Use sorted message array with binary

searches/inserts. This speeds up the mh/maildir operations considerably. Testing on a MH folder with 6000 messages shows the following data: Time of execution List Array+Bsearch % Speedup (times) scan 24.795 7.526 30.3 3.294 refile (to) 10.915 0.068 0.6 160.514 rmm 10.893 0.105 1 103.742

Use sorted message array with binary
searches/inserts. This speeds up the mh/maildir operations considerably. Testing on a MH folder with 6000 messages shows the following data: Time of execution List Array+Bsearch % Speedup (times) scan 24.795 7.526 30.3 3.294 refile (to) 10.915 0.068 0.6 160.514 rmm 10.893 0.105 1 103.742
Sergey Poznyakoff
Commit 84c33414 ... 84c334143ef6fbe6220e3d889c32375f63c8e8e2 authored 2004-01-22 10:54:35 +0000 by Sergey Poznyakoff
Showing 4 changed files with 233 additions and 113 deletions
mailbox/amd.c
mailbox/include/amd.h
mailbox/maildir/mbox.c
mailbox/mh/mbox.c
--- a/mailbox/amd.c
View file @84c3341
+++ b/mailbox/amd.c
View file @84c3341
@@ -111,8 +111,132 @@ static int amd_envelope_date __P((envelope_t envelope, char *buf, size_t len,
 static int amd_envelope_sender __P((envelope_t envelope, char *buf, size_t len,
 				   size_t *psize));
-/* Should be in other header file.  */
+/* Operations on message array */
-extern int amd_message_number __P ((message_t msg, size_t *pnum));
+/* Perform binary search for message MSG on a segment of message array
+   of AMD between the indexes FIRST and LAST inclusively.
+   If found, return 0 and store index of the located entry in the
+   variable PRET. Otherwise, return 1 and place into PRET index of
+   the nearest array element that is less than MSG (in the sense of
+   amd->msg_cmp)
+   Indexes are zero-based. */
+static int
+amd_msg_bsearch (struct _amd_data *amd, off_t first, off_t last,
+		 struct _amd_message *msg,
+		 off_t *pret)
+{
+  off_t mid;
+  int rc;
+  if (last < first)
+    return 1;
+  mid = (first + last) / 2;
+  rc = amd->msg_cmp (amd->msg_array[mid], msg);
+  if (rc > 0)
+    return amd_msg_bsearch (amd, first, mid-1, msg, pret);
+  *pret = mid;
+  if (rc < 0)
+    return amd_msg_bsearch (amd, mid+1, last, msg, pret);
+  /* else */
+  return 0;
+}
+/* Search for message MSG in the message array of AMD.
+   If found, return 0 and store index of the located entry in the
+   variable PRET. Otherwise, return 1 and place into PRET index of
+   the array element that is less than MSG (in the sense of
+   amd->msg_cmp)
+   Index returned in PRET is 1-based, so *PRET == 0 means that MSG
+   is less than the very first element of the message array.
