Re: [Xen-devel] [xen-4.3-testing test] 63948: regressions - FAIL [and 1 more messages]

[Ian Jackson <Ian.Jackson@xxxxxxxxxxxxx>]

Jan Beulich writes ("Re: [Xen-devel] [xen-4.3-testing test] 63948: regressions 
- FAIL"):
> On 11.11.15 at 16:58, <Ian.Jackson@xxxxxxxxxxxxx> wrote:
> > Do we know why this cross-version migration test fails when
> > within-version migrations of both 4.2 and 4.3 succeed ?
> 
> As said in the original reply to the test report, this is due to vMCE
> (validly) rejecting the incoming feature set. On 4.2, regardless of
> CPU vendor, certain features got advertised to the guest. On 4.3
> that advertisement became CPU vendor specific.

And 4.3 checks the incoming advertisement, I see.

> > Can things easily be fixed ?
> 
> Not sure. We could cheat and accept the known bogus value. But
> should we even consider such on a no longer maintained branch?

I wonder how easy it would be to filter out the wrong advertisement,
during outgoing migration, in the toolstack in 4.2.  I appreciate that
4.2 is long-dead.  So maybe we should just leave it.  Users can shut
down and restart their guests.

Ian Campbell writes ("Re: [Xen-devel] [xen-4.3-testing test] 63948: regressions 
- FAIL"):
> On Wed, 2015-11-11 at 15:58 +0000, Ian Jackson wrote:
> > Also we don't know (ATM) whether 4.3->4.4 works.  4.4->4.5 does.

From what Jan says, at least this specific bug does not affect
4.3->4.4.

> Are you basing that on 
> http://logs.test-lab.xenproject.org/osstest/results/history/test-amd64-amd64-migrupgrade/xen-4.4-testing.html
> and 
> http://logs.test-lab.xenproject.org/osstest/results/history/test-amd64-amd64-migrupgrade/xen-4.5-testing.html

Yes.

> Plus:
> [i386]

Yes.

I think we should consider introducing
  test-amd64-<something>-hvm-migrupgrade-xsm

Ian.

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