Date: Sun, 6 Jan 2008 15:25:49 -0800 From: Marcel Moolenaar <xcllnt@mac.com> To: grehan@FreeBSD.org Cc: Rafal Jaworowski <raj@FreeBSD.org>, Perforce Change Reviews <perforce@FreeBSD.org> Subject: Re: PERFORCE change 132485 for review Message-ID: <4CCF7BFD-9DD3-4D69-8113-D5DC3E8F4B27@mac.com> In-Reply-To: <47815371.9020501@freebsd.org> References: <200801041525.m04FPIrt011288@repoman.freebsd.org> <47815371.9020501@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Jan 6, 2008, at 2:17 PM, Peter Grehan wrote: > Hi Rafal, > >> Temporary fix to avoid confusion of 'kstack0' symbol and make e500 >> kernel operational again. kstack0 area requires further >> improvements (space for kstack0 is now statically defined within >> kernel body, does not have guard pages etc., which should be fixed) > > On AIM it is allocated correctly but never switched over. The (not- > yet-implemented) plan was to mimic amd64 and return the value from > powerpc_init() to locore, switching r1 to this value before calling > mi_startup(). I'm planning this. I have done the same on ia64. It helps SMP, because after you switched the BSP to its own stack kstack0 becomes available to start APs on (in particular before you have the MMU set up)... FYI, -- Marcel Moolenaar xcllnt@mac.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4CCF7BFD-9DD3-4D69-8113-D5DC3E8F4B27>