Now that we have functionality to get the minimal and nominal
frequencies, the corresponding fields in the CPPC config can now be
populated. If the HOB isn't present and/or the frequency values
could not be obtained, CPPC_UNSUPPORTED is still used; otherwise the
HOB-provided frequency in MHz is used for those two fields.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Signed-off-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Change-Id: Id3257690a3388d44ceceb7ac4f1db3d49e195caa
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66551
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add common AMD FSP functionality to get the nominal and minimal CPU core
CPPC frequencies. Those functions will be used in the _CPC ACPI object
generation in a follow-up patch.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Signed-off-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Change-Id: I68ebdf610795d2673e0118a732f54f5f719b73c0
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66550
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Checking if the return value of the fsp_find_extension_hob_by_guid call
is NULL should make the code a bit easier to read.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I6bdb07eab6da80f46c57f5d7b3c894b41ac23b8a
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66699
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The PSP currently uses a hard coded GPIO for the TPM IRQ. Not all board
versions use the same GPIO. This method allows the mainboard to pass
in the correct GPIO.
BUG=b:241824257
TEST=Boot guybrush and verify PSP message prints
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: Ie05d095d7f141d6a526d08fbf25eb2652e96aa49
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66614
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
The Cezanne CPPC ACPI table generation code also applies to Sabrina, so
move it to the common AMD SoC code directory so that it can be used for
Sabrina too.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I5ce082a27429948f8af7f55944a1062ba03155da
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66400
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
We want to extend the vb2ex_hwcrypto APIs on the vboot side to allow
passing 0 for the data_size parameter to vb2ex_hwcrypto_digest_init()
(see CL:3825558). This is because not all use cases allow knowing the
amount of data to be hashed beforehand (most notable the metadata hash
for CBFS verification), and some HW crypto engines do not need this
information, so we don't want to preclude them from optimizing these use
cases just because others do.
The new API requirement is that data_size may be 0, which indicates that
the amount of data to be hashed is unknown. If a HW crypto engine cannot
support this case, it should return VB2_ERROR_EX_HWCRYPTO_UNSUPPORTED to
those calls (this patch adds the code to do that to existing HW crypto
implementations). If the passed-in data_size value is non-zero, the HW
crypto implementation can trust that it is accurate.
Also reduce a bit of the console spew for existing HW crypto
implementations, since vboot already logs the same information anyway.
Signed-off-by: Julius Werner <jwerner@chromium.org>
Change-Id: Ieb7597080254b31ef2bdbc0defc91b119c618380
Reviewed-on: https://review.coreboot.org/c/coreboot/+/66621
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
'Mendocino' was an embargoed name and could previously not be used
in references to Skyrim. coreboot has references to sabrina both
in directory structure and in files. This will make life difficult
for people looking for Mendocino support in the long term. The code
name should be replaced with "mendocino".
BUG=b:239072117
TEST=Builds
Cq-Depend: chromium:3764023
Cq-Depend: chromium:3763392
Cq-Depend: chrome-internal:4876777
Signed-off-by: Jon Murphy <jpmurphy@google.com>
Change-Id: I2d0f76fde07a209a79f7e1596cc8064e53f06ada
Reviewed-on: https://review.coreboot.org/c/coreboot/+/65861
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
The default state for the IRQ lines when the eSPI controller comes out
of reset is high. This is because the IRQ lines are shared with the
other IRQ sources using AND gates. This means that in order to not cause
any spurious interrupts or miss any interrupts, the IO-APIC must use a
low polarity trigger.
On zork/guybrush/skyrim the eSPI IRQs are currently working as follows:
* On power on/resume the eSPI controller drives IRQ 1 high.
* eSPI controller gets configured to not invert IRQ 1.
* OS configures IO-APIC IRQ 1 as Edge/High.
* EC writes to HIKDO (Keyboard Data Out) which causes the EC to set IRQ1
high.
* eSPI controller receives IRQ 1 high, doesn't invert it, and leaves IRQ
1 as high. This results in missing the first interrupt.
