[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [xen-unstable bisection] complete build-i386-oldkern

Ian Campbell writes ("Re: [Xen-devel] [xen-unstable bisection] complete 
> On Wed, 2012-05-30 at 09:13 +0100, xen.org wrote:
> >   changeset:   25404:ccad7bed9163
> >   user:        Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>
> >   date:        Tue May 29 16:36:50 2012 +0100
> >       
> >       libxl: introduce libxl__alloc_vdev
> This is probably a false positive due to the race with the missing
> xl.h->_paths.h dependency. See "12988: regressions - FAIL" for details.

No, actually.  If you follow the link
you can see the bisection graph with the tested revisions coloured

The bisector insists on 3 alternating passes/fails with the
before/after revisions before it will finger a changeset.  So it would
have to be a pretty reproduceable race, not a fluke.

If you look at one of the flights listed as `fail' for
`...,ccad7bed9163' - I looked at 12985 [1] - you'll see this:

  cc1: warnings being treated as errors
  libxl.c:1748: error: 'libxl__alloc_vdev' defined but not used
  make[3]: *** [libxl.o] Error 1

And indeed 25404:ccad7bed9163 introduces a static function which isn't
used.  I should have spotted this in review and asked Stefano to add
#ifdefs around it, pending introducing of the caller in the next

Anyway the dependency race is a bug too so I have committed your fix.
libxl tip seems to build for me with make -j4 now.


There isn't a link to this from the bisection page because I'm not
sure how to get dot to do that.  But you can url-hack it by analogy
from any other flight:

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.