Starting vboot earlier in bootblock instead of romstage is usually
preferred (smaller root of trust, among other things). Therefore
unselect VBOOT_STARTS_IN_ROMSTAGE for broadwell. Also remove the unused
BROADWELL_VBOOT_IN_BOOTBLOCK option.
Change-Id: If8feea403ee4cd3a16ed8cb0faf9f4ccb34feaaf
Signed-off-by: Yu-Ping Wu <yupingso@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66779
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
xhci.asl places the SS ports at 11-14, following HS ports 1-10. However,
for Nissa, the kernel detects 12 HS ports 1-12 and 4 SS ports at 13-16,
resulting in the PLD intended for SS ports 1 and 2 being associated with
HS ports 11 and 12.
Changing the asl for SS to 13-16 makes locations associate correctly and
peering work.
BUG=b:234544025
BRANCH=firmware-brya-14505.B
TEST=manually verified on Nissa and Brya devices
Change-Id: I57aef771a7ff086b71a9e90b81e1a3635f832b2f
Signed-off-by: Adam Mills <adamjmills@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66590
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Reviewed-by: Sam McNally <sammc@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This patch drops explicit usage of the address-of operator ('&') while
passing the function pointer (argument 0) to the
`mp_run_on_all_cpus` API.
Note: It's just cosmetic change without any real difference in the operation.
TEST=Able to build and boot Google/Kano where CPU feature programming
is successful on all logical processors.
Signed-off-by: Subrata Banik <subratabanik@google.com>
Change-Id: I2c77959a76d2240ad1bfb7a9d7b9db7e8aee42f7
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66685
Reviewed-by: Lean Sheng Tan <sheng.tan@9elements.com>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Eric Lai <eric_lai@quanta.corp-partner.google.com>
Introduce the `BROADWELL_LPDDR3` Kconfig option along with some wrapper
code to allow mainboards using LPDDR3 DRAM to supply the DQ/DQS maps to
chipset code without having to use `pei_data`. The only mainboard using
LPDDR3 is Google Samus.
Change-Id: I0aaf0ace243c03600430c2a7ab6389a7b20cb432
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/55812
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Mainboards do not need to know about `pei_data` to tell northbridge code
where to find the SPD data. As done on Haswell, add the `mb_get_spd_map`
function and the `struct spd_info` type to retrieve SPD information from
mainboard code without having to use `pei_data` in said mainboard code.
Unlike Haswell MRC, Broadwell MRC uses all positions of the `spd_data`
array, not just the first. The placeholder SPD address for memory-down
seems to be different as well. Adapt the existing code to handle these
variations. Once complete, the abstraction layer for both MRC binaries
will have the same API.
Change-Id: I92a05003a319c354675368cae8e34980bd2f9e10
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/55811
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
There's no generic way to tell whether a mainboard has an EC or not.
Making Kconfig symbols for these options seems overkill, too. So, just
put them on the devicetree. Also, drop unnecessary assignments when the
board's current value is zero, as the struct defaults to zero already.
Change-Id: I8d3b352333bea7ea6f7b0f96d73e6c2d7d1a2cfb
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/55809
Reviewed-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Introduce the `SPD_MEMORY_DOWN` macro to indicate that a slot is used
with memory-down. This enables computing the channel disable masks as
the bits for slots where the SPD address is zero. To preserve current
behavior, zero the SPD addresses for memory-down slots afterwards.
Change-Id: I75b7be7c72062d1a26cfc7b09b79de62de0a9cea
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/55807
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
These files contain no creative content, and therefore have no
copyright. This effectively means that they are in the public
domain.
This commit updates the unlicensable empty (and effectively empty)
files with the CC-PDDX identifier for license compliance scanning.
Signed-off-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Change-Id: I0b76921a32e482b6aed154dddaba368f29ac2207
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66497
Reviewed-by: Michael Niewöhner <foss@mniewoehner.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This reverts commit 844dcb3725.
A power and performance analysis performed on Alder Lake demonstrated
that with an EPP (Energy Performance Preference) at 50% along with
EET (Energy Efficient Turbo) disabled, the overall SoC performance are
similar or better and the SoC uses less power.
For instance some browser benchmark results improved by 2% and some
multi-core tests by 4% while at the same time power consumption
lowered by approximately 7.6%.