+   In other words, when amd_msg_lookup() returns 1, the value in *PRET
+   can be regarded as a 0-based index of the array slot where MSG can
+   be inserted */
+int
+amd_msg_lookup (struct _amd_data *amd, struct _amd_message *msg,
+		 size_t *pret)
+{
+  int rc;
+  off_t i;
+  if (!amd->msg_array)
+    {
+      *pret = 0;
+      return 1;
+    }
+  rc = amd->msg_cmp (msg, amd->msg_array[0]);
+  if (rc < 0)
+    {
+      *pret = 0;
+      return 1;
+    }
+  else if (rc == 0)
+    {
+      *pret = 1;
+      return 0;
+    }
+  rc = amd->msg_cmp (msg, amd->msg_array[amd->msg_count - 1]);
+  if (rc > 0)
+    {
+      *pret = amd->msg_count;
+      return 1;
+    }
+  else if (rc == 0)
+    {
+      *pret = amd->msg_count;
+      return 0;
+    }
+  rc = amd_msg_bsearch (amd, 0, amd->msg_count - 1, msg, &i);
+  *pret = i + 1;
+  return rc;
+}
+#define AMD_MSG_INC 64
+/* Prepare the message array for insertion of a new message
+   at position INDEX (zero based), by moving its contents
+   one slot to the right. If necessary, expand the array by
+   AMD_MSG_INC */
+int
+amd_array_expand (struct _amd_data *amd, size_t index)
+{
+  if (amd->msg_count == amd->msg_max)
+    {
+      struct _amd_message **p;
+      amd->msg_max += AMD_MSG_INC; /* FIXME: configurable? */
+      p = realloc (amd->msg_array, amd->msg_max * amd->msg_size);
+      if (!p)
+	{
+	  amd->msg_max -= AMD_MSG_INC;
+	  return ENOMEM;
+	}
+      amd->msg_array = p;
+    }
+  memmove (&amd->msg_array[index+1], &amd->msg_array[index],
+	   (amd->msg_count-index) * amd->msg_size);
+  amd->msg_count++;
+  return 0;
+}
+/* Shrink the message array by removing element at INDEX-1 and
+   shifting left by one position all the elements on the right of
+   it. */
+int
+amd_array_shrink (struct _amd_data *amd, size_t index)
+{
+  memmove (&amd->msg_array[index-1], &amd->msg_array[index],
+	   (amd->msg_count-index) * amd->msg_size);
+  amd->msg_count--;
+  return 0;
+}
 int
 amd_init_mailbox (mailbox_t mailbox, size_t amd_size, struct _amd_data **pamd) 
@@ -171,20 +295,18 @@ static void
 amd_destroy (mailbox_t mailbox)
 {
  struct _amd_data *amd = mailbox->data;
-  struct _amd_message *msg, *next;
+  size_t i;
  if (!amd)
    return;
  monitor_wrlock (mailbox->monitor);
-  msg = amd->msg_head;
+  for (i = 0; i < amd->msg_count; i++)
-  while (msg)
    {
-      next = msg->next;
+      message_destroy (&amd->msg_array[i]->message, amd->msg_array[i]);
-      message_destroy (&msg->message, msg);
+      free (amd->msg_array[i]);
-      free (msg);
-      msg = next;
    }
+  free (amd->msg_array);
  if (amd->name)
    free (amd->name);
@@ -220,16 +342,13 @@ amd_close (mailbox_t mailbox)
  return 0;
 }
-static struct _amd_message *
+struct _amd_message *
 _amd_get_message (struct _amd_data *amd, size_t msgno)
 {
-  size_t n;
+  msgno--;
-  struct _amd_message *msg;
+  if (msgno >= amd->msg_count)
+    return NULL;
-  for (n = 1, msg = amd->msg_head; msg && n < msgno; n++, msg = msg->next)
+  return amd->msg_array[msgno];
-    ;
-  return msg;
 }
 static int
@@ -440,7 +559,7 @@ _amd_message_save (struct _amd_data *amd, struct _amd_message *mhm, int expunge)
  /* Add imapbase */
  if (amd->next_uid
