Re: [Xen-devel] [PATCH v2 0/5] xen: Break multiboot (v1) dependency and add multiboot2 support

[Roy Franz <roy.franz@xxxxxxxxxx>]

On Wed, Sep 24, 2014 at 10:19 AM, Daniel Kiper <daniel.kiper@xxxxxxxxxx> wrote:
> Hi,
>
> This patch series breaks multiboot (v1) protocol dependency and adds
> multiboot2 support. It lays down the foundation for EFI + GRUB2 + Xen
> development. Detailed description of ideas and thoughts you will
> find in commit message for every patch. If something is not obvious
> please drop me a line.
>
> Most of the requested things are fixed but there are still some minor
> outstanding issues (multiboot2 tags generation, excessive amount of casts
> in xen/arch/x86/boot/reloc.c, etc.; please check commit messages for
> more details). If something is not fixed yet it means that I do not have
> good idea how to do that. In case you spot something which was mentioned
> during previous review and still think that your comment is valid
> in particular case please notify me.
>
> Below you can find reply for Konrad's questions in regards to
> exception request for Xen 4.5 release.
>
>> Couple of questions:
>>
>>  - Since this is mostly XBI code, is there a lot of overlap with the ARM/x86
>>    refactoring of the EFI code? As in, will it require a lot of
>>    rebasing/fixing it up?
>
> I did not checked that right now. However, Roy once told me that this should
> not be very big issue because overlap is not so big. Please check this email
> for more details: 
> http://lists.xen.org/archives/html/xen-devel/2014-08/msg01173.html
>
> Sadly, it was sent more then one month ago so maybe something has changed.
> Roy, any comments on that?
There aren't fundamental conceptual conflicts, but there will be quite
a bit of refactoring/rebasing to do.  Most of my patchset is
re-organizing
x86/efi/boot.c, so rebasing it on this patchset would be a significant
amount of work.   My guess is that the first 10 of the 13 patches
in my series would need be be completely redone, as the code they are
rearranging has been changed by your series.
I think that it would be much less work for you to re-base your series
on mine, than the other way.  The x86/efi/boot.c changes
are a relatively small part of your series, but are the bulk of mine.

I should have another version of my patchset out today.


>
> Anyway, I am more than happy to work on this issue with Roy.
>
>>  - What is the disadvantage of having this in Xen 4.6?
>
> Well, it will require to rebase all this work on latest developments.
> It could be quite difficult because my patches touches a lot of places
> and as I can see there is a chance for substantial changes in some of them.
>
>>  - What is the advantage of having this in Xen 4.5?
>>    [My understanding is that it would allow us to thrash out a lot of
>>     the early bootup issues _now_ and fix them up instead of
>>     having to deal in Xen 4.6 with this, EFI, and SecureBoot - which
>>     is a lot of more moving pieces]
>
> More or less. Additionally, we will have support for multiboot2 protocol
> in place. There is not very big demand for that but as I know at least Intel
> guys were looking for that. IIRC, they needed that for TPM support.
>
> Daniel
>
>  xen/arch/x86/Makefile             |    1 +
>  xen/arch/x86/boot/cmdline.S       |    9 +-
>  xen/arch/x86/boot/head.S          |  150 ++++++++++++++++++++++++++++-----
>  xen/arch/x86/boot/reloc.c         |  248 
> ++++++++++++++++++++++++++++++++++++++++++++-----------
>  xen/arch/x86/boot/x86_64.S        |   10 ++-
>  xen/arch/x86/boot_info.c          |  258 
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>  xen/arch/x86/dmi_scan.c           |    7 +-
>  xen/arch/x86/domain_build.c       |   24 +++---
>  xen/arch/x86/e820.c               |    8 +-
>  xen/arch/x86/efi/boot.c           |  214 
> +++++++++++++++++++++++------------------------
>  xen/arch/x86/efi/efi.h            |    3 -
>  xen/arch/x86/efi/runtime.c        |   52 +++++++++---
>  xen/arch/x86/microcode.c          |   39 ++++-----
>  xen/arch/x86/mpparse.c            |    9 +-
>  xen/arch/x86/platform_hypercall.c |   19 ++---
>  xen/arch/x86/setup.c              |  370 
> +++++++++++++++++++++++++++-------------------------------------------------------
>  xen/arch/x86/x86_64/asm-offsets.c |    5 +-
>  xen/drivers/acpi/osl.c            |    9 +-
>  xen/drivers/video/vesa.c          |    7 +-
>  xen/drivers/video/vga.c           |   18 ++--
>  xen/include/asm-x86/boot_info.h   |  126 ++++++++++++++++++++++++++++
>  xen/include/asm-x86/config.h      |    2 -
>  xen/include/asm-x86/e820.h        |   10 +--
>  xen/include/asm-x86/edd.h         |    6 --
>  xen/include/asm-x86/mbd.h         |   70 ++++++++++++++++
>  xen/include/asm-x86/setup.h       |   10 +--
>  xen/include/xen/efi.h             |   10 ---
>  xen/include/xen/multiboot2.h      |  354 
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>  xen/include/xen/vga.h             |   18 ----
>  xen/include/xsm/xsm.h             |   14 ++--
>  xen/xsm/xsm_core.c                |    6 +-
>  xen/xsm/xsm_policy.c              |   14 ++--
>  32 files changed, 1510 insertions(+), 590 deletions(-)
>
> Daniel Kiper (5):
>       xen/x86: Introduce MultiBoot Data (MBD) type
>       xen/x86: Define e820 entries counter as unsigned int
>       xen/x86: Migrate to boot_info structure
>       xen/x86: Use constant as multiboot protocol identifier
>       xen/x86: Add multiboot2 protocol support
>

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