From mint-bounce@lists.fishpool.fi  Mon Sep  5 00:13:04 2005
X-Original-To: fnaumann@mail.boerde.de
Delivered-To: fnaumann@mail.boerde.de
Subject: [MiNT] Mint Bug?
From: Mark Duckworth <mduckworth@atari-source.com>
To: Mint List <mint@fishpool.com>
Content-Type: text/plain
Date: Sun, 04 Sep 2005 18:09:03 -0400
Message-Id: <1125871743.23405.23.camel@mduckworth.phillypark.net>
Mime-Version: 1.0
X-Mailer: Evolution 2.2.3 (2.2.3-2.1.fc4.nr) 
Content-Transfer-Encoding: 7bit
X-ecartis-version: Ecartis v1.0.0
Sender: mint-bounce@lists.fishpool.fi
Errors-To: mint-bounce@lists.fishpool.fi
X-original-sender: mduckworth@atari-source.com
Precedence: bulk
List-help: <mailto:ecartis@lists.fishpool.fi?Subject=help>
List-unsubscribe: <mailto:mint-request@lists.fishpool.fi?Subject=unsubscribe>
List-Id: <mint.lists.fishpool.fi>
X-List-ID: <mint.lists.fishpool.fi>
X-Virus-Scanned: by amavisd-new at relay.boerde.de
X-Spam-Checker-Version: SpamAssassin 2.63 (2004-01-11) on relay.boerde.de
X-Spam-Status: No, hits=-1.0 tagged_above=-50.5 required=7.0 tests=AWL,
 BAYES_00
X-Spam-Level: 

Hey guys,

This one is VERY INTERESTING!  I'm confused with this one.  Binutils
2.16.1 gets an error linking part of gcc.  This error is not present
with binutils 2.13.  Anyway, I traced it down to a failure of malloc()
itself sometime around after binutils allocates 10megabytes.

So I ran a test.  It allocates 1 megabyte every 1 second, tells you if
the pointer is null or not, and then runs "free".  If the pointer is
null it means malloc failed, incidentally I ran free just to ensure it
wasn't magically a null pointer and the memory was actually allocated.

Here is the output:

root@washi:/root/binutils/binutils-mint/ld>./malloctest
Malloc 1, allocating 1 megabyte
261472256 bytes (255344K) free.
Malloc 2, allocating 1 megabyte
260440064 bytes (254336K) free.
Malloc 3, allocating 1 megabyte
259407872 bytes (253328K) free.
Malloc 4, allocating 1 megabyte
258375680 bytes (252320K) free.
Malloc 5, allocating 1 megabyte
257343488 bytes (251312K) free.
Malloc 6, allocating 1 megabyte
256311296 bytes (250304K) free.
Malloc 7, allocating 1 megabyte
255279104 bytes (249296K) free.
Malloc 8, allocating 1 megabyte
254246912 bytes (248288K) free.
Malloc 9, allocating 1 megabyte
253214720 bytes (247280K) free.
Malloc 10, allocating 1 megabyte
252182528 bytes (246272K) free.
Malloc 11, allocating 1 megabyte
251150336 bytes (245264K) free.
Malloc 12, allocating 1 megabyte
250118144 bytes (244256K) free.
Malloc 13, allocating 1 megabyte
ERROR!: pointer is NULL on count: 13
250118144 bytes (244256K) free.
Malloc 14, allocating 1 megabyte
ERROR!: pointer is NULL on count: 14
250118144 bytes (244256K) free.

You can see at count 13 malloc fails.. it will continue to fail.  To see
if this is a per-process limit, I logged in again via ssh and ran the
test again while the other one was frozen with it's allocated ram.
Every pointer returned was null.

I don't know much about how malloc works, but this test succeeds on all
of my linux boxes to an extremely high count.

For giggles, I tried on aranym too - It SUCCEEDS.  Right up until aranym
runs out of ram completely.  I have to try on a TT as well, but I don't
have that setup right now.

So this is either a CT60 hardware problem since the same failure occurs
on a completely different CT60 falcon, a MiNT/Falcon/CT60 code bug
(something that aranym wouldn't be using), or I'm just nuts.

I spent an awful lot of time bughunting in binutils for incorrect sizes
when it turns out malloc wasn't even succeeding!

Thanks,
Mark