Re: glibc, libstdc++2.8 and the kernel
In article <[🔎] 35E19225.D63DFD55@livenet.net> you wrote:
: My understanding is that *if* you could compile a kernel (sources are
: truly NOT ready yet), you could not then install it *safely*. The
: current bootrom does not have multi kernel support. So, if you hose the
: kernel somehow -- there is little means of safe recovery.
The recovery FAQ indicates a process for doing a diskless boot from a network
server.
If we can build a kernel and point it to a different partition for the root
filesystem, it should be fairly easy to leave the current bits completely
alone, using the chroot to build packages, install the packages onto the
Debian-only partition, then net-boot a Debian kernel that picks up the
Debian-only partition as a root. This is a tad twisted, and it would certainly
be easier if the monitor had multi-boot already, but this should be more than
sufficient to bootstrap ourselves to a base tarball.
I'll sit up tonight and try to network boot the current kernel. If I can get
that to work at all, putting my disk at risk is a no-brainer, since I have
zero interest in running the bits that shipped with the machine beyond getting
Debian bootstrapped. I started by cutting a tape of the machine as received
before I touched anything, so recovery shouldn't be too hard if I hose things
up.
Bdale
Reply to: