From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ian Campbell Subject: Re: [PATCH OSSTEST] get_hostflags: return an empty list when there is no flight/job. Date: Tue, 11 Aug 2015 16:57:57 +0100 Message-ID: <1439308677.9747.292.camel@citrix.com> References: <1438187304-17039-1-git-send-email-ian.campbell@citrix.com> <21947.38096.288929.877968@mariner.uk.xensource.com> <1438357178.30740.70.camel@citrix.com> <1438771857.9747.56.camel@citrix.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1438771857.9747.56.camel@citrix.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Ian Jackson Cc: xen-devel@lists.xen.org List-Id: xen-devel@lists.xenproject.org On Wed, 2015-08-05 at 11:50 +0100, Ian Campbell wrote: > On Fri, 2015-07-31 at 16:39 +0100, Ian Campbell wrote: > > On Fri, 2015-07-31 at 16:31 +0100, Ian Jackson wrote: > > > Ian Campbell writes ("[PATCH OSSTEST] get_hostflags: return an empty > > > list > > > when there is no flight/job."): > > > > From: Ian Campbell > > > > > > > > Otherwise trying to use mg-hosts mkpxedir fails with: > > > > > > I think your proposed fix is incorrect. It is wrong to call > > > get_hostflags outside the context of a job, because get_hostflags is > > > supposed to return the job's host flags for that ident. > > > > > > The bug was introduced by me in 11e788f7 "JobDB/Executive: Improve an > > > internal `die' error", where a refactoring meant that we always call > > > get_hostflags. > > > > > > How about this instead ? > > > > > > diff --git a/Osstest/JobDB/Executive.pm b/Osstest/JobDB/Executive.pm > > > index 1ec947e..cc52f57 100644 > > > --- a/Osstest/JobDB/Executive.pm > > > +++ b/Osstest/JobDB/Executive.pm > > > @@ -128,7 +128,7 @@ sub host_check_allocated ($$) { #method > > > $ho->{Shared} && > > > $ho->{Shared}{State} eq 'ready'; > > > my $harness = get_harness_rev(); > > > - my @flags = get_hostflags($ho->{Ident}); > > > + my @flags = defined($job) ? get_hostflags($ho->{Ident}) : > > > qw(OUTSIDE > > > -JOB); > > > $ho->{SharedReady}= > > > $ho->{SharedMaybeOthers} && > > > !! (grep { $_." ".$harness eq "share-".$ho->{Shared}{Type} } > > > > > > Signed-off-by: Ian Jackson > > > > LGTM, I suppose the use of the OUTSIDE-JOB sentinel value is just for > > the > > benefit of the reader of the following die() should it occur. > > > > Acked-by: Ian Campbell > > I fabricated some sort of commit message and pushed to pretest. > > commit 2e51119a34d06162b69275f38010130193d5501e > Author: Ian Jackson > Date: Wed Aug 5 11:42:10 2015 +0100 > > Executive: Support host_check_allocated outside a job. > > When called outside a job there are no hostflags, so get_hostflags > cannot be used. Instead assume a new pseudo-flag "OUTSIDE-JOB" when > there is no $job. > > Otherwise uses of select_host such as "mg-hosts mkpxedir" fail. > > Signed-off-by: Ian Jackson > Acked-by: Ian Campbell > [ ijc -- wrong commit message ] I really meant wrote here, sorry. It's through the gate now though.