CFR default overrides use SM_OBJ_* kinds, but
write_numeric_option() compared them to CFR_TAG_OPTION_*,
so enum/number/bool overrides were always skipped.
Compare kinds against the expected SM_OBJ_* for each numeric
tag, then apply the override.
Change-Id: I02046974a7b0a3ef32973689833e1b0d38a5d6f4
Signed-off-by: Sean Rhodes <sean@starlabs.systems>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/90911
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
When updating clang to version 21.1.5, we now get the following error
when compiling the CFR code.
```
src/drivers/option/cfr.c:287:19: error: default initialization of an
object of type 'struct sm_object' with const member leaves the object uninitialized
[-Werror,-Wdefault-const-init-field-unsafe]
287 | struct sm_object sm_obj_copy;
```
In this specific case its actually not an issue, but initialize the
variable from the start to avoid the error and arguably make the code
easier to read.
Signed-off-by: Maximilian Brune <maximilian.brune@9elements.com>
Change-Id: Ib0b22e20a7482ab29d997a3e3c3bec4a1db1a4f1
Reviewed-on: https://review.coreboot.org/c/coreboot/+/89958
Reviewed-by: Elyes Haouas <ehaouas@noos.fr>
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
There is no reason I can think to also pass the old object into the
constructor considering that the new objects contains the exact same
contents during that call (it's copied over a few lines above).
Signed-off-by: Maximilian Brune <maximilian.brune@9elements.com>
Change-Id: Idd5b34134b6064c19266448b551248eb29e097fe
Reviewed-on: https://review.coreboot.org/c/coreboot/+/89957
Reviewed-by: Elyes Haouas <ehaouas@noos.fr>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Sean Rhodes <sean@starlabs.systems>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Add a mechanism for mainboards to override default values of CFR
objects defined in SoC or common code without duplicating object
metadata.
Mainboards can now declare a simple override table mapping option
names to new default values:
const struct cfr_default_override mb_cfr_overrides[] = {
CFR_OVERRIDE_BOOL("s0ix_enable", false),
CFR_OVERRIDE_ENUM("pciexp_aspm", ASPM_DISABLE),
CFR_OVERRIDE_END
};
The CFR backend checks this table when writing options and uses the
override value if one exists. All other metadata (name, help text,
enum values, flags) comes from the original object.
Change-Id: Ifb3da90d605f2799bf0207ff58d69bee3415ccc2
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/89933
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Since CFR options require a backend to store the keys/values, select
DRIVERS_EFI_VARIABLE_STORE when edk2 is used as the payload and
SMMSTORE is enabled, so that boards only need to select
DRIVERS_OPTION_CFR in order to have a fully-functioning configuration
setup.
TEST=build samsung/stumpy with DRIVERS_OPTION_CFR selected and edk2
payload used.
Change-Id: Ib8565e4fefb1b3f05e58ab039be8ab0d1bc046f1
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/87410
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
This commit adds support for minimum/maximum limit values as well as
step sizes for CFR number options. Additionally, add a new flag that
specifies the option should be displayed in hexadecimal notation instead
of decimal.
Change-Id: I2e70f1430fb1911f1ad974832f8abfe76f928ac3
Signed-off-by: Filip Brozovic <fbrozovic@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/86039
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Implements a way for CFR options to depend on another option
being set to one or more specific values. This is achieved
by writing a list of values as a varbinary struct.
Change-Id: Iaf7965551490969052eb27c207fa524470d4dd6a
Signed-off-by: Filip Brozovic <fbrozovic@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/85987
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-by: Sean Rhodes <sean@starlabs.systems>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add a version field to the CFR root struct so parsers can check
compatibility when parsing structs.
Change-Id: Ifcb950f1bdedc0ab925f3841befb7e7001c0f7f4
Signed-off-by: Filip Brozovic <fbrozovic@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/86080
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Add a new option backend that uses values stored in CBFS files, similar
to the SeaBIOS runtime config options stored in files with the etc/
prefix. Options should be stored in CBFS with the option/ prefix. Values
can be set using `cbfstool coreboot.rom add-int -n option/<option-name>
-i <value>`. For simplicity, options should be stored in the COREBOOT
(RO) FMAP region, which is the default for cbfstool. This backend is not
available in SMM due to CBFS dependencies on vboot functions which are
not added to SMM, and thus the fallback will be returned by calls to
get_uint_option() in SMM.
Tested with QEMU Q35 by setting various options for "sata_mode" and
observing the console output for the SATA controller mode during
i82801ix_sata initialization.
Change-Id: Ifc0439ee42f13f49ae54d4855d1d9333c39b01f5
Signed-off-by: Nicholas Chin <nic.c3.14@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/85905
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Subrata Banik <subratabanik@google.com>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Reviewed-by: Jérémy Compostella <jeremy.compostella@intel.com>
Fix some typos and also update the naming convention of
`CFR_OPTFLAG_GRAYOUT` to `CFR_OPTFLAG_INACTIVE` as per reviews.
Signed-off-by: Lean Sheng Tan <sheng.tan@9elements.com>
Change-Id: Id66808382b93e32c58024462c18b20c2a89d6d23
Reviewed-on: https://review.coreboot.org/c/coreboot/+/85780
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Christian Walter <christian.walter@9elements.com>
Introduce a mechanism so that coreboot can provide a list of options to
post-coreboot code. The options are grouped together into forms and
have a meaning name and optional help text. This can be used to let
payloads know which options should be displayed in a setup menu,
for instance. Although this system was written to be used with edk2,
it has been designed with flexibility in mind so that other payloads
can also make use of this mechanism. The system currently lacks a way
to describe where to find option values.
This information is stored in a set of data structures specifically
created for this purpose. This format is known as CFR, which means
"coreboot forms representation" or "cursed forms representation".
Although the "forms representation" is borrowed from UEFI, CFR can
be used in non-UEFI scenarios as well.
The data structures are implemented as an extension of cbtables records
to support nesting. It should not break backwards compatibility because
the CFR root record (LB_TAG_CFR_ROOT) size includes all of its children
records. The concept of record nesting is borrowed from the records for
CMOS options. It is not possible to reuse the CMOS records because they
are too closely coupled with CMOS options; using these structures would
needlessly restrict more capable backends to what can be done with CMOS
options, which is undesired.
Because CFR supports variable-length components, directly transforming
options into CFR structures is not a trivial process. Furthermore, CFR
structures need to be written in one go. Because of this, abstractions
exist to generate CFR structures from a set of "setup menu" structures
that are coreboot-specific and could be integrated with the devicetree
at some point. Note that `struct sm_object` is a tagged union. This is
used to have lists of options in an array, as building linked lists of
options at runtime is extremely impractical because options would have
to be added at the end of the linked list to maintain option order. To
avoid mistakes defining `struct sm_object` values, helper macros exist
for supported option types. The macros also provide some type checking
as they initialise specific union members.
It should be possible to extend CFR support for more sophisticated
options like fan curve points. Feedback about this is highly
appreciated.
Change-Id: I304de7d26d79245a2e31a6d01f6c5643b31cb772
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/74121
Reviewed-by: Christian Walter <christian.walter@9elements.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
