On Fri, Jan 13, 2017 at 12:57 PM, Antti Kantee <[email protected]> wrote: > On 10/01/17 05:11, Hui Kang wrote: >>>> >>>> (m2) I kind of cherry-pick the libtun component (I created a PR to >>>> track the commit [1]) and be able to make the compilation work. With >>>> this PR, the rump kernel client can use the tun component of a rump >>>> kernel. This is not a real "cherry-pick" because the upstream netbsd >>>> has made some change to the app-src branch of rumpkernel/src-bsd [2]. >>>> [2] >>>> >>>> https://github.com/huikang/netbsd-src/commit/60a241be2b08a4c81dda8b9b8ec09d2bbe6fba3f#diff-c871c9f5be0e8b75c6fd3de9ea46fa2a >>> >>> >>> >>> Can you expand on why changes to the tap driver are relevant? >> >> >> No, the change to tap driver in upstream is irrelevant. The point is >> to show why a >> blind full sync does work. > > > I still don't get it. There are probably thousands of changes. Picking one > of them doesn't, without further explanation, show why things wouldn't work. > >> I am not sure whether I fully understand the checkout.sh script. But I > > > Check out the history of that file. History-perusing also generally a good > technique when trying to understand something. > >> created a PR with my change to checkout my src-netbsd [3]. However, >> when I run >> >> ./checkout.sh githubdate src >> >> It hangs at checking out upstream netbsd src by csv (the full log of >> the above command is [4]). Could you take a look at this? Thanks > > > I assume you don't have access to the cvs host, so you need to use anoncvs. > Set BUILDRUMP_CVSROOT before running the script.
I figured out the issue and updated the PR [3]. Could you take a look? Thanks. - Hui > > >> [3] https://github.com/rumpkernel/buildrump.sh/pull/85 >> [4] https://paste.fedoraproject.org/524917/84025053/ > >
