Re: [Xen-devel] [xen-unstable test] 15401: regressions - FAIL

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Mon, 2013-02-04 at 11:17 +0000, Jan Beulich wrote:
> >>> On 04.02.13 at 12:06, Ian Campbell <Ian.Campbell@xxxxxxxxxx> wrote:
> > On Fri, 2013-02-01 at 11:44 +0000, Ian Jackson wrote:
> >> xen.org writes ("[xen-unstable test] 15401: regressions - FAIL"):
> >> > flight 15401 xen-unstable real [real]
> >> > http://www.chiark.greenend.org.uk/~xensrcts/logs/15401/ 
> >> > 
> >> > Regressions :-(
> >> > 
> >> > Tests which did not succeed and are blocking,
> >> > including tests which could not be run:
> >> >  test-amd64-amd64-xl-qemut-win  7 windows-install     fail REGR. vs. 
> >> > 15179
> >> 
> >> With some handholding, I managed to get the bisector to work on this.
> >> It found that the original "good" version is unreliable: it built Xen
> >> 5af4f2ab06f3 and in two recent tests on the same host, of the same
> >> build, it failed once and passed once.
> > 
> > Hrm, did it make any progress over the w/e.
> 
> With the failure now being consistent rather than intermittent,
> we almost definitely have a state worse than before.
> 
> >> Under the circumstances it's not clear that the current staging is any
> >> worse than non-staging.  I think we should push the revision reported
> >> in this test (which was otherwise OK according to the tester) to
> >> non-staging, with a manual "hg push".
> > 
> > This sounds like a good idea.
> 
> Wouldn't that set us up for the same problem again when the next
> testing round fails here again?

Yes, that's true.

> 
> Unless Olaf's testing with partial reverts shows otherwise, I'd be up
> for reverting all non-trivial x86 HVM RTC patches I had applied
> recently (where "trivial" to me would be "use RTC_* names instead of
> literal numbers" and "use cached original value in RTC_REG_B writing
> code", albeit the latter may not revert cleanly on its own). Should
> they turn out not to be the culprit, they could always be re-applied
> later.

We should certainly see what Olaf's test shows. I'd also be interested
in what (if anything) the bisector has discovered. But then yes, if we
think these might be the culprit then reverting would be sensible.

Ian.



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