Can someone provide instructions and the necessary environment/toolchain support needed to get Xen to build and boot on a native EFI system (i.e. with no CSM/legacy support)? Joe
>>> On 01.03.12 at 21:40, "Cihula, Joseph" <joseph.cihula@intel.com> wrote: > Can someone provide instructions and the necessary environment/toolchain > support needed to get Xen to build and boot on a native EFI system (i.e. with > no CSM/legacy support)?gcc must be 4.5.x or newer (I''d recommend not trying anything below 4.6.x, as I''m unaware of anyone ever having tried that, and I can''t exclude that there are bugs in the relevant code that was newly added in 4.5.0). ld must support linking native 64-bit EFI applications, which so far no released binutils version does. The single relevant change that''s necessary for this to work in 2.21 can be taken from http://www.sourceware.org/ml/binutils/2011-05/msg00131.html (but there''s at least one other non-essential adjustment in case you want to do anything with the generated binary that involves its symbol table). Next, a suitably enabled Dom0 kernel is needed. Finally, a config file needs to be placed alongside the xen*.efi binary; I''m attaching a simplistic example. Jan _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
On 02/03/2012 08:06, "Jan Beulich" <JBeulich@suse.com> wrote:>>>> On 01.03.12 at 21:40, "Cihula, Joseph" <joseph.cihula@intel.com> wrote: >> Can someone provide instructions and the necessary environment/toolchain >> support needed to get Xen to build and boot on a native EFI system (i.e. with >> no CSM/legacy support)? > > gcc must be 4.5.x or newer (I''d recommend not trying anything below > 4.6.x, as I''m unaware of anyone ever having tried that, and I can''t > exclude that there are bugs in the relevant code that was newly added > in 4.5.0). > > ld must support linking native 64-bit EFI applications, which so far no > released binutils version does. The single relevant change that''s > necessary for this to work in 2.21 can be taken from > http://www.sourceware.org/ml/binutils/2011-05/msg00131.html (but > there''s at least one other non-essential adjustment in case you want > to do anything with the generated binary that involves its symbol > table).Has that patch been picked up by the maintainers? -- Keir> Next, a suitably enabled Dom0 kernel is needed. > > Finally, a config file needs to be placed alongside the xen*.efi > binary; I''m attaching a simplistic example. > > Jan >
>>> On 02.03.12 at 09:22, Keir Fraser <keir.xen@gmail.com> wrote: > On 02/03/2012 08:06, "Jan Beulich" <JBeulich@suse.com> wrote: > >>>>> On 01.03.12 at 21:40, "Cihula, Joseph" <joseph.cihula@intel.com> wrote: >>> Can someone provide instructions and the necessary environment/toolchain >>> support needed to get Xen to build and boot on a native EFI system (i.e. > with >>> no CSM/legacy support)? >> >> gcc must be 4.5.x or newer (I''d recommend not trying anything below >> 4.6.x, as I''m unaware of anyone ever having tried that, and I can''t >> exclude that there are bugs in the relevant code that was newly added >> in 4.5.0). >> >> ld must support linking native 64-bit EFI applications, which so far no >> released binutils version does. The single relevant change that''s >> necessary for this to work in 2.21 can be taken from >> http://www.sourceware.org/ml/binutils/2011-05/msg00131.html (but >> there''s at least one other non-essential adjustment in case you want >> to do anything with the generated binary that involves its symbol >> table). > > Has that patch been picked up by the maintainers?Oh, yes, I committed it a few days after sending (the mail thread above even includes the approval). Jan
>>> On 02.03.12 at 09:22, Keir Fraser <keir.xen@gmail.com> wrote: > On 02/03/2012 08:06, "Jan Beulich" <JBeulich@suse.com> wrote: > >>>>> On 01.03.12 at 21:40, "Cihula, Joseph" <joseph.cihula@intel.com> wrote: >>> Can someone provide instructions and the necessary environment/toolchain >>> support needed to get Xen to build and boot on a native EFI system (i.e. > with >>> no CSM/legacy support)? >> >> gcc must be 4.5.x or newer (I''d recommend not trying anything below >> 4.6.x, as I''m unaware of anyone ever having tried that, and I can''t >> exclude that there are bugs in the relevant code that was newly added >> in 4.5.0). >> >> ld must support linking native 64-bit EFI applications, which so far no >> released binutils version does. The single relevant change that''s >> necessary for this to work in 2.21 can be taken from >> http://www.sourceware.org/ml/binutils/2011-05/msg00131.html (but >> there''s at least one other non-essential adjustment in case you want >> to do anything with the generated binary that involves its symbol >> table). > > Has that patch been picked up by the maintainers?Wait - I was actually wrong: The patch pre-dates 2.22, and hence no change is needed when using this (newest) binutils version. Jan
