On Thu, 2020-05-14 at 15:44 -0400, Bruce Ashfield wrote:
> On Thu, May 14, 2020 at 10:22 AM Bruce Ashfield via
> lists.openembedded.org
> <[email protected]> wrote:
> > On Thu, May 14, 2020 at 10:09 AM Richard Purdie
> > <[email protected]> wrote:
> > > On Wed, 2020-05-13 at 10:24 -0400, [email protected] wrote:
> > > > From: Bruce Ashfield <[email protected]>
> > > > 
> > > > Hi all,
> > > > 
> > > > This is a resend of the v5.4 stable bump to v5.4.38, plus a new
> > > > -stable
> > > > bump to v5.4.40.
> > > > 
> > > > The extra changes to make these pass the AB tests are the kernel
> > > > reproducibility commits.
> > > > 
> > > > The issue we had on the last submission was that IKHEADERS was
> > > > switched
> > > > to a module, and hence was failing reproducibility tests as we didn't
> > > > have the timestamp variables exported for the module build. It has
> > > > always been a problem, just one we hadn't hit.
> > > > 
> > > > I've left the fix as two commits on purpose, since the copy of the
> > > > reproducibility export fixes the problem, and is definitely safe for
> > > > dunfell,
> > > > and the second commit is my effort to remove the duplication of logic
> > > > and
> > > > create a common python routine they can call. I've used the
> > > > techniques for
> > > > subprocess running that I've done in other kernel build elements, but
> > > > if they
> > > > aren't the right ones to use, point me in the right direction.
> > > > 
> > > > All said, these are green on the AB and should be ok for master and
> > > > dunfell.
> > > 
> > > I think this triggers something here:
> > > 
> > > https://autobuilder.yoctoproject.org/typhoon/#/builders/47/builds/1914
> > > 
> > 
> > In that case, just drop my effort at factoring out the source date
> > epoch into a python function.
> > 
> > It looks like the git check went nuts, and since we don't need it, and
> > I won't have time to work on it again for a while, if you toss it in
> > the bin, all should be well.
> 
> I did tweak the python code to handle the git return code more
> gracefully. I did test all the conditional paths before, and they
> worked, so I'm not sure how to trigger the failure here.
> 
> I can try an AB run against the fixed up routine, but what target
> should I be running ? Obviously quick doesn't hit it, or I would have
> seen it.

You need to push a plain oe-core branch to oe-core-contrib and point
the build at that. This is the target:

https://autobuilder.yoctoproject.org/typhoon/#/builders/47

(qemuarm-oecore)

Not setting a DISTRO would also do it.

Cheers,

Richard



-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#138292): 
https://lists.openembedded.org/g/openembedded-core/message/138292
Mute This Topic: https://lists.openembedded.org/mt/74182743/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub  
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to