Get rid of old hybrid graphics driver and use the new one.
1. Disable IGD and PEG in early romstage.
The PEG port will get disabled on devices that do not have a
discrete GPU. The power savings are around ~1Watt.
The disabled IGD does no longer waste GFX stolen memory.
2. Get rid of PCI driver
The Nvidia GPU can be handled by the generic PCI driver and allows
to use the ACPI _ROM generator for Switchable graphics.
3. Settings are stored in devicetree.
One driver for all Lenovo hybrid graphics capable devices.
4. Add support for Thinker1 GPU power handling.
Only boards that do use reference design 2012 are known to be
supported. Needs test on boards that do you use reference design 2013.
Should reduce idle power consumption when using IGD by ~5Watt.
Tested on Lenovo T430 without DGPU. PEG port is disabled.
Needs test on all devices.
Change-Id: Ibf18b75e8afe2568de8498b39a608dac8db3ba73
Signed-off-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-on: https://review.coreboot.org/20794
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Introduce a chip_driver that uses devicetree instead of Kconfig.
The new driver has the following advantages:
* No more wasted IGD GFX stolen memory
* Can be used by T500 series
* Is even run on devices that do not have a dGPU installed
* Can disable unused PEG port on devices without dGPU (and save power)
* Use devicetree instead of Kconfig options
* Support for multiple hybrid GPIO active levels
* Support for backlight control GPIO
* Support for _ROM on Optimus capable devices
The driver is split into romstage part and ramstage part.
Every mainboard has to call the driver in romstage to get the requested
GPU state. The mainboard code then has to toggle GPU power or disable
the IGD or PEG port.
The ramstage part does handle the hygrid graphics GPIO, including
optional backlight mux GPIO. Every GPIO can have it's own active level,
as defined in devicetree. Devices are no longer disabled in ramstage.
The existing hybrid graphics driver does the same configuration and
should not interfere with this commit until it has been removed.
Change-Id: Ie467f9a18b35ab3b8a523dbf51c5575db5b374a5
Signed-off-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-on: https://review.coreboot.org/20792
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Setting up default BARs and DMI init code is done in northbridge
code.
Change-Id: I6cfa3018ca7f5ef351415c4ec6e178ade353f7a7
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/20906
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
This mainboard features is an G43 northbridge, ICH10 southbridge and
Winbond W83627dhg SuperI/O. This board is impossible to flash
internally with vendor bios (BIOS region is WP and other regions like
IFD and ME are read only and inaccessible respectively). Due to either
ICH10 or board layout it is also impossible to do ISP, which requires
desoldering flash chip. To make hacking more easy there is an empty
SPI header next to spi flash pads which can be hooked up to a SPI
flash.
What works:
* 2 DDR2 dimms per channel (tested with 1+2G in CH0 and 2+2G in CH1);
* SATA with AHCI
* Integrated GPU with option rom (extracted from a Gigabyte vendor
bios)
* VGA (on DVI) with NGI if patched to use DVI gmbus port for output
* PCI
* Reboot and S3 resume
* Descriptor mode with ME disable straps and ME region absent (no
working gbe in this configuration though)
* USB.
What does not work:
* GBE (probably requires working ME);
* Analog on DVI port out is shaking, which is not the case with vendor
BIOS (setting clockgen on smbus 0x69 like vendor fixes it).
* Booting with ME enabled (needs raminit patches for that)
Not tested:
* Sound;
* All the rest.
Not coreboot related problems:
* Flashing this board with vendor bios is a PITA and requires
desoldering flash chip;
* In situ programming is not possible.
TESTED with SeaBIOS and Linux 4.10.8
Change-Id: If27280feb7cbf0a88f19fe6a63b1f6dbcf9b60f4
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/19256
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Add Cannon lake, Apollo Lake and GLK CPU device IDs in
common Mp Init code.
BUG=none
BRANCH=none
TEST=Build and boot reef
Change-Id: I22694ced0cf900a55a28d1ecaa177cab2ea9a90c
Signed-off-by: Barnali Sarkar <barnali.sarkar@intel.com>
Reviewed-on: https://review.coreboot.org/20896
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
If all strings in SMBIOS table are empty, smbios_string_table_len
function should return 2, cause every table must end with "\0\0".
