Re: [Qemu-devel] [Qemu-ppc] [PATCH 1/1] Support ppc64le binfmt registration

["Serge E. Hallyn" <serge@hallyn.com>]

Quoting Serge E. Hallyn (serge@hallyn.com):
> Quoting Alexander Graf (agraf@suse.de):
> > 
> > 
> > On 27.08.14 23:35, Peter Maydell wrote:
> > > On 27 August 2014 22:30, Alexander Graf <agraf@suse.de> wrote:
> > >>
> > >>
> > >> On 27.08.14 21:01, Peter Maydell wrote:
> > >>> [adding qemu-ppc]
> > >>>
> > >>> On 27 August 2014 19:45, Serge E. Hallyn <serge@hallyn.com> wrote:
> > >>>> Quoting Peter Maydell (peter.maydell@linaro.org):
> > >>>>> On 27 August 2014 18:55, Serge E. Hallyn <serge@hallyn.com> wrote:
> > >>>>>> This is a part of a patch proposed by jaejunh at
> > >>>>>> https://bugs.launchpad.net/ubuntu/+source/qemu/+bug/1358268
> > >>>>>> which enables running ppc64le binaries through qemu-user.
> > >>>>>
> > >>>>> I'm confused -- wouldn't these be handled by the existing
> > >>>>> "ppc64le" support?
> > >>>
> > >>>> It looks to me like the point was to add a TARGET_ABI32=y
> > >>>> version (and I dunno, something about CONFIG_LIBDECNUMBER=y
> > >>>> for 32-bit).
> > >>>
> > >>> Wouldn't that be a "ppc64leabi32" target?
> > >>
> > >> There is no LE support for 32bit on PPC :).
> > > 
> > > ...in that case what is the original bug reporter actually
> > > asking for?
> > 
> > Not sure - the only thing that remotely resembles what he's describing
> > would be if you enable the 32bit personality:
> > 
> >   $ /uname-a.sh
> >   Linux tempranillo 3.17.0-rc1-2-default+ #1 SMP Thu Jan 1 01:04:00 CET
> >   1970 ppc64le ppc64le ppc64le GNU/Linux
> > 
> >   $ perl -e 'syscall(136, 8); exec "/uname-a.sh"'
> >   Linux tempranillo 3.17.0-rc1-2-default+ #1 SMP Thu Jan 1 01:04:00 CET
> >   1970 ppcle ppcle ppcle GNU/Linux
> > 
> > but that doesn't mean that there is an actual ABI change. The ABI is
> > still ppc64le, as all binaries are still running in 64bit mode.
> > 
> > So before we add anything to QEMU, I would like to first see someone
> > point me to kernel patches that actually enable LE on Linux for 32bit
> > user space. So far I'm not aware of any (and I doubt it makes much sense).
> 
> Thanks, guys, I'll ping him for that.

(Fwiw my reading of the original bug description is that in fact he
is working on the kernel patch in question)