For the Aarch64, there is one machine 'virt', it is primarily meant to
run on KVM and execute virtualization workloads, but we need an
environment as faithful as possible to physical hardware,  to support
firmware and OS development for pysical Aarch64 machines.

This machine comes with:
 - Re-designed memory map.
 - CPU cortex-a57.
 - EL2 and EL3 enabled.
 - GIC version 3.
 - System bus AHCI controller.
 - System bus XHCI controller.
 - CDROM and hard disc on AHCI bus.
 - E1000E ethernet card on PCIE bus.
 - VGA display adaptor on PCIE bus.
 - Only minimal device tree nodes.
And without:
 - virtio deivces.
 - fw_cfg device.
 - ACPI tables.

Arm Trusted Firmware and UEFI porting to this are done accordingly, and
it should supply ACPI tables to load OS, the minimal device tree nodes
supplied from this platform are only to pass the dynamic info reflecting
command line input to firmware, not for loading OS.

v7 changes:
 - edit memory map for PCIE slightly
 - add another secure UART which can be used for RAS and MM from EL0.

v6 changes:
 - rebased to the latest QEMU tree
 - rechecked all the header files included
 - added the newly introduced system bus EHCI controller
 - removed the machine_done callback due to commit 5614ca80
 - updated block comments styles according to checkpatch.pl
 - use Kconfig to add new file
 - use private SBSA* types defination instead of VIRT* in virt.h
   since nobody else using them so they are in the .c file instead
   of a new .h file

v5 changes:
 - removed more lines derived from virt.c
 - designed a new memory map
 - splitted former one patch into two for easier review
 - cancled previous EHCI and new HXCI coming later separately

V4 changes:
 - rebased to v3.0.0
 - removed timer, uart, rtc, *hci device tree nodes
   (others were removerd in v3)
 - other minore codes clean up, mainly unsed header files, comments etc.

V3 changes:
 - rename the platform 'sbsa-ref'
 - move all the codes to a separate file sbsa-ref.c
 - remove paravirtualized fw_cfg device
 - do not supply ACPI tables, since firmware will do it
 - supply only necessary DT nodes
 - and other minor code clean up

Hongbo Zhang (2):
  hw/arm: Add arm SBSA reference machine, skeleton part
  hw/arm: Add arm SBSA reference machine, devices part

 default-configs/arm-softmmu.mak |   1 +
 hw/arm/Kconfig                  |   3 +
 hw/arm/Makefile.objs            |   1 +
 hw/arm/sbsa-ref.c               | 757 ++++++++++++++++++++++++++++++++++++++++
 4 files changed, 762 insertions(+)
 create mode 100644 hw/arm/sbsa-ref.c

-- 
2.7.4

On Thu, 18 Apr 2019 at 05:05, Hongbo Zhang <hongbo.zhang@linaro.org> wrote:
>
> For the Aarch64, there is one machine 'virt', it is primarily meant to
> run on KVM and execute virtualization workloads, but we need an
> environment as faithful as possible to physical hardware,  to support
> firmware and OS development for pysical Aarch64 machines.
>
> This machine comes with:
>  - Re-designed memory map.
>  - CPU cortex-a57.
>  - EL2 and EL3 enabled.
>  - GIC version 3.
>  - System bus AHCI controller.
>  - System bus XHCI controller.
>  - CDROM and hard disc on AHCI bus.
>  - E1000E ethernet card on PCIE bus.
>  - VGA display adaptor on PCIE bus.
>  - Only minimal device tree nodes.
> And without:
>  - virtio deivces.
>  - fw_cfg device.
>  - ACPI tables.

I've had a look through these patches -- my comments are all
pretty minor. There are a few places where I don't really know
the right answer and have asked other people to weigh in on
whether changes (proposed or recent) for the virt board should
be applied here too.

thanks
-- PMM