On Mon, 21 Sep 2020 at 10:04, Ed Maste <[email protected]> wrote: > > On Mon, 21 Sep 2020 at 03:42, Walter von Entferndt > <[email protected]> wrote: > > > > At Sonntag, 20. September 2020, 14:00:00 CEST, Ed Maste <[email protected]> > > wrote: > > > Until pkgbase is in place the most straightforward way to address this > > > issue will probably be for us to ship a new kernel even if not > > > necessary, just to update the version number. > > -1 from all with limited bandwith and/or volume. > > Updates should be as small as possible. > > The updates are delivered as binary patches (bspatch); a patch to > change the kernel version is probably less data than this email.
Colin pointed out on IRC that there's a caveat to this: bspatch allocates enough memory to store the entire original file, and the entire modified file, and performs all of the patching operations in memory. This might fail on extremely memory-constrained systems, with freebsd-update then falling back to fetching the full file. However, this is an implementation detail, not a fundamental aspect of the bsdiff/bspatch format. I believe it should be possible for bspatch to operate in a (perhaps slower) reduced memory mode, if it turns out this is a significant issue in practice. _______________________________________________ [email protected] mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "[email protected]"
