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

Re: [Xen-devel] [PATCH v8 05/32] libxc: introduce a domain loader for HVM guest firmware



On Thu, Oct 08, 2015 at 12:04:30PM +0100, Ian Campbell wrote:
> On Thu, 2015-10-08 at 12:43 +0200, Roger Pau Monné wrote:
> > > However I'm unsure that the presence or absence of ELF notes is 
> > > sufficient,
> > > since there is at least the legacy SHT_NOTE section and then __xen_guest
> > > section (see the tail of elf_xen_parse) as well.
> > 
> > AFAICT NetBSD still uses the __xen_guest section, and it works fine with
> > this change:
> > 
> > http://cvsweb.netbsd.org/bsdweb.cgi/~checkout~/src/sys/arch/amd64/amd64/locore.S?rev=1.78&content-type=text/plain&only_with_tag=MAIN
> > 
> > elfloader recognizes the kernel and it's able to load it without issues.
> > 
> > xc_dom_parse_elf_kernel already calls elf_xen_parse later on, while
> > trying to load the kernel, so the functionality it's exactly the same,
> > it's just that we now simply refuse to try to load a kernel without
> > elfnotes with elfloader.
> > 
> > Without this change the elfloader would try to load a kernel without
> > elfnotes just to fail later while parsing it.
> 
> Great, thanks for confirming.
> 
> > > It's a bit of a shame to now have to parse the ELF twice. How abusive 
> > > would
> > > to be to declare that when a xc_dom ->probe hook returns success it is
> > > entitled to rely on the contents of dom->loader_private being preserved
> > > until ->parse is called. In turn this would imply that the first 
> > > successful
> > > probe would be used rather than e.g. probing everything and then picking a
> > > winner from the successful applicants.
> > 
> > We already do this, we stop testing once a probe returns 0 (see
> > xc_dom_find_loader), and use that as the loader, so there's no
> > functional change in this aspect. IMHO, this approach is not very
> > intrusive, would you like me implement it in a followup patch or rather
> > do it in the series?
> 
> I think this series is big enough that we need to start draining the
> pipeline. So lets address this later.
> 
> On the basis that v7 was Reviewed by Andy and Acked by Wei and that I've
> looked at the delta (only) from that:
> 
> Acked-by: Ian Campbell <ian.campbell@xxxxxxxxxx>
> 

v8

Acked-by: Wei Liu <wei.liu2@xxxxxxxxxx>

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


 


Rackspace

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