> -----Message d'origine----- > De : dman [mailto:[EMAIL PROTECTED] > Envoyé : jeudi 8 novembre 2001 04:35 > À : debian-user@lists.debian.org > Objet : Re: max RAM size > > > On Wed, Nov 07, 2001 at 08:19:21AM -0800, Jeffrey W. Baker wrote: > | > | On Tue, 6 Nov 2001, François THOMAS wrote: > | > | > Hello list > | > > | > I have upgraded the physical amount of RAM on a Potato server (r3). > | > Unfortunately, it looks like only 960M are managed by the > kernel... Is there > | > a *safe* way to make my system manage all the available RAM > (=> 2 Gigs) ? > | > This is a production server, and I cannot afford the risk of > destroying the > | > system (I would prefer to continue using half the available memory !). > | > Any advice ? > | > | Recompile your kernel with higmem enabled. > > IIRC (which is not very probable recently) that is only necessary for > > 4GB of RAM. >
I found the following info on http://www.vmware.com/support/reference/common/big_mem.html : "By default, Linux kernels in the 2.2.x series support 1GB of physical memory. [...2 solutions:] Utilize the CONFIG_2GB option and recompile your kernel as a 2GB kernel Enable the CONFIG_BIGMEM option to map more physical memory" I did not try to pass mem=2000M to the kernel, because i already found in the log something like "Warning only 960M will be used", which sounds like more RAM has been correctly reported by the bios. I think I will test kernel compilation on another box, not a production one. Anyway, I wanted to have a rescue server ready just in case, so it will not be wasted time...
