Provide a helper function to locate PCI devices on a given socket
by their PCI vendor and device IDs and functions to return
information about the current device, like the corresponding stack
and socket.
In addition add functions to return "location" information, like stack
and socket affiliation.
This becomes handy when locating devices and generating ACPI code.
Change-Id: I266360588548ba579f46b228c4d5b3ae6e39a029
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80094
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Shuo Liu <shuo.liu@intel.com>
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Read the VtdBar and add it to the resources of the host bridge PCI
device. The BAR is already marked as PciResourceMem32 in the parent
PCI domain.
This allows easy probing for VTD devices with enabled VtdBars in the
next commit, without the need to look up the stack HOB.
Change-Id: Id579a94e653473f3dd0dccea6e33dc64f792d028
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80550
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
If a downstream LPC device (eg, SIO function) is disabled, we shouldn't
attempt to open PMIO windows for it, as those functions often have
unset IO bases (which default to 0), resulting in false errors like:
[ERROR] LPC IO decode base 0!
TEST=build/boot purism/librem_cnl (Mini v2), verify no LPC IO errors
in cbmem log for disabled SIO functions.
Change-Id: I92c79fc01be21466976f3056242f6d1824878eab
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80646
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Subrata Banik <subratabanik@google.com>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Early EC Sync does not need to be enabled in coreboot as EFS2 is being
enabled in the EC.
BUG=b:326152804
BRANCH=None
TEST=emerge-brox coreboot
To be tested with EC sync enabled
Change-Id: I08bdbe9f3dcea837b0b148adc137c03d3461877a
Signed-off-by: Shelley Chen <shchen@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80689
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The lower bit of the BAR might be used for something else,
like enable bits, so mask the lower 12 bits and align all
base address to 4K.
Confirmed that all BARs have a minimum alignment of 4K, so that
masking the lower bits doesn't change the reported address.
The alignment of the VTD BARs is:
- VTD_MMCFG_BASE_CSR 64 MiB
- VTD_MMIOL_CSR 1 MiB
- VTD_NCMEM_BASE_CSR 64 MiB
- VTD_TSEG_BASE_CSR 1 MiB
- VTD_BAR_CSR 4 KiB
Change-Id: I9a7b963c0074246616968dd15c147f4916297d59
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80548
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
IOAT logics are optimized out for non-IOAT platforms where
CONFIG(HAVE_IOAT_DOMAINS) as false.
This patch puts CONFIG(HAVE_IOAT_DOMAINS) check together ahead
of is_ioat_iio_stack_res() check in the corresponding if
statement to fulfill the optimization outs.
TEST=intel/archercity CRB
Change-Id: I2d16c6ff5320bc9195a1033b6d55e3d997b19b88
Signed-off-by: Shuo Liu <shuo.liu@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80683
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add supported memory parts in mem_parts_used list, and generate SPD ID
for these parts.
DRAM Part Name ID to assign
K3KL8L80CM-MGCT 0 (0000)
K3KL6L60GM-MGCT 1 (0001)
H58G56AK6BX069 2 (0010)
H9JCNNNBK3MLYR-N6E 3 (0011)
BUG=b:319071869
BRANCH=firmware-nissa-15217.B
TEST=Run command "go run ./util/spd_tools/src/part_id_gen/\
part_id_gen.go ADL lp5 \
src/mainboard/google/brya/variants/glassway/memory/ \
src/mainboard/google/brya/variants/glassway/memory/\
mem_parts_used.txt"
Change-Id: I00ae3efe8e554f44cee5a27ac88c5d65eb95f7fb
Signed-off-by: Daniel Peng <Daniel_Peng@pegatron.corp-partner.google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80686
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Eric Lai <ericllai@google.com>
Reviewed-by: Daniel Peng <daniel_peng@pegatron.corp-partner.google.com>
The Rotation Matrix allows the specification of a 3x3 matrix
representing the orientation of devices, such as accelerometers.
Each value in the matrix can be one of -1, 0, or 1, indicating the
transformation applied to the device's axes.
It is expected by Linux and required for the OS to interpret
the data from the device correctly. It is used by various drivers,
mainly in `iio/accel`.
It was tested on Ubuntu, by rotating the device and verifying the
orientation was correct.
Signed-off-by: Sean Rhodes <sean@starlabs.systems>
Change-Id: Id4a940d999a0e300a6fe21269f18bab6e3c0523c
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80179
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
This adds an optimization to lzma decode to also read from the boot
medium in chunks of 8 bytes if that is the general purpose register
length instead of always 4 bytes. It depends on the cache / memory / spi
controller whether this is faster, but it's likely to be either the same
or faster.
TESTED
- google/vilboz: cached boot medium
64bit before - 32bit - 64bit after
load FSP-M: 35,674 - 35,595 - 34,690
load ramstage: 42,134 - 43,378 - 40,882
load FSP-S: 24,954 - 25,496 - 24,368
- foxconn/g41m: uncached boot medium for testing
64bit before - 32bit - 64bit after
load ramstage: 51,164 - 51,872 - 51,894
Change-Id: I890c075307c0aec877618d9902ea352ae42a3bfa
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/70175
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Lean Sheng Tan <sheng.tan@9elements.com>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Keeping the PM timer enabled will disqualify an ADL system from entering
S0i3, and will also cause an increase in power during suspend states.
The PM timer is not required for brya boards, therefore disabling it.
