From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jesse Barnes Date: Thu, 08 May 2003 00:00:45 +0000 Subject: Re: [Linux-ia64] Re: [PATCH] head.S fix for unusual load addrs Message-Id: List-Id: References: In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: linux-ia64@vger.kernel.org On Wed, May 07, 2003 at 04:51:21PM -0700, David Mosberger wrote: > Hi Tony, > > Great summary! Yeah, thanks. It's a good outline of our options. > Tony> 3) David's suggestion of boot-time relocation. Probably > Tony> simplest to implement this in elilo, but if you are really > Tony> good at PIC asm code it could be done in the kernel startup > Tony> sequence. > > Agreed. Anyone who's ever debugged the ld.so startup code will agree > that it's much better to do the relocation in elilo. > > Tony> pros) Just like linking kernel at a new address. Avoids the > Tony> __tpa() issue. Doesn't invalidate any assumptions about how > Tony> to get from virtual to physical addresses and back again. > > Tony> cons) Nobody has implemented it. > > I think: > > - kernel text addresses vary depending on physical memory conf. > > is potentially a bigger issue. Implementing the relocator in elilo > should be really easy (just take the kernel loader and massage it a > bit). > > How many tools are out there that depend on System.map or similar? I > suppose you could just generate a fresh System.map at boot time (or > have something like /proc/System.map). Ahh, this might be a good way to go. We'd basically be relinking the kernel at load time with a new KERNEL_START, right? I haven't looked at the new module loader code yet (in fact, I've never done any linker stuff), but I will now. Thanks, Jesse