Also replace "eos" field type in smbios structures
from char to u8.
Change-Id: Ia3178b0030aa71e1ff11a3fd3d102942f0027eb1
Signed-off-by: Konstantin Aladyshev <aladyshev22@gmail.com>
Reviewed-on: https://review.coreboot.org/20840
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Change-Id: I730a8a150134cc1ef8fb3872728bb0586ac7b210
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/19732
Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Update the existing tianocore payload (which didn't do any more than
adding an elf payload with a specific name) to fetch and build the
UEFI corebootPayloadPackage, using the coreboot toolchain for
compilation. Only checkout the commit when changing commit IDs or if
version is master, instead of every time it builds.
Currently working if patches are merged into the upstream edk2
repository (to be included in a follow-on patch).
Change-Id: I0bf4cedec2d6821ae2a04184ebb5cf88979ccee3
Signed-off-by: Martin Roth <martinroth@google.com>
Signed-off-by: Philipp Deppenwiese <zaolin@das-labor.org>
Signed-off-by: Evelyn Huang <evhuang@google.com>
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/15057
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
- Check for original LinuxBIOS header from LANL
- Check for the BSD style license used in x86emu
- Add top level checks for src/cpu and src/vboot
- Add drivers dir check for aspeed, emulation, and sil
- Add mainboard dir check for digitallogic
Change-Id: Ie1ef6a39fd1d184815d22235d43780caedb2ca95
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/19122
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Chris Ching <chingcodes@google.com>
Reviewed-by: Werner Zeh <werner.zeh@siemens.com>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
libpayload needs a static copy of the out of line function
`font_glyph_filled()` in every TU that needs it. So make it static
inline.
This fixes a build error by gcc (Debian 7.1.0-12) 7.1.0 from Debian
Sid/unstable. This happens with any libpayload based payload like
coreinfo, nvramcui or tint.
```
[…]
LPCC build/coreinfo.elf (LINK)
/src/coreboot/payloads/coreinfo/build/libpayload/bin/../lib/libpayload.a(corebootfb.libc.o): In function `corebootfb_putchar':
/src/coreboot/payloads/libpayload/drivers/video/corebootfb.c:173: undefined reference to `font_glyph_filled'
[…]
```
Change-Id: I931f0f17b33abafdc49aa755a0dad65e28820750
Signed-off-by: Paul Menzel <pmenzel@molgen.mpg.de>
Reviewed-on: https://review.coreboot.org/20897
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Updating the common smihandler to handler gpi events which
originally were going to be left to each soc to handle. After
some more analysis the gpi handler can also be commonized.
Change-Id: I6273fe846587137938bbcffa3a92736b91982574
Signed-off-by: Brandon Breitenstein <brandon.breitenstein@intel.com>
Reviewed-on: https://review.coreboot.org/20917
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The device_t type is used in smihandler.h, but the header
is not including a header which provides that type. Fix
that by #includeing <device/device.h>
Change-Id: I89af949b088bf569d330f2ea74b1b8ae97e1ed52
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/20933
Reviewed-by: Furquan Shaikh <furquan@google.com>
Reviewed-by: Subrata Banik <subrata.banik@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Each line of spd config overwriting the slot 0 instead
of appending it.
Change-Id: I0124aa34f1d4fcb30810fb7eef03d4828a7ac430
Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
Tested-on: lenovo x1 carbon gen 1 with elpida 8GB
Fixes: e9787ff61f ("mb/lenovo/x1_carbon_gen1: Add 4GiB SPD index 2")
Reviewed-on: https://review.coreboot.org/20918
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Add smm common code for io trap handling. This feature is
mainly used by big core intel platforms. Commonizing io trap
handling will make bring up of big core platforms more
seamless.
Change-Id: I83bcf22107291ea181b347fac40d57d7ea138de1
Signed-off-by: Brandon Breitenstein <brandon.breitenstein@intel.com>
Reviewed-on: https://review.coreboot.org/20848
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The newline lint check takes forever as it checks every file
in the git repo -- not just the files changed. To add insult to
injury these checks are performed on every commit. Allow one to
skip this check by setting the LINT_SKIP_NEWLINE_CHECK environment
variable.