-      && (!amd->msg_head || (amd->msg_head == mhm))) /*FIXME*/
+      && (!amd->msg_array || (amd->msg_array[0] == mhm))) /*FIXME*/
    {
      nbytes += fprintf (fp, "X-IMAPbase: %lu %u\n",
 			 (unsigned long) amd->uidvalidity,
@@ -556,7 +675,12 @@ amd_append_message (mailbox_t mailbox, message_t msg)
  status = _amd_message_save (amd, mhm, 0);
  mhm->message = NULL;
  /* Insert and re-scan the message */
-  _amd_message_insert (amd, mhm);
+  status = _amd_message_insert (amd, mhm);
+  if (status)
+    {
+      free (mhm);
+      return status;
+    }
  if (amd->msg_finish_delivery)
    amd->msg_finish_delivery (amd, mhm);
@@ -588,8 +712,7 @@ static int
 amd_messages_recent (mailbox_t mailbox, size_t *pcount)
 {
  struct _amd_data *amd = mailbox->data;
-  struct _amd_message *mhm;
+  size_t count, i;
-  size_t count;
  /* If we did not start a scanning yet do it now.  */
  if (amd->msg_count == 0)
@@ -599,9 +722,9 @@ amd_messages_recent (mailbox_t mailbox, size_t *pcount)
 	return status;
    }
  count = 0;
-  for (mhm = amd->msg_head; mhm; mhm = mhm->next)
+  for (i = 0; i < amd->msg_count; i++)
    {
-      if (MU_ATTRIBUTE_IS_UNSEEN(mhm->attr_flags))
+      if (MU_ATTRIBUTE_IS_UNSEEN(amd->msg_array[i]->attr_flags))
 	count++;
    }
  *pcount = count;
@@ -613,8 +736,7 @@ static int
 amd_message_unseen (mailbox_t mailbox, size_t *pmsgno)
 {
  struct _amd_data *amd = mailbox->data;
-  struct _amd_message *mhm;
+  size_t i;
-  size_t i, unseen;
  /* If we did not start a scanning yet do it now.  */
  if (amd->msg_count == 0)
@@ -623,15 +745,15 @@ amd_message_unseen (mailbox_t mailbox, size_t *pmsgno)
      if (status != 0)
 	return status;
    }
-  for (unseen = i = 1, mhm = amd->msg_head; mhm; i++, mhm = mhm->next)
+  for (i = 0; i < amd->msg_count; i++)
    {
-      if (MU_ATTRIBUTE_IS_UNREAD(mhm->attr_flags))
+      if (MU_ATTRIBUTE_IS_UNREAD(amd->msg_array[0]->attr_flags))
 	{
-	  unseen = i;
+	  *pmsgno = i + 1;
 	  break;
 	}
    }
-  *pmsgno = unseen;
  return 0;
 }
@@ -640,6 +762,7 @@ amd_expunge (mailbox_t mailbox)
 {
  struct _amd_data *amd = mailbox->data;
  struct _amd_message *mhm;
+  size_t i;
  if (amd == NULL)
    return EINVAL;
@@ -648,20 +771,18 @@ amd_expunge (mailbox_t mailbox)
    return 0;
  /* Find the first dirty(modified) message.  */
-  for (mhm = amd->msg_head; mhm; mhm = mhm->next)
+  for (i = 0; i < amd->msg_count; i++)
    {
+      mhm = amd->msg_array[i];
      if ((mhm->attr_flags & MU_ATTRIBUTE_MODIFIED) ||
 	  (mhm->attr_flags & MU_ATTRIBUTE_DELETED) ||
 	  (mhm->message && message_is_modified (mhm->message)))
 	break;
    }
-  if (!mhm)
+  while (i < amd->msg_count)
-    return 0; /* Nothing changed, just return.  */
-  while (mhm)
    {
-      struct _amd_message *next = mhm->next;
+      mhm = amd->msg_array[i];
      if (mhm->attr_flags & MU_ATTRIBUTE_DELETED)
 	{
@@ -676,15 +797,19 @@ amd_expunge (mailbox_t mailbox)
 	      free (new_name);
 	    }
 	  _amd_message_delete (amd, mhm);
+	  /* Do not increase i! */
 	}
-      else if ((mhm->attr_flags & MU_ATTRIBUTE_MODIFIED)
+      else
+	{
+	  if ((mhm->attr_flags & MU_ATTRIBUTE_MODIFIED)
 	      || (mhm->message && message_is_modified (mhm->message)))
 	    {
 	      _amd_attach_message (mailbox, mhm, NULL);
