Date: Tue, 10 Nov 2009 22:32:28 -0600 From: Nathan Whitehorn <nwhitehorn@freebsd.org> To: Peter Grehan <grehan@freebsd.org> Cc: FreeBSD PowerPC ML <freebsd-ppc@freebsd.org> Subject: Re: 8.0-RC2 on Rackmac3,1 - fatal kernel trap Message-ID: <4AFA3E5C.4060003@freebsd.org> In-Reply-To: <4AF9E9F5.8000207@freebsd.org> References: <aff3f9930911041456o732ea890n6cda52783e8db286@mail.gmail.com> <4AF219DF.2040309@freebsd.org> <B06424FF-FE87-441E-A8C3-ED401A8561FE@gmail.com> <aff3f9930911060851l12e52e87n653b8f96ba8cb7d7@mail.gmail.com> <4AF46399.20803@freebsd.org> <aff3f9930911061219g61be95a9l40884c854c7b4971@mail.gmail.com> <4AF575D2.5020101@freebsd.org> <4AF577F6.2080708@freebsd.org> <4AF90970.4080708@freebsd.org> <4AF9E9F5.8000207@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Peter Grehan wrote: > Hi Nathan, > >> This problem should be fixed now. > > Almost :) My 7,2 and 11,2 G5's now hit the panic: > > 7,2: 120 translations > > 11,2: 126 translations > > If you are concerned about BSS usage for holding the translations, it > might be possible to bump up the loader's unit of allocation from 16 > pages to 64 when on a G5. See the history of that at > > http://svn.freebsd.org/viewvc/base/head/sys/boot/ofw/libofw/ofw_copy.c?view=log > Drat. I think this is something inside OF that only applies to CD booting, since 64 was more than enough for netbooting the kernel or booting from a hard drive, so changes to the loader probably won't affect anything. Also, it is now too late to get any changes to this code into 8.0, which is unfortunate. I guess the size of the buffer just needs to be bumped to 128 (or more?). I'm a little concerned about bloating the kernel here. I think there is enough headroom in the stack that it can be moved to the local stack frame, so that's a possibility. It can't be allocated dynamically, since there is no way to add memory that OF can use at that point in the boot process. -Nathan
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4AFA3E5C.4060003>