Hi Ard, Rafael

On Thu, Feb 16, 2017 at 5:14 AM, Rafael J. Wysocki [off-list ref] wrote:

On Wednesday, February 15, 2017 05:44:28 PM Ard Biesheuvel wrote:

quoted

Hello Bhupesh,

On 15 February 2017 at 13:04, Bhupesh Sharma [off-list ref] wrote:

quoted

This patchset presents a RFC approach to move the ACPI BGRT table
handling related code out of 'arch/x86' and place it inside
'drivers/firmware/efi' so that it can be used to support
ACPI BGRT table and BGRT logo across x86 and AARCH64 ARCHs.

Note that this patchset is primarily tested on Qemu VMs for x86_64
and AARCH64 environments, while I try to get my hands on a real
AARCH64 hardware which supports a graphic card and whose boot firmware
I can modify to support the ACPI BGRT table to be passed to the kernel.

Here are some details/outputs about/from the test environment:

AARCH64:
========
1. I used the latest Qemu (master branch) and compiled it for aarch64 with softmmu support.
   Here is the command line I use to launch the same:

# qemu-system-aarch64 -cpu cortex-a57 -M virt -m 2048 -bios QEMU_EFI.fd -kernel Image -initrd rootfs.cpio -serial stdio --append "acpi=force" -device virtio-gpu-pci

where, QEMU_EFI.fd is the DEBUG build of EDK2 ArmVirtQemu.dsc (master branch)
       rootfs.cpio is the AARCH64 build of Buildroot (Little Endian AARCH64)
       -device virtio-gpu-pci provides the Graphical support required to have BGRT logo.


2. Here are some test results once Linux boots up, which show a valid BGRT table/logo being
   seen by the kernel:

# acpidump
...

BGRT @ 0x0000000000000000
  0000: 42 47 52 54 38 00 00 00 01 B4 49 4E 54 45 4C 20  BGRT8.....INTEL
  0010: 45 44 4B 32 20 20 20 20 02 00 00 00 20 20 20 20  EDK2    ....
  0020: 13 00 00 01 01 00 01 00 00 80 B0 BA 00 00 00 00  ................
  0030: 2F 01 00 00 0F 01 00 00                          /.......

# dmesg | grep -i BGRT
[    0.000000] ACPI: BGRT 0x00000000B8640000 000038 (v01 INTEL  EDK2     00000002      01000013)


# ls -lah /sys/firmware/acpi/bgrt/image
-r--r--r--    1 root     root       32.9K Feb 15 11:27 /sys/firmware/acpi/bgrt/image

x86:
====
1. I used the latest Qemu (master branch) and compiled it for x86.
   Here is the command line I use to launch the same:

# qemu-system-x86_64 -enable-kvm -sdl -m 2048 -vga std -boot c -cpu host -bios OVMF_CODE.fd -kernel bzImage -initrd rootfs.cpio -serial stdio

where, OVMF_CODE.fd is the RELEASE build of EDK2 OvmfPkg (master branch)
       rootfs.cpio is the x86_64 build of Buildroot

2. Similar test results are seen as noted for the AARCH64 case above, which show a valid BGRT table/logo being
   seen by the kernel.

Cc: Ard Biesheuvel <redacted>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Matt Fleming <redacted>
Cc: Dan Williams <redacted>
Cc: Dave Young <redacted>
Cc: Thomas Gleixner <redacted>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Will Deacon <redacted>

Bhupesh Sharma (2):
  x86/efi-bgrt: Move efi-bgrt handling out of arch/x86
  bgrt: Make ACPI BGRT parsing code common for ARCHs

 arch/arm64/kernel/acpi.c                               | 3 +++
 arch/x86/kernel/acpi/boot.c                            | 6 ------
 arch/x86/platform/efi/Makefile                         | 1 -
 drivers/acpi/Kconfig                                   | 2 +-
 drivers/acpi/bgrt.c                                    | 6 ++++++
 drivers/firmware/efi/Makefile                          | 1 +
 {arch/x86/platform => drivers/firmware}/efi/efi-bgrt.c | 0
 include/linux/efi-bgrt.h                               | 2 ++
 8 files changed, 13 insertions(+), 8 deletions(-)
 rename {arch/x86/platform => drivers/firmware}/efi/efi-bgrt.c (100%)

These patches look fine to me, and I think we should be able to merge
them without any changes.

Reviewed-by: Ard Biesheuvel <redacted>

You only failed to cc the ACPI maintainers, please add Len and Rafael next time.

Oops. Will take care in future.

No worries.

The changes are fine by me.

Thanks Rafael and Ard.

Regards,
Bhupesh