Re: [PATCH 3/3] init: use do_mount() instead of ksys_mount()
From: Nick Desaulniers
Date: Tue Dec 17 2019 - 13:05:13 EST
On Mon, Dec 16, 2019 at 10:43 PM Dominik Brodowski
<linux@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> On Mon, Dec 16, 2019 at 01:12:28PM -0800, Nick Desaulniers wrote:
> > Shouldn't patches bake for a while in -next? (That way we catch regressions
> > before they hit mainline?)
> >
> > This lit up our CI this morning.
> >
> > https://travis-ci.com/ClangBuiltLinux/continuous-integration/builds
> >
> > (Apologies for missing context, replying via lore.kernel.org directions.)
> > https://lore.kernel.org/lkml/20191212135724.331342-4-linux@xxxxxxxxxxxxxxxxxxxx/
>
> A fix for this issue is already upstream, 7de7de7ca0ae .
I appreciate that. I was just surprised to have no advanced notice;
-next is our "canary in the coalmine." Mainline is usually pretty
stable; if it goes red then we're missing testing coverage somewhere.
For mainline to break suddenly implies that either a pull was merged
from a branch that wasn't flowing into -next, or someone got [un]lucky
with the merge window, or something worse. I saw -next failed the
same day as mainline for the same reason, so I'm going to give you the
benefit of the doubt and chalk it up to luck with timing of the merge
window.
In the future, please give patches more time to soak in -next. We
expect -next to be noisy, mainline not so much.
--
Thanks,
~Nick Desaulniers