Re: 2.4.21-pre4 comparison bugs (More of those)

[Oleg Drokin <green@namesys.com>]

Hello!

On Sun, Feb 09, 2003 at 10:01:30PM +0000, Alan Cox wrote:
> > This time I changed the type of variable to signed type whenever
> > I felt it was appropriate.
> > When I was not sure (or unsigned type was in some commonly used
> > structure), I still used a cast just to highlight a problem, so that someone
> > more knowledgeable created better fix.
> > See the patch.
> > Mostly we do incorrect stuff on errors. Sigh, nobody likes errors ;)
> Hiding them is even worse than having them there visible and unfixed.
> Changing the sign on stuff holding physical addresses is actually
> introducing real bugs

I assume you are speaking of slram stuff here.
I thought that slram was not designed to work with parts of RAM past 2G border.
(as far as I remember, slram was used on old x86 HW to convert uncached RAM
beyond 64M (256M for some systems?) into kind of a ramdisk.)

Bye,
    Oleg