BRANCH=firmware-brya-14505.B
BUG=b:240669428
TEST=verify that ETT is disabled
`iotools rdmsr 0 0x1fc'
Signed-off-by: Jeremy Compostella <jeremy.compostella@intel.com>
Change-Id: I96a72009aaf96d4237d57f4d5c8b1f41f87174d1
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66281
Reviewed-by: Zhixing Ma <zhixing.ma@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Selma Bensaid <selma.bensaid@intel.com>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
This change provides access to IOE through P2SB Sideband interface for
Meteor Lake TCSS functions of pad configuration and Thunderbolt
authentication. There is a policy of locking the P2SB access at the end
of platform initialization. The tbt_authentication is read from IOM
register through IOE P2SB at early silicon initialization phase and its
usage is deferred to usb4 driver.
BUG=b:213574324
TEST=Built coreboot and validated booting to OS successfully on MTLRVP
board. No boot hung was observed.
Change-Id: Icd644c945bd293a8b9c4a364aaed99ec4a7c12f9
Signed-off-by: John Zhao <john.zhao@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66410
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Subrata Banik <subratabanik@google.com>
Reviewed-by: Tarun Tuli <taruntuli@google.com>
Delete the Thunderbolt authentication function ioe_tcss_valid_tbt_auth
from the common block. Meteor Lake Platform will implement it.
BUG=b:213574324
TEST=Built coreboot image successfully.
Change-Id: I97a289faa6351fe562f91d8478b72c9403ce88cb
Signed-off-by: John Zhao <john.zhao@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66416
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Tarun Tuli <taruntuli@google.com>
DDR5 memory modules have two separate 32-bit channels (40-bit on ECC
memory modules), and the SPD info refers to one channel: the primary
bus width is 32 (or 40) bits and the "DIMM size" is halved. On Alder
Lake, there are 2 memory controllers with 4 32-bit channels each for
DDR5. FSP has 16 positions to store SPD data, some of which are only
used with LPDDR4/LPDDR5.
To try to make things less confusing, FSP abstracts the DDR5 channels
so that the configuration works like on DDR4. This is done by copying
each DIMM's SPD data to the other half-channel. Thus, fix the wrapper
parameters for DDR5 accordingly.
Tested on AlderLake-P DDR5 RVP (board ID 0x12), both DIMM slots now
function properly. Without this patch, only the top slot would work.
Change-Id: I5f01cd77388b89ba34d91c2dc5fb843fe9db9826
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: Lean Sheng Tan <sheng.tan@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66608
Reviewed-by: Lean Sheng Tan <sheng.tan@9elements.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
From Meteorlake, IGD BAR0(GTTMMADR) is changed to 64bit prefetchable.
Due to the prefetchable attribute, resource allocation for IGD BAR0 is
assigned WC memory and it causes kernel driver failure.
For avoiding kernel driver failure, ignore prefetch PCI attribute
for IGD BAR0 to assign UC memory.
We're working on publishing below information.
- IGD BAR0(GTTMMADR) is changed to 64bit prefetchable BAR
- GTTMMADDR BAR should be always mapped as UC memory although
marked Pre-fetchable.
BUG=b:241746156
TEST=boot to OS and check guc driver loading successful
Signed-off-by: Wonkyu Kim <wonkyu.kim@intel.com>
Change-Id: I76d816d51f32f99c5ebcca54f13ec6d4ba77bba5
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66403
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Subrata Banik <subratabanik@google.com>
Reviewed-by: Tarun Tuli <taruntuli@google.com>
The software used to read the document listing the VR settings turns
out to not be perfectly compatible. Indeed, it displays a value of 55A
for RPL-P 282 15W GT ICC MAX while the correct value actually is 40A.
After a thorough review using the software used to create the
document, it is the only value presenting a discrepancy.
BRANCH=firmware-brya-14505.B
BUG=b:239797178
TEST=build and boot
Signed-off-by: Jeremy Compostella <jeremy.compostella@intel.com>
Change-Id: Iee293c87a66f0cd32714766e3ad81eee1a411723
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66057
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Reviewed-by: Anil Kumar K <anil.kumar.k@intel.corp-partner.google.com>
Reviewed-by: Bora Guvendik <bora.guvendik@intel.com>
Reviewed-by: Sumeet R Pawnikar <sumeet.r.pawnikar@intel.com>
Support for feature "In-Band ECC" not available for Tiger Lake
Similar to Elkhart Lake, Tiger Lake also provides this feature.
Ported from Elkhart Lake (CB:55668)
Bug = N/A
TEST = Build and boot Siemens AS-TGL1
Change-Id: Ie54d5f6a9747fad0105d0f8bf725be611bb8cf60
Signed-off-by: Frans Hendriks <fhendriks@eltan.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66372
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
UFS storage devices require the bRefClkFreq attribute to be set to
operate correctly in high speed mode. The correct value is determined by
what the SoC / board supports. For the ADL UFS controller, it is
19.2 MHz.
a) Introduce a new ACPI property "ref-clk-freq".
b) Add support to configure this property using an SoC Kconfig.
Kernel patch:
https://web.archive.org/web/20220801060732/https://lore.kernel.org/all/
20220715210230.1.I365d113d275117dee8fd055ce4fc7e6aebd0bce9@changeid/
BUG=b:238262674
TEST=Build,boot Nirwen and dump SSDT entries and check that the kernel
correctly parses ref-clk-freq as 19.2 MHz.
Scope (\_SB.PCI0)
{
Device (UFS)
{
Name (_ADR, 0x0000000000120007) // _ADR: Address
Name (_DDN, "UFS Controller") // _DDN: DOS Device Name
Name (_DSD, Package (0x02) // _DSD: Device-Specific Data
{
ToUUID ("daffd814-6eba-4d8c-8a91-bc9bbf4aa301")
/* Device Properties for _DSD */,
Package (0x01)
{
Package (0x02)
{
"ref-clk-freq",
0x0124F800
}
}
})
}
}
Signed-off-by: Meera Ravindranath <meera.ravindranath@intel.com>
Signed-off-by: Reka Norman <rekanorman@chromium.org>
Change-Id: I80c338a8a61f161b0feb6c5a3ca00cf5e0cfb36c
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66051
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Ronak Kanabar <ronak.kanabar@intel.com>
Reviewed-by: Eric Lai <eric_lai@quanta.corp-partner.google.com>
Copy the constraint from ADL-S to ADL-P.
Fixes the following warning in Linux on System76 oryp9, which has an
NVIDIA GPU on the bridge.
pcieport 0000:00:01.0: can't derive routing for PCI INT A
This, in turn, resolves an IRQ conflict with the PCH HDA device that
would cause a stack track on every boot.
irq 10: nobody cared (try booting with the "irqpoll" option)
<snip>
[<00000000bf549647>] azx_interrupt [snd_hda_codec]
Disabling IRQ #10
Change-Id: I550c80105ff861d051170ed748149aeb25a545db
Signed-off-by: Tim Crawford <tcrawford@system76.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66285
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add TDP and Power Limit settings for ADL-S 8+8 150W, 4+0 and 2+0.
The System Agent PCI IDs were not present in older 2.1 revision of
DOC #619501. Now that the mapping of these IDs to SKUs is known, fill
the missing TDPs and Power Limit settings based on DOC #626343.
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Change-Id: I23dd8478e60bcc81a1048f2f6e6717dd281d1a69
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66053
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Sumeet R Pawnikar <sumeet.r.pawnikar@intel.com>
The patch updates platform_is_resuming() API such that platform resume
state is determined from the saved state (CBMEM) instead of checking PMC
registers (PM1_STS & PM1_CNT) as they are getting cleared (before/early)
ramstage.
coreboot sends DISCONNECT IPC command which times out during resume (S3)
if system has servoV4 connected on port0. The issue occurs only during
the first cycle of resume (S3) test cycle after cold boot due to side
effect of platform_is_resuming() API that is not determining the resume
(S3) state correctly in ramstage.
PM1_STS and PM1_CNT register gets cleared at the start of ramstage.
platform_is_resuming() function was checks the cleared register value
and fails the condition of resume (S3) resulting in sending DISCONNECT
IPC command. Checking the platform resume state from the CBMEM saved
state using acpe_get_sleep_type() function helps cross verify the
system previous state at the later part of ramstage.
localhost ~ # cbmem -c | grep ERROR
[ERROR] EC returned error result code 3
[ERROR] PMC IPC timeout after 1000 ms
[ERROR] PMC IPC command 0x200a7 failed
[ERROR] pmc_send_ipc_cmd failed
[ERROR] Failed to setup port:0 to initial state
[ERROR] PMC IPC timeout after 1000 ms
[ERROR] PMC IPC command 0x200a7 failed
[ERROR] pmc_send_ipc_cmd failed
[ERROR] Failed to setup port:1 to initial state
[ERROR] GENERIC: 0.0 missing read_resources
[ERROR] PMC IPC timeout after 1000 ms
[ERROR] PMC IPC command 0xd0 failed
[ERROR] PMC: Failed sending PCI Enumeration Done Command
BUG=b:227289581
TEST=Verified system boots to OS and verified below tests on
Redrix (ADL-P) and Nivviks (ADL-N)
1. coreboot doesn't send the DISCONNECT during S3 resume
2. suspend S3 passes with both suzyq and servoV4 connected
3. After S3 resume, system detects the pen drive with Superspeed
4. After system resumes from S3, hot-plug the pen drive, system detects
the pen drive
Signed-off-by: Harsha B R <harsha.b.r@intel.com>
Change-Id: I353ab49073bc4b5288943e19a75efa04bd809227
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66126
Reviewed-by: Subrata Banik <subratabanik@google.com>
Reviewed-by: Sridhar Siricilla <sridhar.siricilla@intel.com>
Reviewed-by: Krishna P Bhat D <krishna.p.bhat.d@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
There are 3 more CPU PCIe RP UPDs that are the current code is not setting,
and some boards may want to set these, so this patch adds support to set
these UPDs. The default values for any existing boards using these UPDs
should not change with this patch.
