfer_object(&p->car);
      transfer_object(&p->cdr);
      return sizeof(pair_t);
    }
  }
}

void collect_garbage(size_t min_free)
{
  static bool collecting = false;

  gc_root_t *root;
  char *object_ptr;

  //debug(("Collecting garbage...\n"));

  ++gc_stats.collections;
  gc_stats.total_ticks -= clock();

  /* Recursive calls to collector should never occur */
  if (collecting)
  {
    debug(("Ran out of memory while collecting garbage!\n"));
    abort();
  }
  else
    collecting = true;

  /* Swap ranges; new "current" range is initially empty, old one is full */
  gc_current_range = 1 - gc_current_range;
  gc_free_ptr      = (char*)&gc_ranges[gc_current_range][0];
  gc_free_space    = gc_soft_limit;
  object_ptr       = gc_free_ptr;

  /* Transfer GC roots (if necessary) */
  root = gc_root_list.next;
  
  while (root != &gc_root_list)
  {
    transfer_object(&root->value);
    root = root->next;
  }

  /* Keep transferring until no more objects in the new range refer to the old one */
  while (object_ptr < gc_free_ptr)
  {
    object_ptr += gc_align(transfer_children((object_t*)object_ptr));
  }
  
  //debug(("Finished collection with %d bytes to spare (out of %d bytes).\n", gc_free_space, gc_soft_limit));

  {
    size_t bytes_used = gc_soft_limit - gc_free_space;
    size_t min_limit = bytes_used + min_free;
    size_t new_limit = (5 * min_limit) / 3;

    if (new_limit > gc_max_size)
      new_limit = gc_max_size;
    else if (new_limit < gc_min_size)
      new_limit = gc_min_size;

    gc_free_space = (gc_free_space + new_limit) - gc_soft_limit;
    gc_soft_limit = new_limit;
  }

  if (gc_free_space < min_free)
  {
    out_of_memory();
  }

  if (gc_soft_limit > gc_stats.high_water)
  {
    gc_stats.high_water = gc_soft_limit;
  }

  /* Done collecting. */
  collecting = false;

  gc_stats.total_ticks += clock();
}

/* vim:set sw=2 expandtab: */