Commit c37b0e3 [soc/intel/skylake: Generate ACPI DMAR table]
only generates DMAR tables for boards using FSP 2.0, which
leaves out Skylake Chromebooks, which use FSP 1.1.
Correct this omission by adding the same functionality for
FSP 1.1 boards.
Test: build/boot on U-series Skylake Chromebook, observe
IOMMU fully functional with intel_iommu=on kernel parameter.
Change-Id: I68837f58aac357fa3f58979fe92d8993fae58640
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/30230
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
FSP support two SATA modes as AHCI mode (0) and RAID mode (1), make it
more clear in header file.
Change-Id: I1edcadc0048df839da145260b60f9f7720d981fe
Signed-off-by: Lijian Zhao <lijian.zhao@intel.com>
Reviewed-on: https://review.coreboot.org/c/30093
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
CPU ratio will be fixed to non-turbo max value if CpuRatio UPD had been
set to zero.
BUG=N/A
TEST=Boot up into sarien system, cat /proc/cpuinfo and cpu frequency is
changing.
Signed-off-by: Lijian Zhao <lijian.zhao@intel.com>
Change-Id: I3e82293c8b6027ddf9a528d0654fe46f233dcb82
Reviewed-on: https://review.coreboot.org/c/30216
Reviewed-by: Nick Vaccaro <nvaccaro@google.com>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
RAM is reserved for Chromeos even when Chrome is not used.
Use CONFIG_CHROMEOS to determine if RAM must be reserved.
BUG=N/A
TEST=Intel CherryHill CRB
Change-Id: I3f55bf96ab2ec66cddbb54de03455a9bfd194682
Signed-off-by: Frans Hendriks <fhendriks@eltan.com>
Reviewed-on: https://review.coreboot.org/c/29332
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
commit msg copied from
commit id: 64c9f1584c
The GPIO drivers in Windows and Linux for the Icelake CPU
have a sparse GPIO map and do not allocate pins contiguously.
Each GPIO group is allocated as 32 pads regardless of whether
the hardware actually has that many in the group.
It appears this originated with a bug in Windows/UEFI and was
carried over to Linux in order to work with existing firmware:
https://lore.kernel.org/patchwork/patch/855244/
In order to support using ACPI GPIOs it is necessary for coreboot
to be compatible with this implementation. The GPIO groups that
are usable by the OS are declared with a pad base which is then
used to compute the number for ACPI GPIOs.
Change-Id: I94fafd8af13cf229f5c467de5179aed021465739
Signed-off-by: Subrata Banik <subrata.banik@intel.com>
Reviewed-on: https://review.coreboot.org/c/30276
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
The Intel GMA ACPI opregion address needs to be set on S3 resume,
otherwise the Windows display driver fails to re-initialize correctly.
Fix by ensuring the address is set correctly regardless of display
init type used (GOP or VBIOS).
Test: build/boot on google/edgar, ensure internal display functional
following S3 resume under Windows 10.
Change-Id: I471c44e8ba4514e4a2ddf6739109b759145598ed
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/30233
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Mirrors addition to Braswell SoC in commit d3d0f07.
Test: build/boot Windows 10 on Baytrail ChromeOS device, verify Windows shows
virtualization as enabled.
Change-Id: Ia1fafa73325814fed30b2ac91290b682dd8eab04
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/30228
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
It seems they are not included anywhere, Jenkins?
Change-Id: I629cdeb337fce381c69bd1ba0520e524ccdd90dd
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/c/26756
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
The GPIO pin map for CNL-H does not match with the OS expected
pin numbers. This has been updated to match what is used by the
Linux kernel pinctrl driver and the pad base has been set for
the GPIO groups to match the sparse GPIO map used by the kernel.
I do not have CNL-H hardware to test this so it is verified against
the kernel driver at drivers/pinctrl/intel/pinctrl-cannonlake.c
Change-Id: Ife7d3090d654b0b88c6911befa08bf6abd4f2ff9
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: https://review.coreboot.org/c/30134
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The GPIO drivers in Windows and Linux for the Cannonlake CPU
have a sparse GPIO map and do not allocate pins contiguously.
Each GPIO group is allocated as 32 pads regardless of whether
the hardware actually has that many in the group.
It appears this originated with a bug in Windows/UEFI and was
carried over to Linux in order to work with existing firmware:
https://lore.kernel.org/patchwork/patch/855244/
In order to support using ACPI GPIOs it is necessary for coreboot
to be compatible with this implementation. The GPIO groups that
are usable by the OS are declared with a pad base which is then
used to compute the number for ACPI GPIOs.
BUG=b:120686247
TEST=tested with write protect GPIO on sarien board. Before this
change the ACPI pin number was 220 which did not correspond to the
pin number in Linux. After this change the ACPI number is 303,
which maps to the correct GPIO in Linux. Now the GPIO value reported
by the kernel changes when the WP pin is toggled in hardware.
Change-Id: I4f1a9e118d7e48f2445ccbb62a12a22e9a832c51
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: https://review.coreboot.org/c/30133
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
In some situations the GPIO pad numbers used by the OS are not
contiguous and coreboot must provide a way for ACPI to provide
the expected GPIO number to the OS.
