Re: [xen-4.11-testing test] 136385: regressions - FAIL

From: Julien Grall <julien.grall@arm.com>
To: Ian Jackson <ian.jackson@citrix.com>,
	Jan Beulich <JBeulich@suse.com>,
	Stefano Stabellini <sstabellini@kernel.org>
Cc: "xen-devel@lists.xenproject.org" <xen-devel@lists.xenproject.org>
Subject: Re: [xen-4.11-testing test] 136385: regressions - FAIL
Date: Mon, 20 May 2019 15:35:40 +0100	[thread overview]
Message-ID: <7bea4485-b0da-ed92-336d-b2307c444cb1@arm.com> (raw)
In-Reply-To: <23778.33200.47146.497482@mariner.uk.xensource.com>

Hi Ian,

On 20/05/2019 11:30, Ian Jackson wrote:
> osstest service owner writes ("[xen-4.11-testing test] 136385: regressions - FAIL"):
>> flight 136385 xen-4.11-testing real [real]
>> http://logs.test-lab.xenproject.org/osstest/logs/136385/
>>
>> Regressions :-(
>>
>> Tests which did not succeed and are blocking,
>> including tests which could not be run:
>>   test-amd64-amd64-xl-qemut-debianhvm-amd64 10 debian-hvm-install fail REGR. vs. 135683
>>   test-amd64-amd64-xl-qemut-debianhvm-i386-xsm 10 debian-hvm-install fail REGR. vs. 135683
> 
> These are a known bug.  The Debian "stretch" ("stable") i386 kernel
> simply does not work under Xen HVM.  I don't thnk this is anything to
> do with the 4.11 Xen commits under test here.  It is appearing here in
> this flight because I reorganised the amd64/i386 test jobs.
> 
>>   test-armhf-armhf-xl-credit2 16 guest-start/debian.repeat fail REGR. vs. 135683
> 
> I looked at this.  I thought it might be the known xenstore memory
> handling bug but it doesn't seem to be.  I looked at the logs and the
> guest seems simply not to have started at all.  There is nothing from
> it in its pv console.  Adding some ARM folks.  I have a memory of some
> allegation that there is a fundamental cache mismanagement problem
> which might cause this, in which case we probably ought to force push.

There are indeed some cache mismanagement on Xen 4.11, but they were never 
reliable. In this case, they seem to be reliable. So this may well be different.

Before the last 2 flights, the kernel was compiled natively. Now it is 
cross-compiled on x86 machine. So there are a difference in the way the kernel 
is built.

Could we try a flight with kernel built from the cubietruck?

On a side note, I noticed that the name of the file [1] leads to think it is a 
gzip, but it is actually a plain text.

Cheers,

[1] 
http://logs.test-lab.xenproject.org/osstest/logs/136385/test-armhf-armhf-xl-credit2/cubietruck-gleizes---var-log-xen-console-guest-debian.guest.osstest.log.gz

> 
> Ian.
> 

-- 
Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel