When cbmem is initialized in romstage and postcar placed in the stage
cache + cbmem where it is run, the assumption is made that these are
all in UC memory such that calling INVD in postcar is OK.
For performance reasons (e.g. postcar decompression) it is desirable
to cache cbmem and the stage cache during romstage.
Another reason is that AGESA sets up MTRR during romstage to cache all
dram, which is currently worked around by using additional MTRR's to
make that UC.
TESTED on asus/p5ql-em, up/squared on both regular and S3 resume
bootpath. Sometimes there are minimal performance improvements
when cbmem is cached (few ms).
Change-Id: I7ff2a57aee620908b71829457ea0f5a0c410ec5b
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/37196
Reviewed-by: Lean Sheng Tan <sheng.tan@9elements.com>
Reviewed-by: Kapil Porwal <kapilporwal@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Instead of having a magic entry in the CPU device ID table list to tell
find_cpu_driver that it has reached the end of the list, introduce and
use CPU_TABLE_END. Since the vendor entry in the CPU device ID struct is
compared against X86_VENDOR_INVALID which is 0, use X86_VENDOR_INVALID
instead of the 0 in the CPU_TABLE_END definition.
TEST=Timeless build for Mandolin results in identical image.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Suggested-by: Angel Pons <th3fanbus@gmail.com>
Change-Id: I0cae6d65b2265cf5ebf90fe1a9d885d0c489eb92
Reviewed-on: https://review.coreboot.org/c/coreboot/+/72888
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Introduce a macro to get the raw CPUID leaf 1 EAX value from a given set
of CPU family, model and stepping. The processor type in bits 12 and 13
is assumed to be always be zero; at least this is the case for all
CPUIDs that are currently in the coreboot tree. This can be used to
make the device values in the CPU device ID tables easier to read.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Idab77453712b14983b1d02ca365f7924239fc2bf
Reviewed-on: https://review.coreboot.org/c/coreboot/+/72856
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Instead of always doing exact matches between the CPUID read in
identify_cpu and the device entries of the CPU device ID table,
offer the possibility to use a bit mask in the CPUID matching. This
allows covering all steppings of a CPU family/model with one entry and
avoids that case of a missing new stepping causing the CPUs not being
properly initialized.
Some of the CPU device ID tables can now be deduplicated using the
CPUID_ALL_STEPPINGS_MASK define, but that's outside of the scope of this
patch.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I0540b514ca42591c0d3468307a82b5612585f614
Reviewed-on: https://review.coreboot.org/c/coreboot/+/72847
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Now that we support >1 Xeon-SP, XEON_SP_COMMON_BASE no longer
reflects the socket type. This uses SOC_INTEL_* Kconfig variables and
returns the correct socket type for Cooper Lake-SP.
Signed-off-by: David Hendricks <ddaveh@amazon.com>
Change-Id: I142de5f040f3b76e352f27c00fe9e50787df5712
Reviewed-on: https://review.coreboot.org/c/coreboot/+/72498
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Jonathan Zhang <jonzhang@fb.com>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
A mainboard port needs to:
- select `CONFIG_MAINBOARD_HAS_EARLY_LIBGFXINIT'
- implement the Ada package `GMA.Mainboard' with a single function
`ports' that returns a list of ports to be probed for displays.
- set the desired `GFX_GMA_DEFAULT_MMIO' IO memory address to use
in romstage (and ramstage) for the graphic device.
BUG=b:252792591
BRANCH=firmware-brya-14505.B
TEST=libgfxinit compiles in romstage.
libgfxinit successfully executes in romstage and ramstage using
the requested MMIO setting on skolas.
Change-Id: I3c2101de10dc5df54fe873e43bbe0f1c4dccff44
Signed-off-by: Jeremy Compostella <jeremy.compostella@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/70276
Reviewed-by: Tarun Tuli <taruntuli@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
It's hidden behind the configuration option `CONFIG_ROMSTAGE_LIBHWBASE'.
This also adds some glue code to use the coreboot console for debug
output and our monotonic timer framework as timer backend.
Running Ada code in romstage and more particular libhwbase brings a few
challenges as global initialized variables are not supported in
Cache-As-Ram mode.
1. The libhwbase dynamic mmio driver implementation makes the Gnat
compiler generate some global initialized variables.
For this reason, when compiled for romstage or for romstage and
ramstage the static mmio driver is enforced (`HWBASE_STATIC_MMIO').
2. The Gnat compiler generates elaboration functions to initialize
program data at runtime. These elaboration functions are called by
the romstage_adainit() function.
The data references symbols suffixed by `_E'. Even though these
symbols, at compilation time, do not contain any data and are
filled with zeros, the Gnat compiler installs them in the .data
section.
Since these symbols are actually filled with zeros, it is safe to
install them in the .bss section.
cf. https://docs.adacore.com/gnat_ugn-docs/html/gnat_ugn/gnat_ugn/elaboration_order_handling_in_gnat.html#elaboration-code
This patch requires the libhwbase
https://review.coreboot.org/c/libhwbase/+/69854 CL.
BUG=b:252792591
BRANCH=firmware-brya-14505.B
TEST=libhwbae compiles for romstage and loads successfully
Change-Id: I670249d33506e886a683e55d1589cb2bf9b16aa3
Signed-off-by: Jeremy Compostella <jeremy.compostella@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/70275
Reviewed-by: Boris Mittelberg <bmbm@google.com>
Reviewed-by: Nick Vaccaro <nvaccaro@google.com>
Reviewed-by: Tarun Tuli <taruntuli@google.com>
Reviewed-by: Zhixing Ma <zhixing.ma@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Properly handle meminfo DIMMs with `dimm_size` of 0, which represent
empty slots. This allows platform code to create dummy meminfo DIMMs
so that SMBIOS tables have type 17 entries for empty DIMM slots.
Change-Id: I17ae83edf94483bd2eeef5524ff82721c196b8ba
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/64035
Reviewed-by: Jonathan Zhang <jonzhang@fb.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This reverts commit 095c931cf1.
Previously cpu_info() was implemented with a struct on top of an
aligned stack. As FSP changed the stack value cpu_info() could not be
used in FSP context (which PPI is). Now cpu_info() uses GDT segments,
which FSP does not touch so it can be used.
This also exports cpu_infos from cpu.c as it's a convenient way to get
the struct device * for a certain index.
TESTED on aldrvp: FSP-S works and is able to run code on APs.
Change-Id: I3a40156ba275b572d7d1913d8c17c24b4c8f6d78
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/69509
Reviewed-by: Subrata Banik <subratabanik@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Taken from Linux which also updated these files.
Clang only works with this syntax, so this fixes builds for arm.
TESTED on qemu vexpress-a9 and verstage on google/vilboz with
BUILD_TIMELESS=1, binaries remain the same.
Change-Id: Ia320dc2c460c99d934b8f17dee7748a9def4e750
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63058
Reviewed-by: Martin L Roth <gaumless@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