To do this each GPIO group can now have a pad base value, which
will be used as the starting pin number for this group and it
is added to the relative pin number of this GPIO to compute the
ACPI pin number for a particular GPIO.
By default this change has no effect because the existing uses
of INTEL_GPP() will set the pad base to PAD_BASE_NONE and the
GPIO number is used as the ACPI pin number without translation.
BUG=b:120686247
TEST=tested on a sarien(cannonlake) board
Change-Id: I25f73df45ffae18c5721a00ca230a6b07c250bab
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: https://review.coreboot.org/c/30131
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This patch introduces 3 helper function for cpuid(1) :
1. cpu_get_cpuid() -> to get processor id (from cpuid.eax)
2. cpu_get_feature_flags_ecx -> to get processor feature flag (from cpuid.ecx)
3. cpu_get_feature_flags_edx -> to get processor feature flag (from cpuid.edx)
Above 3 helper functions are targeted to replace majority of cpuid(1)
references.
Change-Id: Ib96a7c79dadb1feff0b8d58aa408b355fbb3bc50
Signed-off-by: Subrata Banik <subrata.banik@intel.com>
Reviewed-on: https://review.coreboot.org/c/30123
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
The kernel GPIO driver only expects some GPIO communities to be exported
in the _CRS and it will not work correctly if the other communities are
exported.
CNL-LP: GPIO communities 0, 1, 4
CNL-H: GPIO communities 0, 1, 3, 4
Additionally one of the pin offset values was incorrect in GPIO
community 1 for CNL-LP. This doesn't have any specific failure mode but
it was found when auditing the GPIO code.
Details of the kernel expected map can be found in the linux kernel at
drivers/pinctrl/intel/pinctrl-cannonlake.c
BUG=b:120686247
TEST=check /sys/kernel/debug/pinctrl/INT34BB:00/pins to ensure that
pins >= 198 are not reading all zeros for the pin config registers.
Change-Id: Ie1a2f3b9f9f4b24a9fc57e468dee50e99753912f
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: https://review.coreboot.org/c/30112
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
Recently there has been a change to print ME version. But the stage at
which the version is printed causes the HECI device to remain in D0 state.
This in turn prevents the SoC from entering S0ix state.
This change moves printing ME version a little earlier so that the HECI
device is put into D0i3 state by FSP and the SoC can enter S0ix state
successfully.
BRANCH=octopus
BUG=b:120571529
TEST=Ensure that the ME version gets printed in BIOS logs. Ensure that
the device boots to ChromeOS. Ensure that the device enters S0ix
successfully(using suspend_stress_test -c 25).
Change-Id: I85bc45003a040c8347f929457792d78a9a077c6c
Signed-off-by: Karthikeyan Ramasubramanian <kramasub@google.com>
Reviewed-on: https://review.coreboot.org/c/30074
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
The input clock for the I2C controllers was set at 133MHz but should
really be 216MHz according to the kernel:
https://patchwork.kernel.org/patch/10408729/
"Intel Cannon Lake PCH has much higher 216 MHz input clock to LPSS I2C
than Sunrisepoint which uses 120 MHz. Preliminary information was that
both share the same clock rate but actual silicon implements elevated
rate for better support for 3.4 MHz high-speed I2C."
This change was tested on a sarien board where an I2C trackpad that was
measuring ~700MHz on I2C and is now measuring ~380MHz.
Change-Id: I792d1f013da5538a2b8157e2f99b754ca7b6bf70
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: https://review.coreboot.org/c/30061
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
Reviewed-by: Lijian Zhao <lijian.zhao@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
As part of the memory mapped BIOS region is covered by SRAM, check
that CBFS always fits the effectively mapped region of flash. This
is usually taken care of by reserving the SRAM range in the FMAP
(e.g. as BIOS_UNUSABLE), but can be missed.
Change-Id: If5a5b553ad4853723bf13349c809c4f6154aa5f2
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: https://review.coreboot.org/c/30055
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Use common code to tear down CAR.
Change-Id: I62a70ae35fe92808f180f2b5f21c5899a96c2c16
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/29930
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
This patch does the following:
- improve the style by removing tabs in front of jmp addresses
- Make the code for zeroing variable MTRR more readable (copied from
cpu/intel/car)
- Fetch PHYSMASK high from cpuid instead of Kconfig
Change-Id: I6ba67bb8b049c3f25b856f6ebb1399d275764f54
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/29928
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
This does the following:
- Reuse the cpu/intel/car/non-evict CAR setup and exit.
- Use postcar_frame functions to set up the postcar frame
Change-Id: I428832a2d7e46ce61a7f9bd498b609feb4518eb0
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/29927
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Instead of SMMSTORE_APM_CNT use APM_CNT_SMMSTORE and define it in
cpu/x86/smm.h
Change-Id: Iabc0c9662284ed3ac2933001e64524011a5bf420
Signed-off-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-on: https://review.coreboot.org/c/30023
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Instead of ELOG_GSMI_APM_CNT use APM_CNT_ELOG_GSMI and define it in
cpu/x86/smm.h
Change-Id: I3a3e2f823c91b475d1e15b8c20e9cf5f3fd9de83
Signed-off-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-on: https://review.coreboot.org/c/30022
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Cache BIOS region can boost boot performance, however it can't be over
16MB, according to processor EDS vol1(Apollolake/Skylake/WhiskeyLake),
FLASH+APIC LT will be less than 20MB under 4G. Set the maxiam to 16GB
to save numbers of mtrr entries.
