Re: GCC 4.8 out of memory

Frank Wille <[email protected]>
Newsgroups gmane.os.netbsd.ports.m68k
Organization Privates Internet Ostwestfalen/Lippe e.V.
Message-ID <[email protected]>
Michael van Elst wrote:

> On Tue, Jun 02, 2015 at 04:16:48PM +0200, Frank Wille wrote:
> > On Mon, 1 Jun 2015 20:28:13 +0200
> > Michael van Elst <mlelstv%serpens.de@localhost> wrote:
> > > It is just a limit. The only constraint is that the text+data+stack
> > > address space must be handled by pmap which is limited to 512MB.
> > 
> > Wasn't that 224MB for 040/060?
>
> that was apparently fixed:
>
> revision 1.41
> date: 2010-12-28 19:31:05 +0100;  author: mhitch;  state: Exp;  lines: +2
-3;
> branches:  1.41.2;  1.41.4;
> Raise address of user stack to the highest safe (for 68040 mmu) address,
> which increases available user address space by 256MB.  One file in
> pkgsrc/lang/php53 was too large for the amiga.  Also remove an old
> presumably outdated comment.

There is still a strange 224MB limit somewhere. The last vmparam.h patch,
which extends MAXDSIZ to 416MB, does not work!

I noticed that programs are now running out of memory after allocating not
much more than 30MB, although I have 128MB RAM and 512MB swap.

So I made a series of tests with kernels having a different MAXDSIZ and
wrote a small test program which tries to allocate as much memory as
possible with malloc(). This is the result:

MAXDSIZ   max. malloc-size
416       63
260       186
256       190
240       206

Seem like the optimal value for MAXDSIZ would be arround 224MB, which gives
a maximum data segment of 224MB.

Can somebody explain that?

-- 
Frank Wille
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.