On Fri, 2004-10-22 at 00:31 +0200, Frank Lichtenheld wrote: > On Sat, Oct 02, 2004 at 06:01:31PM -0400, Andres Salomon wrote: > > The kernel team is considering dropping 386 support (the 80386 > > processor, not the i386 arch) from Debian. Currently, in order to > > support 386, we include a 486 emulation patch (the patch can be viewed > [...] > > Comments? Thoughts? > > I think the following should pretty much outline the current opinion > of the release team (at least vorlon and me agreed explicetly on it): > > We're in favor of keeping a -386 kernel image that is compiled with > the patch activated and therefor runs on real i386 machines. It > should be mentioned in the release notes and in the description > of the option in the kernel config that it has known security risks > and that there may no fix for this available in the near or even far > future. That leaves i386 users the choice whether they want to accept > the risk or if they want to stay on older software (which probably has > its own risks). As the patch doesn't has any affects on all other > machines we think this is an acceptable solution. >
Works for me. > Has the current image compiled the patch in? (I haven't checked > that yet) Yes, it does. > If yes, there should be no problem at all to implement this solution > (as long as the patch works). If no, the d-i team will have to speak Heh, that's the rumor. Can't say that I've actually tested it. :) > up and say if a new kernel image could still be added before release > with reasonable effort. As most 386 machines will already fail to > satisfy other requirements of d-i (as RAM), it may even be acceptable > only support 386 via upgrades or manual installation... > > I will begin next week with some upgrade tests from woody on a 386 > machine and could then handle the further steps like creation of > a upgrade-i386 directory with backported modutils, initrd-tools > and a current kernel-image. If someone from the kernel or glibc team had access to a real 386, we might be able to make (userspace) support work. Would it be possible to get access to this machine? -- Andres Salomon <[EMAIL PROTECTED]>
signature.asc
Description: This is a digitally signed message part