Change-Id: I1cd5c50627b73d68b0da2fcb2d9546eb874a7959
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/20903
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
This patch uses struct device explicitly for the ramstage functions
as that's the actual type it's working on. Additionally, the
declarations for types and functions are fully exposed so that
compliation units don't have to guard certain functions from use
because it's being compiled for multiple stages.
Change-Id: I8db23ed400a59073e1e66522d020a5928f71f3a6
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/20902
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
In _FSL ACPI notification 0x83 was incorrectly being sent to DPTF.
When there should be no notification on fan speed change.
Change-Id: I66efa7a7feb911a458829a54dbd0afefabd42394
Signed-off-by: Sumeet Pawnikar <sumeet.r.pawnikar@intel.com>
Reviewed-on: https://review.coreboot.org/20875
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
the __must_check function attribute is pretty much straight from the
linux kernel - used to encourage callers to consume function return
values.
Change-Id: I1812d957b745d6bebe2a8d34a9c4862316aa8530
Signed-off-by: Caveh Jalali <caveh@google.com>
Reviewed-on: https://review.coreboot.org/20881
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
With this patch apollolake uses the common PMC util
code.No regression observed on a APL platform.
Change-Id: I322a25a8b608d7fe98bec626c6696e723357a9d2
Signed-off-by: Shaunak Saha <shaunak.saha@intel.com>
Reviewed-on: https://review.coreboot.org/19375
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
PMC util code is very similar accross different intel SOC's.
This patch is an effort to move those code in common place
so that it can be shared accross different intel platforms
instead of duplicating for each platform. This patch adds
pmclib.c file which contains the pmc utility functions
common accross SOC's. The config for common PMC is
SOC_INTEL_COMMON_BLOCK_PMC which can be defined in SOC's
Kconfig file in order to use the common PMC util code.
Change-Id: Ic3d96fc23a98c30e8ea0969a7be09d217eeaa889
Signed-off-by: Shaunak Saha <shaunak.saha@intel.com>
Reviewed-on: https://review.coreboot.org/19349
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Internal PME is detected when bit PME_B0_STS is set. Following devices
causes internal PME.
- Integrated LAN
- HD Audio/Audio DSP
- SATA
- XHCI ('USB3')
- ME Maskable Host Wake
In SPT, PCIEXPWAK_STS bit isn't getting set due to known bug.
So scan all PCIe RP for PME status bit & update event log accordingly.
BUG=b:36992859
TEST=Build for Soraka, Verify resume due to PME on root port is logged
in elog.
Change-Id: I879a7c332e62ab598942b29d31bad84619b35ea7
Signed-off-by: Naresh G Solanki <naresh.solanki@intel.com>
Reviewed-on: https://review.coreboot.org/20532
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Ensure that the VLF bit will be cleared after a power loss event even if
the stopwatch has expired before the code to clear the bit is reached.
Change-Id: Ib2cfdabf4cd4df834395d6a102c6ae70568e71db
Signed-off-by: Werner Zeh <werner.zeh@siemens.com>
Reviewed-on: https://review.coreboot.org/20893
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Mario Scheithauer <mario.scheithauer@siemens.com>
The schematics isn't available for the board, but other L*00 series boards
seem to use a different, compared to T*00 series, GPIO layout.
As it has never been tested, remove the broken driver.
Change-Id: I4bfa02fdbc5da5b556010c2f300faaf6dc845b80
Signed-off-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-on: https://review.coreboot.org/20798
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Configuring the clockgen like vendor bios fixes the issue where the
display wobbles from left to right on the analog VGA output.
Note: This seems to be common issue/requirement on Intel devices from
that generation (also happens on dg43gt).
TESTED on Intel D510MO.