* When the x86 reads from HIKDO, the EC deasserts IRQ1. This causes the
eSPI controller to set IRQ1 to low. We are now primed to catch the
next edge high interrupt. This is generally not a problem since the
linux driver will probe the 8042 with interrupts off.
On S3/S0i3 resume since the eSPI controller comes out of reset driving
the IRQ lines high, we trigger a spurious IRQ since the IO-APIC is
configured to trigger on edge high. This results in the 8042 controller
getting incorrectly marked as a wake trigger.
By configuring the IO-APIC to use low polarity interrupts, we no longer
lose the first interrupt. This also means we can use a level interrupt
to match what the EC actually asserts.
We use the `Interrupt` keyword instead of the `IRQ` keyword in the ACPI
because the linux kernel will ignore the level/polarity parameters
for the `IRQ` keyword and default to `edge/high. `Interrupt` doesn't
have this problem.
The PIC is not currently configured anywhere and it defaults to an
edge/high trigger. We could add some code to configure the PICs trigger
register, but I don't think we need the functionality right now.
For zork and guybrush, this change is a no-op. eSPI is configured in
verstage which is located in RO, and we have already locked RO for
these devices. We will need to figure out how to properly set the
`vw_irq_polarity` for these devices.
BUG=b:218874489, b:160595155, b:184752352, b:157984427, b:238818104
TEST=On zork, guybrush and skyrim
$ suspend_stress_test --post_resume_command 'cat /sys/devices/platform/i8042/serio0/wakeup/wakeup35/active_count'
Verify keyboard works as expected and no interrupt storms are observed.
On morphius I verified keyboard and mouse work on windows as well.
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I4608a7684e34ebb389e0e55ceba7e7441939afe7
Reviewed-on: https://review.coreboot.org/c/coreboot/+/54924
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Add static check to ensure the reserved APOB DRAM space is the same size
as the MRC_CACHE region specified in the fmap.
Update sabrina APOB DRAM size to match the fmap.
TEST: Timeless builds identical. Test build with a larger MRC_CACHE than
APOB DRAM failed the assert as expected.
Signed-off-by: Fred Reitberger <reitbergerfred@gmail.com>
Change-Id: Ia14f6ef94b9062df0612fe96098b1012085ccf9c
Reviewed-on: https://review.coreboot.org/c/coreboot/+/65878
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The SPI_RESTRICTED_CMD register is not a PCI configuration register. It
is memory mapped from the SPI bar.
Verified against PPR 55570 rev 3.16, PPR 56569 rev 3.03, and PPR 57243
rev 1.50
TEST=Compile tested only
Change-Id: I7c88aaea9ddac200644bb368be3bd4e9be47fd7b
Signed-off-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63305
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
On SoCs where PSP use A/B recovery layout, PSP expects PSP L2 directory
address relative to the start of the SPI ROM. Unfortunately there is
nothing in the EFS2 header to help identify such SoCs. Hence add a
config item to statically identify such SoCs.
Also when PSP uses A/B recovery layout, BIOS L2 directory is an entry in
the PSP L2 directory. Hence the address of BIOS L2 directory is not part
of EFS2 header. Thankfully PSP is able to identify the BIOS L2 directory
itself and does not expect PSP verstage to pass the address. Modify PSP
verstage to handle these updates.
BUG=b:217414563
TEST=Build Skyrim BIOS image. Ensure that PSP verstage returned the PSP
L2 directory as expected.
Change-Id: I2f856a62055c80b8e2db91c983832611a5f0389c
Signed-off-by: Karthikeyan Ramasubramanian <kramasub@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/65865
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The APOB in sabrina is larger than in cezanne/picasso and no longer
fits in the previously allocated 64K space for it. Other symbols are
placed immediately after the APOB region and end up corrupting the APOB
data on sabrina.
Add a Kconfig option to specify the APOB size in DRAM to reserve enough
memory and increase the size for sabrina to 128K
TEST=Timeless builds are identical for mandolin/majolica for PCO/CZN.
Build chausie and verify symbols do not overlap _apob region
BUG=b:224056176
Signed-off-by: Fred Reitberger <reitbergerfred@gmail.com>
Change-Id: Ia5dbacae67ff02fc8a6ec84b9007110ca254daa3
Reviewed-on: https://review.coreboot.org/c/coreboot/+/65852
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Since there are many identifiers whose name contain "__unused" in
headers of musl libc, introducing a macro which expands "__unused" to
the source of a util may have disastrous effect during its compiling
under a musl-based platform.