The UPDs are:
- CpuPcieRpDetectTimeoutMs
- CpuPcieRpAspm
- CpuPcieRpSlotImplemented
Signed-off-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Change-Id: Id48019f984e8e53ff3ce0c3c23e02dab65112c99
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66197
Reviewed-by: Lean Sheng Tan <sheng.tan@9elements.com>
Reviewed-by: Subrata Banik <subratabanik@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The current "normal" EPB (six) setting resulted in the desired out of
box power and performance for several CPU generations.
However, a power and performance analysis on Alder Lake and Raptor
Lake CPUs demonstrates that this value results in undesirable higher
uncore power and that seven is a more appropriate value.
Note: the Linux kernel "4ecc933b x86: intel_epb: Allow model specific
normal EPB value" patch sets the EPB to 7 for Alder Lake.
BRANCH=firmware-brya-14505.B
BUG=b:239853069
TEST=verify that EPB is set by coreboot
Signed-off-by: Jeremy Compostella <jeremy.compostella@intel.com>
Change-Id: I5784656903d4c58bedc5063ee3ef310a99711050
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66059
Reviewed-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Reviewed-by: Cliff Huang <cliff.huang@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
According to document 619503 ADL EDS Vol2, bit 18 of MSR_POWER_CTL
must be set to be able to set the Energy/Performance Bias using MSR
IA32_ENERGY_PERF_BIAS.
Note that since this bit was not set until this patch, the
`set_energy_perf_bias(ENERGY_POLICY_NORMAL);' call in
`soc_core_init()` was systematically failing.
BRANCH=firmware-brya-14505.B
BUG=b:239853069
TEST=verify that EPB is set by coreboot
Signed-off-by: Jeremy Compostella <jeremy.compostella@intel.com>
Change-Id: Ic24abdd7f63f4707b8996da4755a26be148efe4a
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66058
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
This patch fixes the documentation discrepancy of GPIO reset type
between PCH EDS and GPIO BWG.
As per GPIO BWG, there are four GPIO reset types in Meteor Lake as
below:
- Power Good - (Value 00)
- Deep - (Value 01)
- Host Reset/PLTRST - (Value 10)
- Global Reset for GPP - (Value 11)
Also, dropped the need for having dedicated reset type for GPIO
community 3. As per the MTL EDS, all GPIO communities have the same
reset type.
BUG=b:213293047
TEST=Able to build and boot Google/Rex without below error msg.
[ERROR] gpio_pad_reset_config_override: Logical to Chipset mapping
not found
Signed-off-by: Subrata Banik <subratabanik@google.com>
Change-Id: Id7ea16d89b6f01b00a7b7c52945f6e01e8db6cbd
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66155
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Tarun Tuli <taruntuli@google.com>
Reviewed-by: Eric Lai <eric_lai@quanta.corp-partner.google.com>
Reviewed-by: Jamie Ryu <jamie.m.ryu@intel.com>
Reviewed-by: Will Kim <norwayforest92@gmail.com>
This patch updates the GPIO lock configuration recommendation
kconfig string to ensure the SoC user can select the correct
config as applicable for the SoC.
Note: From MTL onwards GPIO lock config can be performed using
PCR write (MMIO write) and the GPIO team has confirmed this.
BUG=b:213596994
TEST=Able to fix below GPIO lock config error msg on MTL with
`SOC_INTEL_COMMON_BLOCK_GPIO_LOCK_USING_PCR` kconfig enabled.
[INFO ] Locking pad configuration using SBI
[INFO ] gpio_pad_config_lock_using_sbi: Locking pad 73
configuration
[ERROR] SBI Failure: Transaction Status = 1
[ERROR] Failed to lock GPIO PAD, response = 1
Signed-off-by: Subrata Banik <subratabanik@google.com>
Change-Id: Icab1e4849b8e08ee1c695c924599f1513774178f
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66113
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Tarun Tuli <taruntuli@google.com>
Reviewed-by: Eric Lai <eric_lai@quanta.corp-partner.google.com>