 	      mhm->deleted = mhm->attr_flags & MU_ATTRIBUTE_DELETED;
 	      _amd_message_save (amd, mhm, 1);
 	    }
-      mhm = next;
+	  i++; /* Move to the next message */
+	}
    }
  return 0;
@@ -695,6 +820,7 @@ amd_save_attributes (mailbox_t mailbox)
 {
  struct _amd_data *amd = mailbox->data;
  struct _amd_message *mhm;
+  size_t i;
  if (amd == NULL)
    return EINVAL;
@@ -703,19 +829,17 @@ amd_save_attributes (mailbox_t mailbox)
    return 0;
  /* Find the first dirty(modified) message.  */
-  for (mhm = amd->msg_head; mhm; mhm = mhm->next)
+  for (i = 0; i < amd->msg_count; i++)
    {
+      mhm = amd->msg_array[i];
      if ((mhm->attr_flags & MU_ATTRIBUTE_MODIFIED)
 	  || (mhm->message && message_is_modified (mhm->message)))
 	break;
    }
-  if (!mhm)
+  for ( ; i < amd->msg_count; i++)
-    return 0; /* Nothing changed, just return.  */
-  while (mhm)
    {
-      struct _amd_message *next = mhm->next;
+      mhm = amd->msg_array[i];
      if ((mhm->attr_flags & MU_ATTRIBUTE_MODIFIED)
 	  || (mhm->message && message_is_modified (mhm->message)))
@@ -724,7 +848,6 @@ amd_save_attributes (mailbox_t mailbox)
 	  mhm->deleted = mhm->attr_flags & MU_ATTRIBUTE_DELETED;
 	  _amd_message_save (amd, mhm, 0);
 	}
-      mhm = next;
    }
  return 0;
@@ -782,61 +905,52 @@ amd_cleanup (void *arg)
 }
 /* Insert message msg into the message list on the appropriate position */
-void
+int
 _amd_message_insert (struct _amd_data *amd, struct _amd_message *msg)
 {
-  struct _amd_message *p;
+  size_t index;
-  struct _amd_message *prev;
-  for (p = amd->msg_head; p && amd->msg_cmp (p, msg) < 0; p = p->next)
-    ;
-  if (!p)
+  if (amd_msg_lookup (amd, msg, &index))
    {
-      msg->next = NULL;
+      /* Not found. Index is the index of the array cell where msg
-      msg->prev = amd->msg_tail;
+	 must be placed */
-      amd->msg_tail = msg;
+      int rc = amd_array_expand (amd, index);
-      if (!amd->msg_head)
+      if (rc)
-	amd->msg_head = msg;
+	return rc;
+      amd->msg_array[index] = msg;
+      msg->amd = amd;
    }
  else
    {
-      msg->next = p;
+      /*FIXME: Found? Shouldn't happen */
-      msg->prev = p->prev;
+      return EEXIST;
-      p->prev = msg;
    }
-  if ((prev = msg->prev) != NULL)
+  return 0;
-    prev->next = msg;
-  else
-    amd->msg_head = msg;
-  msg->amd = amd;
-  amd->msg_count++;
 }
 static void
 _amd_message_delete (struct _amd_data *amd, struct _amd_message *msg)
 {
-  struct _amd_message *p;
+  size_t index;
-  struct _amd_message **pp = amd_pool_lookup (msg);
+  struct _amd_message **pp;
-  if (pp)
+  if (amd_msg_lookup (amd, msg, &index))
-    *pp = NULL;
+    {
+      /* FIXME: Not found? */
+      return;
+    }
-  if ((p = msg->next) != NULL)
+  msg = _amd_get_message (amd, index);
-    p->prev = msg->prev;
-  else
-    amd->msg_tail = msg->prev;
-  if ((p = msg->prev) != NULL)
+  pp = amd_pool_lookup (msg);
-    p->next = msg->next;
+  if (pp)
-  else
+    *pp = NULL;
-    amd->msg_head = msg->next;
  message_destroy (&msg->message, msg);
  if (amd->msg_free)
    amd->msg_free (msg);
  free (msg);
-  amd->msg_count--;
+  amd_array_shrink (amd, index);
 }
 /* Scan given message and fill amd_message_t fields.