However, it is hard to detect musl at build time as musl is notorious
for having explicitly been refusing to add a macro like "__MUSL__" to
announce its own presence.
Using __always_unused and __maybe_unused for everything may be a good
idea. This is how it works in the Linux kernel, so that would at least
make us match some other standard rather than doing our own thing
(especially since the other compiler.h shorthand macros are also
inspired by Linux).
Signed-off-by: Bill XIE <persmule@hardenedlinux.org>
Change-Id: I547ae3371d7568f5aed732ceefe0130a339716a9
Reviewed-on: https://review.coreboot.org/c/coreboot/+/65717
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Reviewed-by: Julius Werner <jwerner@chromium.org>
This avoids searching the HOB output multiple times when calling
smm_region().
Change-Id: Iad09c3aa3298745ba3ba7012e6bb8cfb8785d525
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/65787
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reverting commit 1e25fd426a ("soc/amd/common/block/psp: introduce
AMD_SOC_SEPARATE_EFS_SECTION").
A better solution was used in commit c17330c1dd ("mb/amd/chausie: Add
EC blob into CBFS"), and this is no longer necessary.
TEST: Boot chausie
Signed-off-by: Fred Reitberger <reitbergerfred@gmail.com>
Change-Id: I27a8622a1f0d871690b181a79adca225a20996ea
Reviewed-on: https://review.coreboot.org/c/coreboot/+/65492
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Return the correct processor family code for smbios per System
Management BIOS (SMBIOS) Reference Specification DSP0134 revision 3.5.0.
BUG=b:234409052
TEST=Boot chausie to chromeos and verify "dmidecode -t processor"
outputs the correct processor family.
Signed-off-by: Fred Reitberger <reitbergerfred@gmail.com>
Change-Id: I617ce3e23f4b28a197034756d285339595d3b53b
Reviewed-on: https://review.coreboot.org/c/coreboot/+/65364
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
There is a lot of going back-and-forth with the KiB arguments, start
the work to migrate away from this.
Change-Id: I329864d36137e9a99b5640f4f504c45a02060a40
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/64658
Reviewed-by: Werner Zeh <werner.zeh@siemens.com>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Commit 96f7b96866 (soc/amd/common/block/
cpu/: Make ucode update more generic) removed the code that used the
SOC_AMD_COMMON_BLOCK_UCODE_SIZE Kconfig value. Drop the now unused
Kconfig option.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I079f229678452ff20d8bb282804cd2e49555a6fb
Reviewed-on: https://review.coreboot.org/c/coreboot/+/65255
Reviewed-by: ritul guru <ritul.bits@gmail.com>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Jason Glenesk <jason.glenesk@gmail.com>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Without this, calls to i2c_link() and runtime i2c detection fails on
AMD common platform boards.
Test: Runtime i2c detection of correct touchpad model succeeds on
google/zork.
Change-Id: I238b680b2afb4b9d3e5ac75fe9e630b2adc74860
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/65232
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
All AMD SoCs which select SOC_AMD_COMMON_BLOCK_I2C also select
DRIVERS_I2C_DESIGNWARE, so make the pairing explicit by moving the
selection into SOC_AMD_COMMON_BLOCK_I2C. This will facilitating adding
the Designware I2C bus ops handler in a subsequent commit.
Change-Id: Ice30c8806766deb9a6ba617c3e633ab069af3b46
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/65231
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
TSEG does not need to be aligned to 128KiB but to its size, as the MSR
works like an MTRR. 128KiB is a minimum TSEG size however.
TESTED on google/vilboz.
Change-Id: I30854111bb47f0cb14b07f71cedacd629432e0f4
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/64865
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Use the equivalent cpuid in the microcode header to name the update file
in cbfs. This allows the SOC to directly locate its microcode file when
there are multiple processor revisions.