BUG=b:119267832
TEST=Build and boot up fine on whiskeylake rvp platform.
Change-Id: I46a47c8bf66b14fb2fcb7b6b1d30d02886c450a4
Signed-off-by: Lijian Zhao <lijian.zhao@intel.com>
Reviewed-on: https://review.coreboot.org/c/29944
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
Add the ACPI device names for the USB ports to match what
is in the DSDT so USB ports can be defined in the SSDT.
Change-Id: Ibb323bbd324811fa3178b0cba3d7f0a315169486
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: https://review.coreboot.org/c/29997
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
Increase the bootblock size to 48K to match skylake. With UART
enabled we are very near the 32K limit, and with upcoming changes
to add USB devices in devicetree for a cannonlake board it is over
the current 32K limit.
Change-Id: I155cb0a6af1746af6833fa9f35c2ea6fe0bc709f
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: https://review.coreboot.org/c/29996
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
Define the constants that DPTF expects from the SOC in order to
use the common DPTF ACPI code. For cannonlake this indicates
the CPU device is called B0D4 and is at PCI address 00:04.0.
Change-Id: I43c2f8dd7281d3e9f791ab01478ee7823fd6b128
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: https://review.coreboot.org/c/29759
Reviewed-by: Furquan Shaikh <furquan@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add a method to convert from 1/10 Kelvin to Celsius. This
is useful for EC devices where the sensor temperature are
stored in Celsius instead of Kelvin.
Change-Id: I6b1154f5ba13416131a029966d6d5c1598904881
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: https://review.coreboot.org/c/29758
Reviewed-by: Furquan Shaikh <furquan@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
In order to support using the common ACPI code on more platforms
than just Apollo Lake the DPTF code needs to be told what the
PCI address is for the CPU thermal device.
Change-Id: I638f2387330bbc42f64eb0fb676ee32c5df6572e
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: https://review.coreboot.org/c/29757
Reviewed-by: Furquan Shaikh <furquan@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Backlight control of internal panels likely won't work as configuration
for that seems absent in coreboot. Also, libgfxinit doesn't support any
MIPI/DSI connections, yet, and neither Gemini Lake.
TEST=Booted work-in-progress port kontron/mal10 with VGA text and
linear framebuffer modes. DP display came up.
Change-Id: I7b111f1cdac4d18f2fc3089f57aebf3ad1739e5d
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: https://review.coreboot.org/c/29903
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This change adds support to print ME version if UART_DEBUG is
enabled. Check for UART_DEBUG is necessary because talking to ME to
get the firmware version adds ~0.6 seconds to boot time.
TEST=Verified on octopus that ME version printed is correct.
Change-Id: I41217371558da1af694a2705e005429155d62838
Signed-off-by: Furquan Shaikh <furquan@google.com>
Reviewed-on: https://review.coreboot.org/c/29989
Reviewed-by: Justin TerAvest <teravest@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Sync the FSP settings with what coreboot does. Why both FSP and coreboot
configure this redundantly stays a secret.
TEST=Set SERIRQ_CONTINUOUS on kontron/mal10. A CPLD connected to LPC
works correctly now, but was confused by the wrong settings before
because the FSP defaults allowed to disable the LPC clock.
Change-Id: Id1c7180f460678bf0f9458228591050dd628c052
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: https://review.coreboot.org/c/29901
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Werner Zeh <werner.zeh@siemens.com>
Use acpigen_write_processor_cnot to implement notifications to the CPU.
Change-Id: I01e4397b9a1c15eff4b856cbc697fa2b4bc9761f
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/29893
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Use acpigen_write_processor_cnot to implement notifications to the CPU.
Change-Id: If482c64e7133cc6d82472d121ac138fc1b60a183
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/29892
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-by: David Guckian
Use acpigen_write_processor_cnot to implement notifications to the CPU.
Change-Id: Id64f9857bbd7db520c94de949db8f823f71d6dae
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/29891
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Use acpigen_write_processor_cnot to implement notifications to the CPU.
Automatically generate \PPKG in SSDT.
Change-Id: I79d2eed9b89b420554ce10d1fc0f151b1872afe2
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/29890
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Use acpigen_write_processor_cnot to implement notifications to the CPU.
Change-Id: I93c11e89da34c5432c6ce0415998b47bad339763
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/29889
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Use acpigen_write_processor_cnot to implement notifications to the CPU.
Change-Id: Ifb5a5c1255f9a922063293bf430e849909468eaf
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/29888
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Use acpigen_write_processor_cnot to implement notifications to the CPU.
Change-Id: I182585fd09e4ce848c860d00eb612e8f5fdde35e
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/29884
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