@@ -935,7 +1049,7 @@ amd_is_updated (mailbox_t mailbox)
  struct stat st;
  struct _amd_data *amd = mailbox->data;
-  if (!amd->msg_head)
+  if (amd->msg_count == 0)
    return 0;
  if (stat (amd->name, &st) < 0)
--- a/mailbox/include/amd.h
View file @84c3341
+++ b/mailbox/include/amd.h
View file @84c3341
@@ -39,9 +39,6 @@ do \
 struct _amd_data;
 struct _amd_message
 {
-  struct _amd_message *next;
-  struct _amd_message *prev;
  stream_t stream;          /* Associated file stream */
  off_t body_start;         /* Offset of body start in the message file */
  off_t body_end;           /* Offset of body end (size of file, effectively)*/
@@ -72,8 +69,8 @@ struct _amd_data
  /* List of messages: */
  size_t msg_count; /* number of messages in the list */
-  struct _amd_message *msg_head;  /* First */
+  size_t msg_max;   /* maximum message buffer capacity */
-  struct _amd_message *msg_tail;  /* Last */
+  struct _amd_message **msg_array;
  unsigned long uidvalidity;
@@ -92,9 +89,13 @@ struct _amd_data
 int amd_init_mailbox __P((mailbox_t mailbox, size_t mhd_size,
 			  struct _amd_data **pmhd));
-void _amd_message_insert __P((struct _amd_data *mhd,
+int _amd_message_insert __P((struct _amd_data *mhd,
 			     struct _amd_message *msg));
 int amd_message_stream_open __P((struct _amd_message *mhm));
 void amd_message_stream_close __P((struct _amd_message *mhm));
 void amd_cleanup (void *arg);
 int amd_url_init (url_t url, const char *scheme);
+struct _amd_message *_amd_get_message __P((struct _amd_data *amd,
+					   size_t msgno));
+int amd_msg_lookup __P((struct _amd_data *amd, struct _amd_message *msg,
+			size_t *pret));
--- a/mailbox/maildir/mbox.c
View file @84c3341
+++ b/mailbox/maildir/mbox.c
View file @84c3341
@@ -324,7 +324,8 @@ maildir_message_uid (message_t msg, size_t *puid)
 static size_t
 maildir_next_uid (struct _amd_data *amd)
 {
-  struct _maildir_message *msg = (struct _maildir_message *) amd->msg_tail;
+  struct _maildir_message *msg = (struct _maildir_message *)
+                                   _amd_get_message (amd, amd->msg_count);
  return (msg ? msg->uid : 0) + 1;
 }
@@ -489,13 +490,17 @@ maildir_message_lookup (struct _amd_data *amd, char *file_name)
  char *p = strchr (file_name, ':');
  size_t length = p ? p - file_name : strlen (file_name);
  int rc;
+  size_t i;
-  for (msg = (struct _maildir_message *) amd->msg_head;
+  /*FIXME!*/
-       msg
+  for (i = 0; i < amd->msg_count; i++)
-	 && strlen (msg->file_name) >= length
+    {
-	 && (rc = memcmp (msg->file_name, file_name, length)) < 0;
+      msg = (struct _maildir_message *) amd->msg_array[i];
-       msg = (struct _maildir_message *) msg->amd_message.next)
-    ;
+      if (strlen (msg->file_name) <= length
+	  && (rc = memcmp (msg->file_name, file_name, length)) == 0)
+	break;
+    }
  return rc == 0 ? msg : NULL;
 }
@@ -588,8 +593,8 @@ maildir_scan0 (mailbox_t mailbox, size_t msgno ARG_UNUSED, size_t *pcount,