Change-Id: I08449f0b8b90a1781e6dec91140bf219ea07aaf9
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/19595
Reviewed-by: Sumeet R Pawnikar <sumeet.r.pawnikar@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Taken from crosstool-ng and and re-applied to 0.8.13
Change-Id: I43ad23b66ec10915cc68be8c395825f4f9161725
Signed-off-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-on: https://review.coreboot.org/20800
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Among other improvements, the new GDB has a new GDB/CLI
command to erase flash memory and a few Python scripting
enhancements.
Change-Id: Ie3852d5bc74617d3e05509e19bbd1caa281da3e0
Signed-off-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-on: https://review.coreboot.org/20799
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Martin Roth <martinroth@google.com>
The SPI controller initialization in finalize_chipset was failing
because FSP was setting FLOCKDN before finalize_chipset was called.
Hence move finalize_chipset to get called from BS_POST_DEVICE so that it is
called before FSP notify function-Ready To Boot state.
TEST: run flashrom with -VVV and observe supported opcodes and SPI
flash chip are reported correctly, and write/erase operations succeeed.
Original-Change-Id: I3c0297f3f2258cf77cf00db367f11ff4d1d9dc77
Original-Signed-off-by: Hannah Williams <hannah.williams@intel.com>
Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Change-Id: I690fb4bf9e78bb58811c704179ba8b8f25ce95cc
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/20891
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
When resuming from ACPI S3 suspend, EHCI controller
may be in D3 power-management state. Bring it to D0
early so it is functional for console.
NOTE: D3hot->D0 transition was observed to reset
previous programming of PCI_COMMAND register.
Change-Id: Id177ce61926beb057fe67ba42a306d8e565d2657
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/20827
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Always sanity check for EHCI class device and move
PCI function power enablement up.
Change-Id: I1eebe813fbb420738af2d572178213fc660f392a
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/20826
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
There is assumption of static EHCI_BAR_INDEX, try to
clean it up by bringing BAR programming at one spot.
Change-Id: Ie16090536ac5470c24720a54813015250ae2d0dd
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/20825
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Uses common hardware access functions to make smbus block read and
write available in romstage.
Those are needed to reconfigure the clockgen on smbus offset 0x69,
which is sometimes needed for things like CPU C-states or analog
display out to work properly.
Change-Id: I0a06178d2474ce65972de157cb437b42f3354da0
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/19594
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Damien Zammit <damien@zamaudio.com>
Reviewed-by: Sumeet R Pawnikar <sumeet.r.pawnikar@intel.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
All Intel southbridges implement the same SMBus functions.
This patch replaces all these similar and mostly identical
implementations with a common file.
This also makes i2c block read available to all those southbridges.
If the northbridge has to read a lot of SPD bytes sequentially, using
this function can reduce the time being spent to read SPD five-fold.
Change-Id: I93bb186e04e8c32dff04fc1abe4b5ecbc4c9c962
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/19258
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
This fixes indentation and whitespaces before opening parentheses.
Change-Id: I8940f712c0161419ee0c383b7bc9eb581967366e
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/20852
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Starting with RK3399, PMUGPIO pull registers use the same write mask
format as normal GRF registers, so they need to use RK_CLRSETBITS()
rather than clrsetbits_le32().
BRANCH=None
BUG=None
TEST=boot from scarlet
Change-Id: Ibe391273d58ab35df993e149187d67497fcf2acc
Signed-off-by: Lin Huang <hl@rock-chips.com>
Reviewed-on: https://review.coreboot.org/20871
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Julius Werner <jwerner@chromium.org>
some gpio irq need to set input pull initialization status
to guarantee to get the right irq trigger. let's add this argument
in gpio_input_irq() function
BRANCH=None
BUG=None
TEST=boot from bob
Change-Id: I9b8e6497f07146dafdb447a6ea10d039a2a2fa33
Signed-off-by: Lin Huang <hl@rock-chips.com>
Reviewed-on: https://review.coreboot.org/20866
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
If TTYS0_BAUD is not configured, then by default use baudrate of 115200.
BUG=b:64030366
Change-Id: Ida4c7ae77aba5dfd4ec331e22a54ce43a91bde00
Signed-off-by: Furquan Shaikh <furquan@chromium.org>
Reviewed-on: https://review.coreboot.org/20834
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Patrick Rudolph