From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <andrieu@ijm.jussieu.fr>
Delivered-To: caml-list@yquem.inria.fr
Received: from nez-perce.inria.fr (nez-perce.inria.fr [192.93.2.78])
	by yquem.inria.fr (Postfix) with ESMTP id 4A24CBB81
	for <caml-list@yquem.inria.fr>; Sun,  2 Jan 2005 00:59:37 +0100 (CET)
Received: from kraid.nerim.net (smtp-106-saturday.nerim.net [62.4.16.106])
	by nez-perce.inria.fr (8.13.0/8.13.0) with ESMTP id j01NxbeT016431
	for <caml-list@yquem.inria.fr>; Sun, 2 Jan 2005 00:59:37 +0100
Received: from localhost (karryall.dnsalias.org [62.4.18.180])
	by kraid.nerim.net (Postfix) with ESMTP
	id C91C141F8F; Sun,  2 Jan 2005 00:59:34 +0100 (CET)
Date: Sun, 02 Jan 2005 00:59:35 +0100 (CET)
Message-Id: <20050102.005935.74752788.oandrieu@nerim.net>
To: jonathan.roewen@gmail.com
Cc: caml-list@yquem.inria.fr
Subject: Re: [Caml-list] Garbage Collecting
From: Olivier Andrieu <andrieu@ijm.jussieu.fr>
In-Reply-To: <ad8cfe7e05010115045ef0e999@mail.gmail.com>
References: <ad8cfe7e05010115045ef0e999@mail.gmail.com>
X-Mailer: Mew version 4.0.64 on Emacs 21.3 / Mule 5.0 (SAKAKI)
Mime-Version: 1.0
Content-Type: Text/Plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Miltered: at nez-perce with ID 41D73969.000 by Joe's j-chkmail (http://j-chkmail.ensmp.fr)!
X-Spam: no; 0.00; caml-list:01 garbage:01 andrieu:01 andrieu:01 ijm:01 ocaml:01 printf:01 fib:01 fib:01 printf:01 minor:01 computes:01 allocates:01 runtime:01 ocaml:01 
X-Spam-Checker-Version: SpamAssassin 3.0.0 (2004-09-13) on yquem.inria.fr
X-Spam-Status: No, score=0.0 required=5.0 tests=none autolearn=disabled 
	version=3.0.0
X-Spam-Level: 

 > Jonathan Roewen [Sun, 2 Jan 2005]:
 > 
 > Hi,
 > 
 > We're trying to figure out why memory doesn't get collected in our
 > OS by the OCaml GC, and seems like it doesn't want to reclaim any
 > memory.  We also suspect that Gc.allocated_bytes has a memory leak.
 > 
 > Here is our test case:
 > (* File mod.ml -- some ``useful'' Caml functions *)
 > open Printf
 > 
 > let fib (n:int) =1 + n
 > 
 > ;;
 > 
 > let mem = ref 0;;
 > let do_stuff () = 
 > (* 	making this loop more increases mem usage *)
 > 	for i=1 to 900000 do
 > 		fib i;
 > 		int_of_float (Gc.allocated_bytes()); (* A *)
 > 	done;
 > 	mem := int_of_float (Gc.allocated_bytes())
 > ;;
 > 
 > let do_results() = 
 > 	printf "allocated %d\n" !mem;
 > 	Gc.full_major(); (* B1 *)
 > 	Gc.compact();
 > 	Gc.major();
 > 	Gc.minor(); (*B2 *)
 > 	printf "allocated %d\n" (int_of_float (Gc.allocated_bytes()))
 > ;;
 > do_stuff ();;
 > do_results ();;
 > 
 > My results:
 > Test 1)
 > allocated 57600272
 > allocated 57600788
 > Test 2)
 > allocated 272
 > allocated 956
 > Test 3)
 > allocated 272
 > allocated 744
 > 
 > Where test 1 is using the above code; test 2 is with line A commented
 > out; and test 3 is with both line A and lines B1-B2 commented out.
 > 
 > As you can see from our tests, that invoking the Gc increases memory,
 > and Gc.allocated_bytes appears to gobble it up like candy.

Well, what do you expect ? Gc.allocated_bytes is a pretty
regular-looking function : it computes stuff, using a couple of
intermediate values, so it allocates a couple of bytes to store
these. Now, if you stick this a loop, number_of_iterations times
couple_of_bytes bytes get allocated. This has nothing to do with
reclaiming memory actually, Gc.allocated_bytes simply keeps track of
how many bytes were requested to the runtime; all those bytes are not
"live".

 > Would appreciate someone shedding some light on this topic, as
 > writing our OS in OCaml is going to have some major issues if it
 > won't reclaim memory.

I wouldn't worry. There is some anecdotal evidence that the OCaml GC
is not a fraud and that it _does_ reclaim memory.

-- 
   Olivier