Re: [Xen-devel] [xen-unstable test] 97737: regressions - FAIL

[Julien Grall <julien.grall@xxxxxxx>]

On 25/07/16 12:53, Wei Liu wrote:
> On Mon, Jul 25, 2016 at 12:34:53PM +0100, Julien Grall wrote:
> > On 25/07/16 12:11, Wei Liu wrote:
> > > Thanks for investigating.
> > >
> > > There are only two arm related changes in the range being tested:
> > >
> > > * a43cc8f - (origin/smoke) arm/traps: fix bug in dump_guest_s1_walk handling of level 
> > > 2 page tables (5 days ago) <Jonathan Daugherty>
> > > * 60e06f2 - arm/traps: fix bug in dump_guest_s1_walk L1 page table offset computation 
> > > (5 days ago) <Jonathan Daugherty>
> > >
> > > They don't look very suspicious.
> > The modified function is not called in the hypervisor at all. It's only here
> > for manual debugging.
> >
> > Although, this may change the offset of some function (assuming we have an
> > hidden bug).
> >
> > > If you need help navigating osstest test report, please let me know.
> > I have noticed that there is 2 kernel BUG in the logs (with one host reboot
> > in the middle). Can you detail what the exact the test?
> What I normally do is to look at the summary page of the failed test to
> identify the failed step and the time.
>
> In this case:
>
> http://logs.test-lab.xenproject.org/osstest/logs/97737/test-armhf-armhf-xl/info.html
>
> The time stamp said the failed step started at 2016-07-21 19:30:10 Z,
> and then I look at the output of failed step log to look for time stamp
> that the test failed.  Then I would look for output between these two
> time stamps in various log.
>
> I now realise the log I pasted in was not from the failed test. I wanted
> to paste in the second kernel oops, which should be the culprit that
> the test failed. The two oops was the same one, though.
>
> To identify  which test step was running when the first oops happened,
> the same technique applies.
>
> It seems that the oops happened during ts-debian-install according to
> time stamps.
I looked at the new xen-unstable report [1], test-armhf-armhf-xl is 
passing (though it is on arndale and not a cubietruck).
Is there any way to re-run the previous test on the cubietruck? Just to 
confirm the bug is reproducable.
Regards,

[1] http://logs.test-lab.xenproject.org/osstest/logs/99605/

--
Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel