[PATCH v3] ci: support Travis CI's aarch64 builds

[Paul Chaignon]

On Fri, Nov 15, 2019 at 07:13:15PM +0300, Dmitry V. Levin wrote:
> On Fri, Nov 15, 2019 at 02:18:21PM +0100, Paul Chaignon wrote:
> > On Thu, Nov 14, 2019 at 10:20:50PM +0300, Dmitry V. Levin wrote:
> > > On Thu, Nov 14, 2019 at 06:37:10PM +0100, Paul Chaignon wrote:
> > > > Travis CI now supports aarch64 builds, launched with the new "arch: arm64"
> > > > tag.  These builds are running inside LXD only for the moment.
> > > > 
> > > > The aarch64 builds tend to be a lot slower than their x86_64 counterparts,
> > > > so several timeout values need to be increased.  I've also noticed a
> > > > smaller performance gain in filter_seccomp-perf.test for some builds.
> > > > 
> > > > Some tests occasionally take more than 10 minutes, hitting Travis CI's
> > > > limit on output inactivity.  Using "travis_wait 40" removes that limit for
> > > > 40 minutes (I've not seen any build last more 40 minutes).
> > > 
> > > It's not quite clear from Travis CI documentation [1] whether
> > > "travis_wait 40" extends the "no output timeout" by extra 40 minutes
> > > or it extends the timeout to 40 minutes.
> > 
> > If I understand the documentation correctly, it's neither: "travis_wait X"
> > gives us X minutes during which the "no output timeout" doesn't apply,
> > because a background process is writing to the build logs every minute.
> > If "travis_wait X" was simply extending the "no output timeout" from 10 to
> > X, then we could choose a value lower than 40 (because it's unlikely that
> > any single test actually takes 40 minutes).
> 
> OK, so "travis_wait 40" essentially delays the moment of starting
> "no output timeout" by 40 minutes.  Given that the build timeout
> is 50 minutes, it effectively turns "no output timeout" off.

Yes.  That's my understanding as well.

Paul