Date: Tue, 15 Jan 2013 08:49:28 -0800 From: Marcel Moolenaar <marcel@xcllnt.net> To: Lawrence Stewart <lstewart@freebsd.org>, Robert Noland <rnoland@freebsd.org> Cc: Marcel Moolenaar <marcel@freebsd.org> Subject: Re: svn commit: r245399 - projects/altix2/sys/geom/part Message-ID: <C8FA6792-669F-4C88-88BF-8ABA7B0DD371@xcllnt.net> In-Reply-To: <50F55F11.3040005@freebsd.org> References: <201301132336.r0DNaLb7038996@svn.freebsd.org> <50F55F11.3040005@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Jan 15, 2013, at 5:52 AM, Lawrence Stewart <lstewart@freebsd.org> wrote: > On 01/14/13 10:36, Marcel Moolenaar wrote: >> Author: marcel >> Date: Sun Jan 13 23:36:21 2013 >> New Revision: 245399 >> URL: http://svnweb.freebsd.org/changeset/base/245399 >> >> Log: >> Marking the PMBR as active may be needed for some BIOSes to boot from the >> disk, but also has some EFI implementations reject the GPT altogether. >> Make the behaviour controllable by a sysctll and have it default to 1 >> on i386 and amd64 and 0 otherwise. This means that we mark the PMBR as >> active on i386 and amd64 by default, but remain compliant most of the >> time. > > Would I be correct in thinking this provides a relevant workaround to > the issue I reported a few months back? > > http://lists.freebsd.org/pipermail/freebsd-fs/2012-August/014864.html > > Sounds like you got bitten by the same problem as me? Excellent description! Yes, we're bitten by the same thing. [I moved the src commit aliases to BCC so that they'll drop off and I added rnoland@] Robert: with r198097 you changed the GPT scheme to mark the PMBR partition as active. This trips up with at least some (U)EFI environments (both ia64 & x86). Can you describe the scenario you we're fixing so that we can better assess what to do with what appears to be a compatibility conflict. In general: I think we need to define what we're going to support and how. If for example we're on a x86 with a legacy BIOS and we want to partition the disk with GPT, but find that we cannot boot unless we make GPT non-standard then we should think twice about supporting that. Put differently: if there are BIOSes around that don't work well with GPT, then we should not try to make it work at the expense of "strict" compliant implementations. This is not to say anything about r198097 -- I don't know what the underlying reason for that change is precisely. -- Marcel Moolenaar marcel@xcllnt.net
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?C8FA6792-669F-4C88-88BF-8ABA7B0DD371>