Re: Memory upgrade and kernel panic

[ktb <x.y.f@home.com>]

On Thu, Feb 01, 2001 at 09:26:17AM -0500, Hall Stevenson wrote:
> > On Thu, Feb 01, 2001 at 02:19:54PM +0100, Joris
> > Lambrecht wrote:
> > > i think you should try adding something like
> > >
> > > mem="256"
> > >
> >
> > Just to clarify that would be -
> > append="mem=256M" but..
> >
> > The kernel or bios or whatever takes up part of the first
> > part of the memory.  So if you add the full amount of
> > RAM to your /etc/lilo.conf the kernel will freak out.  Back
> > off a MB.  For example I have 128 MB of RAM and my
> > lilo.conf line reads -
> > append="mem=127M"
> 
> That append line hasn't been needed since kernel 2.0.36, has it ?? I
> remember when I first started with Linux and Redhat supplied 2.0.34 or
> so (Redhat 5.0) and I needed it since I had 128mb RAM and it only saw
> 64mb. Once I upgraded to 2.0.36, I removed that append line from
> /etc/lilo.conf and it saw all of memory just fine.
> 

I have read that.  In my experience it hasn't worked for me yet, without
the line.

> Can *parts* of memory be bad, as in the 2nd half ?? Trying "mem=127M",
> or "mem=63M", and so on might tell something...
> 

Yes parts can be bad.  This can manifest itself, for example, if you are doing an
install and it funks out at exactly the same place each time.  You make
a good point, you might be able to run a bad stick if you set the append
line to read before the bad spot.  In this case the bad spot must be in
the first MB so it wouldn't tell us much.  I would put a stick of the
old ram in.  Boot the computer, add the append line to lilo.conf, run
lilo.  Shut down put the 256MB stick in and see if it boots.  If not the
stick must be bad.  My thoughts are it is probably bad but to me it
would be worth a try.
kent
   
-- 
I'd really love ta wana help ya Flanders but... Homer Simpson