From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <3C4C7C4E.4020506@embeddededge.com> Date: Mon, 21 Jan 2002 15:38:38 -0500 From: Dan Malek MIME-Version: 1.0 To: Mark Pilon Cc: linuxppc-embedded@lists.linuxppc.org Subject: Re: KERNELBASE & large memory systems References: <3C4A96D7.9010501@embeddededge.com> <3C4C2EB6.6F3ACADE@midrivers.com> Content-Type: text/plain; charset=us-ascii; format=flowed Sender: owner-linuxppc-embedded@lists.linuxppc.org List-Id: Mark Pilon wrote: > I've heard that one shouldn't modify the definition of KERNELBASE > as it screws up the MM code; That was true until yesterday. Matt Porter checked in some configuration changes that if you know what you are doing will change the virtual mapping windows on most processors (not 4xx or 8xx yet). They are in the linuxppc_2_4_devel kernel. > I'm wondering how people out there are handling designing systems > (and getting linux running) on large pools of memory -- like > 1 - 2 Gb or so. The only general answer to that question is the solution depends upon your configuration. A combination of highmem and kernel address base changes are used. This will also affect drivers if you have some custom application and you are trying to use a driver that makes assumptions about the "normal" kernel mapping. The "know what you are doing" is critical to your success :-). Good Luck. -- Dan ** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/