Re: More toolchain issues

["Michael L. Hitch" <mhitch%lightning.msu.montana.edu@localhost>]

On Tue, 29 Sep 2009, Frank Wille wrote:

> >  I think
> > the amiga could easily move USRSTACK to 0x1c000000 [any larger will
> > run into a pmap limitiation that can allow user programs to crash a
> > 680[46]0 system].
>
> Ok. I'm all for it! This gives us an adress aspace of more than 256MB,
> which should be enough for 68k Amigas.
>
> AFAICS mac68k still offers SunOS-compatibility, although their USRSTACK
> is not at 0x0e000000.

  Actually, the amiga can handle USRSTACK at 0x1e000000 without the 
ability of a user program causing a panic.

  This does come at a slight cost though - more kernel memory will be
used (I don't have the size at the moment).

  But....  I don't think it's really necessary.  If the mmap() error 
return bug is fixed, jemalloc will fall back to using sbrk() when the 
mmap() fails, and can thus utilize all the alloted datasize space
[which will be in addition to what ever memory is available between the 
shared libraries and user stack].

  I wonder how many of the other m68k ports using the pmap_motorola pmap
with 4KB pagesize are being used on 680[46]0 systems with 5.0 or -current. 
It seems to me that jemalloc can easily panic the system with programs 
that attempt to allocate huge amounts of memory.

--
Michael L. Hitch                        mhitch%montana.edu@localhost
Computer Consultant
Information Technology Center
Montana State University        Bozeman, MT     USA