Message-Id: <m0zMYAJ-000S4WC@inti.gov.ar>
Comments: Authenticated sender is <salvador AT natacha DOT inti DOT gov DOT ar>
From: "Salvador Eduardo Tropea (SET)" <salvador AT inti DOT gov DOT ar>
Organization: INTI
To: bitland AT aol DOT com (Bitland), djgpp AT delorie DOT com
Date: Fri, 25 Sep 1998 10:48:00 +0000
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Subject: Re: malloc/new once again
In-reply-to: <19980925035923.00799.00001778@ng85.aol.com>
Precedence: bulk

bitland AT aol DOT com (Bitland) wrote:

> I got version 2.8.1 and still have malloc-problems. 

You are confusing djgpp and gcc versions. GCC 2.8.1 doesn't imply a change in 
malloc, you should try the *ALPHA* djgpp 2.02 to see a change in malloc. But 
remmember: is alpha and it means "we are not sure if it is stable".

> Allocating memory in an
> absolutely simple program and freeing it in the next moment, displaying the
> memory with  __dpmi_get_free_memory_information(&dfmi); gives confusing
> memory 

Because you don't know how malloc works ;-))

> values and then the memory slowly gets low.

Looking at your code that's natural because:

1) free doesn't return the memory to the DPMI server.
2) You are using a random size so you are asking for different chunck sizes 
and hence the memory chuncks returned to the heap (with free) aren't 
re-used 100% of the time.
 
> Here's the main loop (simulating making allegro bitmaps), with MAXSIZE 1024,
> CDEPTH 32 and MAPS so, that you cannot allocate more than you have, but be sure
> to get closely to the limit, then the effect comes faster.
> 
>   do
>   {
>     for (i=0;i<MAPS;i++)
>     {
>       hm[i] = malloc((CDEPTH>>3)*(random()%MAXSIZE * (random()%MAXSIZE)));
>       if (!hm[i])     
>         textout(screen,font,"there it is !",200,20,COLWHITE);   /* via allegro
> */
>     }
>     checkmem("allocated");
> 
>     for (i=0;i<MAPS;i++)
>       if (hm[i])
>         free(hm[i]);
>     checkmem("freed    ");
> 
>   } while (!keypressed());
> 
> After this, you SOMETIMES cannot get a large memory block !

Because you heavilly fragmented the memory.

> And 'there it is' shouldn't appear at all, what if allocated with 'new' without
> a newhandler ? The program crashes...

Crashes or says virtual memory exhausted?
 
> Any idea without a do-it-yourself-memory-manager ?

A) Try 2.02 .
B) Avoid this silly allocation scheme.

SET
------------------------------------ 0 --------------------------------
Visit my home page: http://set-soft.home.ml.org/
or
http://www.geocities.com/SiliconValley/Vista/6552/
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Alternative e-mail: set-soft AT usa DOT net set AT computer DOT org
ICQ: 2951574
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA
TE: +(541) 759 0013