[Wine]old Wine from CVS

[Julian Hall]

----- Original Message ----- 
From: "Duane Clark" <dclark at akamail.com>
To: <wine-users at winehq.com>
Sent: Saturday, October 30, 2004 8:46 PM
Subject: Re: [Wine]old Wine from CVS


> Take what I say here with a caution; I am not really very knowledgable
> in this area. In general, I believe on modern operating systems
> including Windows ;) and computers, that is not really possible. Each
> program has it's own virtual address space, generally of some very large
> amount like 2-3GB, even if you don't have that much real memory. And one
> program is not able to encroach on another's address space, even with
> bad programming. It sounds like magic, but it is really sophisticated
> memory management that actually seems to work. In general, memory
> corruption is a problem where a program writes over something within
> it's own address space that it was not supposed to, causing itself to
> crash but nothing else. And that appears to be what is happening here.
>
> That is of course a separate problem from memory exhaustion. But even
> then all that should happen is disk swapping, which slows program
> operation to a crawl, rather than a crash. All three of us have the same
> symptoms, so that is unlikely to be the problem. In my case, I have 1GB,
> and don't come close to exhausting memory.

Hi Duane,

I'm sorry to correct you, but it is indeed possible.  Through bad
programming/carelessness, Netscape 2.0* was designed to use the same memory
address area as certain Cirrus Logic graphics cards.  I was working in ISP
Tech Support at the time, and in every case we had to set the graphics card
to use an alternate configuration, or advise the customer to use Internet
Explorer, then in its' infancy so not the option we generally preferred.
While hopefully programmers take more care these days I would not rule out
the possibility.

I'm not saying two items, be they programs or hardware devices, *can* use
the same memory addresses at the same time.  I'm saying they can *try*.  I'm
saying is that it IS possible for a programmer to inadvertently tell his
software to use the same memory address as another program/hardware device
is already using, thus creating a conflict.

Take a very simple example:

If a graphics card was using addresses 5-10.  A programmer wrote a piece of
software with various routines, one of which was told to use address 10.
The program would not crash until the user selected the routine using
address 10.

I accidentally caused a similar problem with Windows 98 on one occasion.  As
you may know, Win98 will not boot on a system with more than 768Mb.  I had
1.5Gb, so I editted the win.ini to tell the system I only had 256Mb (the
recommended solution).  Windows would then boot.  However the graphics card
complained bitterly because the memory addresses it chose *before* Windows
booted were no longer available.  Another example of software playing with
hardware, and vice versa.

Kind regards,

Julian