Fixes: 0e90580 (soc/intel: transition full control over PM Timer from
FSP to coreboot)
This mirrors an identical commit for google/brya: 1ce0f3aab7
("mb/google/brya: Fix S0i3 regression")
TEST=Boot Linux on google/drobit, verify S0i3 counter incrementing after
exiting S0ix suspend states.
Change-Id: I644e42388c0f6127512bf52e774b79721601ecc9
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80612
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Eric Lai <ericllai@google.com>
Reviewed-by: Felix Singer <service+coreboot-gerrit@felixsinger.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Macros MAX_ACPI_MEMORY_AFFINITY_COUNT and MAX_SRAT_MEM_ENTRIES_PER_IMC
are ACPI table specific, and could be used across Xeon-SP SoCs.
This patch moves their definition from FSP header to Xeon-SP layer
ACPI header.
TEST=intel/archercity CRB
Signed-off-by: Shuo Liu <shuo.liu@intel.com>
Change-Id: I6c3a84b04a452bc8d4217947a7d12f050c94b56b
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80629
Reviewed-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
IOAT is the term for the on-chip accelerator technology of
Xeon-SP. In CPX and SPR, IOAT stack is also named as DINO stack.
Different SoC has different check criteria for IOAT stacks,
this patch introduces an util function to abstract these differences
as well as cleaning up the usage of names.
TEST=intel/archercity CRB
Change-Id: I376928ad89b68b294734000678dad6f070d3c97d
Signed-off-by: Shuo Liu <shuo.liu@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80578
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
It's not needed other than for booting w/SeaBIOS, where it is already
selected by default, and enabling it with edk2 payload prevents Linux/
Windows from fully entering S0ix.
TEST=build/boot purism/librem_cnl (Mini v2), verify Win11/Linux able
to enter and exit S0ix properly.
Change-Id: I974a82bedc4e06f48ce801f2bc0c29afbd80ffcf
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80602
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin L Roth <gaumless@gmail.com>
Reviewed-by: Felix Singer <service+coreboot-gerrit@felixsinger.de>
Only call fill_pds() once to prevent leaking memory. Previously it was
called for every active stack on every socket.
Only call dump_pds() once to prevent spamming the console with the same
information.
Drop the return value since it's always returning success.
Change-Id: Ifa9609e9da086dc9731556014ea9b320b270d776
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80547
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Shuo Liu <shuo.liu@intel.com>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
The struct map_entry has two zero'd entries due to the ifdef
being used. Do not read those entries and do not print those
entries.
Fixes a NULL string being printed along as the vendor and device
ID of the PCI device.
Change-Id: Id87ced76af552c0d064538f8140d1b78724fb833
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80546
Reviewed-by: Shuo Liu <shuo.liu@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Now that the baseboard uses chipset devicetree references, remove
all references whose value is identical to the chipset devicetree
default or the baseboard default, since they are pointless clutter.
TEST=build/boot purism/librem_cnl (Mini v2), verify output of lspci
and lsusb unchanged before and after patch.
Change-Id: I12498e7261dafd7ee59fe79926532399392d1b09
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80600
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Martin L Roth <gaumless@gmail.com>
Reviewed-by: Felix Singer <service+coreboot-gerrit@felixsinger.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Now that the board uses chipset devicetree references, remove all
references whose value is identical to the chipset devicetree default,
since they are pointless clutter.
TEST=build/boot purism/librem_cnl (Mini v2), run lspci and verify output
unchanged before and after patch.
Change-Id: I6c656d227962548cebde61f1d82333837adbbf56
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80599
Reviewed-by: Felix Singer <service+coreboot-gerrit@felixsinger.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This patch eliminates coreboot from loading microcode from RW CBFS
(when the RO descriptor is locked, which indicates a fixed RO image)
because the kernel can already patch the microcode on BSPs and APs
while booting to OS.
This may be a chance to lower the burden on the AP FW side because
patching microcode on in-field devices is subject to firmware updates,
which are rarely published and, if required, must go through the
firmware qualification testing procedure (which is costly, unlike
kernel updates for ucode updates).
1. The FIT loads the necessary microcode from the RO during reset.
2. Reloading microcode from RW CBFS impacts boot time
(~60ms, core-dependent).
3. The kernel can still load microcode updates.
ChromeOS devices leverage RO+RW-A/RW-B booting. The RO's microcode is
sufficient for initial boot, and the kernel can apply updates later.
BUG=none
TEST=Verified boot optimization; in-field devices skip RW-CBFS microcode
loading when RO is locked.
Change-Id: Ia859809970406fca3fa14e6fa8e766ab16d94c8a
Signed-off-by: Subrata Banik <subratabanik@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80567
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: YH Lin <yueherngl@google.com>
Add an ACPI stub containing the TCPU device in proper scope, along with
the device status, on boards not using the DPTF driver, so that there
exists an ACPI device to be referenced from the PEPD LPI constraint
list.
Adding the stub fixes an AE_NOT_FOUND ACPI error under Linux for
_SB.PCI0.TCPU on boards with the SA thermal device enabled but which do
not use the Intel DPTF driver.
TEST=build/boot Linux,Win11 on purism/librem_cnl (Librem Mini v2).
Change-Id: I926d0461e5e0dfaf606102575c2be555a6bfb695
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80553
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Singer <service+coreboot-gerrit@felixsinger.de>
Reviewed-by: Nico Huber <nico.h@gmx.de>
