Bernd Ernesti <netbsd%lists.veego.de@localhost> writes: > On Sat, Dec 26, 2009 at 07:23:59PM -0500, Greg Troxel wrote: >> >> yancm%sdf.lonestar.org@localhost writes: >> >> It's not just /boot, but also bootxx_*. > this part is not ok because you can have your own options while using > installboot(8). True, so one needs to be careful and store the options, or have installboot not change the options on update. >> > Is there any substantial risk to updating /boot with/after every >> > distribution install? >> >> There's always the risk of trouble, but kernel and /sbin/init could be >> trouble too. /boot does not have a history of being scary. > > There was some trouble in the last months when it went over a 64k size. Good point; I forgot that. >> > Matthias Scheler suggests that: >> > "After the 6.0 release most users should have a module aware >> > boot loader" >> > (see: http://mail-index.netbsd.org/current-users/2009/12/22/msg011788.html) >> > >> > How does that happen? >> >> Indeed - it usually seems one needs a /boot and maybe bootxx_foo from a >> relatively recent system to boot the next one. > > Hmm, the one change here which requires an update from an older version > was the modules. If you use a kernel without MODULAR support it still > would have worked. Sure, but I meant the change somewhere between 2.0 to 4.0 on sparc. I tend to upgrade systems in place for years and so have had some really ancient (1.6) bootblocks.
Attachment:
pgpKc8jQ3QqmG.pgp
Description: PGP signature