Re: pvgrub "Error 9: Unknown boot failure" booting Debian Jessie kernel (Was: Re: [PATCH v5 6/9] libxc: create unmapped initrd in domain builder if supported)

[Wei Liu <wei.liu2@citrix.com>]

On Tue, Dec 01, 2015 at 10:31:20AM +0000, Ian Campbell wrote:
> On Tue, 2015-12-01 at 10:21 +0000, Wei Liu wrote:
> > On Tue, Dec 01, 2015 at 10:04:38AM +0000, Ian Campbell wrote:
> > > On Tue, 2015-12-01 at 10:01 +0000, Ian Campbell wrote:
> > > > 
> > > > > > I'll try that and your suggested patch below as well once I get
> > > > > > to the
> > > > > > office this morning.
> > > > > 
> > > > > Thanks.
> > > > 
> > > > The BUG_ON doesn't seem to be triggering. I'm not seeing pfn==0x4d80
> > > > going
> > > > anywhere near kexec_allocate, the highest is 0x4c0f.
> > > > 
> > > > Maybe the issue is that the ->allocate hook (==kexec_allocate) isn't
> > > > called
> > > > from xc_dom_alloc_pad?
> > > 
> > > That seems like it might be the answer, this patchlet fixes it for me:
> > > 
> > > diff --git a/tools/libxc/xc_dom_core.c b/tools/libxc/xc_dom_core.c
> > > index 5d6c3ba..6d3f97a 100644
> > > --- a/tools/libxc/xc_dom_core.c
> > > +++ b/tools/libxc/xc_dom_core.c
> > > @@ -579,7 +579,13 @@ static int xc_dom_alloc_pad(struct xc_dom_image
> > > *dom, xen_vaddr_t boundary)
> > >      }
> > >      pages = (boundary - dom->virt_alloc_end) / page_size;
> > >  
> > > -    return xc_dom_chk_alloc_pages(dom, "padding", pages);
> > > +    if ( xc_dom_chk_alloc_pages(dom, "padding", pages) )
> > > +        return -1;
> > > +
> > > +    if (dom->allocate)
> > > +        dom->allocate(dom);
> > > +
> > > +    return 0;
> > >  }
> > >  
> > >  int xc_dom_alloc_segment(struct xc_dom_image *dom,
> > 
> > Currently there are three places that call dom->allocate (if we include
> > the call in the proposed diff). I think it would be better if we push
> > dom->allocate down to xc_dom_chk_alloc_pages, then refactor
> > xc_dom_alloc_page to use xc_dom_chk_alloc_pages.
> 
> Yes, I was thinking along the same lines after the quick hack above
> confirmed it.
> 
> > Just some thought after a quick look at the code. I will see what I can
> > do after confirming this is the culprit.
> 
> I don't mind finishing this one off (although I'm also equally happy to
> hand it over ;-))

Already working on a patch. :-)