On 02/03/2012 08:48, "Jan Beulich" <JBeulich@suse.com> wrote:>>>> On 02.03.12 at 09:22, Keir Fraser <keir.xen@gmail.com> wrote: >> On 02/03/2012 08:06, "Jan Beulich" <JBeulich@suse.com> wrote: >> >>>>>> On 01.03.12 at 21:40, "Cihula, Joseph" <joseph.cihula@intel.com> wrote: >>>> Can someone provide instructions and the necessary environment/toolchain >>>> support needed to get Xen to build and boot on a native EFI system (i.e. >> with >>>> no CSM/legacy support)? >>> >>> gcc must be 4.5.x or newer (I''d recommend not trying anything below >>> 4.6.x, as I''m unaware of anyone ever having tried that, and I can''t >>> exclude that there are bugs in the relevant code that was newly added >>> in 4.5.0). >>> >>> ld must support linking native 64-bit EFI applications, which so far no >>> released binutils version does. The single relevant change that''s >>> necessary for this to work in 2.21 can be taken from >>> http://www.sourceware.org/ml/binutils/2011-05/msg00131.html (but >>> there''s at least one other non-essential adjustment in case you want >>> to do anything with the generated binary that involves its symbol >>> table). >> >> Has that patch been picked up by the maintainers? > > Oh, yes, I committed it a few days after sending (the mail thread above > even includes the approval).Ah, so I see. I didn''t know you had commit rights to binutils. :-) Looks like a binutils release must be about due, and that would include it? -- Keir> Jan >
>>> On 02.03.12 at 09:57, Keir Fraser <keir.xen@gmail.com> wrote: > On 02/03/2012 08:48, "Jan Beulich" <JBeulich@suse.com> wrote: > >>>>> On 02.03.12 at 09:22, Keir Fraser <keir.xen@gmail.com> wrote: >>> On 02/03/2012 08:06, "Jan Beulich" <JBeulich@suse.com> wrote: >>> >>>>>>> On 01.03.12 at 21:40, "Cihula, Joseph" <joseph.cihula@intel.com> wrote: >>>>> Can someone provide instructions and the necessary environment/toolchain >>>>> support needed to get Xen to build and boot on a native EFI system (i.e. >>> with >>>>> no CSM/legacy support)? >>>> >>>> gcc must be 4.5.x or newer (I''d recommend not trying anything below >>>> 4.6.x, as I''m unaware of anyone ever having tried that, and I can''t >>>> exclude that there are bugs in the relevant code that was newly added >>>> in 4.5.0). >>>> >>>> ld must support linking native 64-bit EFI applications, which so far no >>>> released binutils version does. The single relevant change that''s >>>> necessary for this to work in 2.21 can be taken from >>>> http://www.sourceware.org/ml/binutils/2011-05/msg00131.html (but >>>> there''s at least one other non-essential adjustment in case you want >>>> to do anything with the generated binary that involves its symbol >>>> table). >>> >>> Has that patch been picked up by the maintainers? >> >> Oh, yes, I committed it a few days after sending (the mail thread above >> even includes the approval). > > Ah, so I see. I didn''t know you had commit rights to binutils. :-) > > Looks like a binutils release must be about due, and that would include it?See my later second reply - I forgot that there already was one. Jan
On 02/03/2012 09:06, "Jan Beulich" <JBeulich@suse.com> wrote:>>>> Has that patch been picked up by the maintainers? >>> >>> Oh, yes, I committed it a few days after sending (the mail thread above >>> even includes the approval). >> >> Ah, so I see. I didn''t know you had commit rights to binutils. :-) >> >> Looks like a binutils release must be about due, and that would include it? > > See my later second reply - I forgot that there already was one.Ah, I did wonder. That''s good news. -- Keir
Possibly Parallel Threads
- [PATCH] VT-d: make remap_entry_to_msi_msg() return consistent message
- [PATCH 0/6] x86/HVM: miscellaneous RTC emulation adjustments
- [PATCH] x86: add hypercall to query current underlying pCPU''s frequency
- [PATCH] x86/apic: remove DMI checks in bigsmp driver for obsolete systems
- fsincos emulation on AMD CPUs