On Tue, Nov 25, 2003 at 06:51:15PM -0800, walt wrote: > Brooks Davis answered: > > walt asked: > >>What does 'buildworld' give us that the new kernel might need? > > >The correct toolchain including the compiler and config(8). > > Okay, thanks, that helps. > > Just thinking out loud about worst-case examples for people who > do routinely use 'make world' (like I have for several years). > > I found out first-hand why installworld quits halfway through > when the new executables won't run on the old kernel. I need no > further education on that point ;-) > > I'm thinking, though, that doing 'make kernel' first has a much > lower potential for disaster than 'make world': if I reboot after > a 'make kernel' and the new kernel won't run on the old world, then > all I need to do to recover is to boot the old kernel again and > 'make buildworld'. Seems difficult to do any real harm this way. > > Is this completely wrong-headed? Am I missing something important? If you don't build world before building a kernel you may not be able to build a kernel because you may not have toolchain features required to build it. If it builds, it's probably ok, but it may not build, so your should do a "make buildworld" first. As the statfs change demonstrates, you should install a new kernel before attempting to install a new world. These requirements lead to our recomended order. No other order will insure that you can recover at each step. It it worth noting, that neither the buildworld or buildkernel targets have any effect on the running system other then consuming resources so they are always safe. They may not work, but they won't break your install. It's the installation stage that can get you in trouble if you don't do it right. -- Brooks -- Any statement of the form "X is the one, true Y" is FALSE. PGP fingerprint 655D 519C 26A7 82E7 2529 9BF0 5D8E 8BE9 F238 1AD4
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:31 UTC