  if (do_notify)
    {
-      struct _amd_message *mp;
+      size_t i;
-      for (mp = amd->msg_head; mp; mp = mp->next)
+      for (i = 0; i < amd->msg_count++; i++)
 	{
 	  DISPATCH_ADD_MSG(mailbox, amd);
 	}
@@ -609,11 +614,11 @@ maildir_scan0 (mailbox_t mailbox, size_t msgno ARG_UNUSED, size_t *pcount,
 	  amd->uidvalidity = (unsigned long) time (NULL);
 	  /* FIXME amd->uidnext = amd->msg_count + 1;*/
 	  /* Tell that we have been modified for expunging.  */
-	  if (amd->msg_head)
+	  if (amd->msg_count)
 	    {
-	      amd_message_stream_open (amd->msg_head);
+	      amd_message_stream_open (amd->msg_array[0]);
-	      amd_message_stream_close (amd->msg_head);
+	      amd_message_stream_close (amd->msg_array[0]);
-	      amd->msg_head->attr_flags |= MU_ATTRIBUTE_MODIFIED;
+	      amd->msg_array[0]->attr_flags |= MU_ATTRIBUTE_MODIFIED;
 	    }
 	}
    }
--- a/mailbox/mh/mbox.c
View file @84c3341
+++ b/mailbox/mh/mbox.c
View file @84c3341
@@ -85,7 +85,8 @@ mh_message_cmp (struct _amd_message *a, struct _amd_message *b)
 static size_t
 _mh_next_seq (struct _amd_data *amd)
 {
-  struct _mh_message *msg = (struct _mh_message *) amd->msg_tail;
+  struct _mh_message *msg = (struct _mh_message *)
+                              _amd_get_message (amd, amd->msg_count);
  return (msg ? msg->seq_number : 0) + 1;
 }
@@ -111,16 +112,14 @@ _mh_message_name (struct _amd_message *amsg, int deleted)
 static struct _mh_message *
 _mh_get_message_seq (struct _amd_data *amd, size_t seq)
 {
-  struct _mh_message *msg;
+  struct _mh_message msg;
+  size_t index;
-  for (msg = (struct _mh_message *) amd->msg_head;
-       msg && msg->seq_number < seq;
-       msg = (struct _mh_message *) msg->amd_message.next)
-    ;
-  if (msg)
+  msg.seq_number = seq;
-    return msg->seq_number == seq ? msg : NULL;
+  if (amd_msg_lookup (amd, (struct _amd_message*) &msg, &index))
    return NULL;
+  return (struct _mh_message *) _amd_get_message (amd, index);
 }
 /* Scan the mailbox */
@@ -205,8 +204,9 @@ mh_scan0 (mailbox_t mailbox, size_t msgno ARG_UNUSED, size_t *pcount,
  if (do_notify)
    {
-      struct _amd_message *mp;
+      size_t i;
-      for (mp = amd->msg_head; mp; mp = mp->next)
+      for (i = 0; i < amd->msg_count; i++)
 	{
 	  DISPATCH_ADD_MSG(mailbox, amd);
 	}
@@ -225,11 +225,11 @@ mh_scan0 (mailbox_t mailbox, size_t msgno ARG_UNUSED, size_t *pcount,
 	{
 	  amd->uidvalidity = (unsigned long)time (NULL);
 	  /* Tell that we have been modified for expunging.  */
-	  if (amd->msg_head)
+	  if (amd->msg_count)
 	    {
-	      amd_message_stream_open (amd->msg_head);
+	      amd_message_stream_open (amd->msg_array[0]);
-	      amd_message_stream_close (amd->msg_head);
+	      amd_message_stream_close (amd->msg_array[0]);
-	      amd->msg_head->attr_flags |= MU_ATTRIBUTE_MODIFIED;
+	      amd->msg_array[0]->attr_flags |= MU_ATTRIBUTE_MODIFIED;
 	    }
 	}
    }