Re: [Xen-devel] [PATCH 4/9] raisin: add a component to build qemu_traditional

[Ian Campbell <ian.campbell@xxxxxxxxxx>]

On Thu, 2015-04-16 at 10:53 +0100, Stefano Stabellini wrote:
> On Thu, 16 Apr 2015, Ian Campbell wrote:
> > On Wed, 2015-04-15 at 18:42 +0100, Stefano Stabellini wrote:
> > > On Wed, 15 Apr 2015, George Dunlap wrote:
> > > > On 04/15/2015 05:11 PM, Stefano Stabellini wrote:
> > > > > On Wed, 15 Apr 2015, Ian Campbell wrote:
> > > > >> On Wed, 2015-04-15 at 16:49 +0100, Andrew Cooper wrote:
> > > > >>>>> It would be nice if there was a --wtih-system-qemu-trad= option, 
> > > > >>>>> but
> > > > >>>>> that has been looked upon unfavourably by the maintainers.
> > > > >>>> Thanks for feedback!!
> > > > >>>> I think Raisin needs something like --with-system-qemu-trad=, so 
> > > > >>>> if Xen
> > > > >>>> doesn't provide it, we'll have to carry a patch in Raisin to do it.
> > > > >>>>
> > > > >>>> Do you have a link to the discussion or previously rejected patch?
> > > > >>>
> > > > >>> I am not sure that there was ever a patch - it may have been a
> > > > >>> discussion I had with ijc in person.
> > > > >>
> > > > >> Possibly I can't remember myself either.
> > > > >>
> > > > >>> I would be +1 for a --with-system-qemu-trad= option to mirror the
> > > > >>> --with-system-qemu, especially as I am now not the only advocate of 
> > > > >>> one.
> > > > >>
> > > > >> In a pre-raisin world I suppose I thought it didn't make too much 
> > > > >> sense
> > > > >> to go through that churn if we weren't going to stop _all_
> > > > >> cloning/downloading from within xen.git.
> > > > >>
> > > > >> Personally I would have considered qemu-trad very much part of the 
> > > > >> long
> > > > >> tail wrt moving things out of xen.git, but anyway.
> > > > >>
> > > > >> I'd like to see raisin used/supported by osstest before we start
> > > > >> accepting thoe sorts of xen.git patches being suggested here.
> > > > > 
> > > > > I think we'll have to carry those patches in Raisin anyway, otherwise
> > > > > how are we supposed to be able to build xen 4.5? I don't think is
> > > > > reasonable to be able to build only 4.6 onward.
> > > > 
> > > > Why isn't that reasonable?  It's a completely new tool.
> > > > 
> > > > Speaking of which -- are we going to try to make any version of raisin
> > > > build any random version of Xen, or are we going to have "releases" of
> > > > raisin (and subsequent branches and tags in the raisin.git tree) for
> > > > each version we want to support?
> > > 
> > > If we cannot do any better, we can resort to that, but as a user I like
> > > the fact that I can use raisin to build any relatively modern xen
> > > version by simply specifying a tag.
> > > 
> > > We can figure out a way to detect whether we need to patch Xen and only
> > > in that case do it.
> > 
> > When Raisin is building Xen 4.5 or earlier it should just let xen.git
> > take care of cloning and building qemu-trad (and anything else which is
> > in xen 4.5 by default). Thinking about patching xen 4.5 on the fly for
> > this is just madness.
> 
> But then the external qemu-traditional build would silently overwrite
> the qemu-traditional binary produced by xen-unstable or fail during
> installation. Both options are not very nice.

No, because if it is building Xen 4.5 then raisin shouldn't be building
qemu-trad at all, it should let xen.git do it.

Ian.


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