TEST: Loaded a chausie with sabrina, cezanne, and picasso microcode
files and booted. Verified that only the sabrina microcode file was
successfully loaded
Change-Id: I84a2480cf8274d53ffdab7864135c1bf001241e6
Signed-off-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63589
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
All targets now use cbmem for the BERT region, so the implementation can
be common.
This also drops the obsolete comment about the need to have bert in a
reserved region (cbmem gets fixed to be in a reserved region).
Change-Id: I6f33d9e05a02492a1c91fb7af94aadaa9acd2931
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/64602
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
On platforms where the bootblock is not included in CBFS anymore
(because it is part of another firmware section (IFWI or a different
CBFS), the CRTM measurement fails.
This patch adds a new function to provide a way at SoC level to measure
the bootblock. Following patches will add functionality to retrieve the
bootblock from the SoC related location and measure it from there.
In this way the really executed code will be measured.
Change-Id: I6d0da1e95a9588eb5228f63151bb04bfccfcf04b
Signed-off-by: Werner Zeh <werner.zeh@siemens.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/64492
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
There are efforts to have bootflows that do not follow a traditional
bootblock-romstage-postcar-ramstage model. As part of that CBMEM
initialisation hooks will need to move from romstage to bootblock.
The interface towards platforms and drivers will change to use one of
CBMEM_CREATION_HOOK() or CBMEM_READY_HOOK(). Former will only be called
in the first stage with CBMEM available.
Change-Id: Ie24bf4e818ca69f539196c3a814f3c52d4103d7e
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63375
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The psp_notify_boot_done call is done at the entry of BS_PAYLOAD_BOOT,
so it's not guaranteed that the psp_set_spl_fuse call is done before the
psp_notify_boot_done call. Moving the psp_set_spl_fuse call makes sure
that it's done before the psp_notify_boot_done call. This also brings
the psp_set_spl_fuse call in line with the enable_secure_boot call that
sends the PSB fusing command to the PSP.
TEST=None
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Id76b462608c3d788cd90e73a64d18c8e8b89dbfd
Reviewed-on: https://review.coreboot.org/c/coreboot/+/64395
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The MP_IRQ flags can be used in the MP table and the ACPI MADT table.
Move them into acpi.h to avoid pulling in the full mpspec.h which is
only available on x86.
BUG=b:218874489, b:160595155
TEST=Build
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I4f1091b7629a6446fa399720b0270556a926401a
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63845
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add Platform Secure Boot (PSB) enablement via the PSP if it is not
already enabled. Upon receiving psb command, PSP will program PSB fuses
as long as BIOS signing key token is valid.
Refer AMD PSB user guide doc# 56654, Revision# 1.00, this document is
only available with NDA customers.
Change-Id: I30aac29a22a5800d5995a78c50fdecd660a3d4eb
Signed-off-by: Ritul Guru <ritul.bits@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60968
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Setting up postcar MTRRs is done when invd is already called so there
is no reason to do this in assembly anymore.
This also drops the custom code for Quark to set up MTRRs.
TESTED on foxconn/g41m and hermes/prodrive that MTRR are properly set
in postcar & ramstage.
Change-Id: I5ec10e84118197a04de0a5194336ef8bb049bba4
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/54299
Reviewed-by: Elyes Haouas <ehaouas@noos.fr>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The bootblock.elf file gets embedded in the BIOSPSP part and loaded by
the PSP in dram. The top aligned bootblock in cbfs is unused.
Tested on Cezanne/Guybrush.
Change-Id: I72f0092e0e3628b388f6da6a417c2857a510b187
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63226
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
This avoids compiler warnings on 64bit builds that complains about
casting pointer to non matching integer size.
Change-Id: I29fdb73ae1c0508796a21b650bf4fd1ac6688021
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63726
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Commit 198cc26e49 (soc/amd/common/block/
psp/psp_gen2: use SMN access to PSP) changed how the PSP registers are
accessed. Since the new method doesn't need to rely on a MMIO base
address to be configured, the read will always be successful and so
soc_read_c2p38 doesn't need to return an error status and can directly
return the value instead.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I1abace04668947ba3223a107461a27dddc0a9d83
Reviewed-on: https://review.coreboot.org/c/coreboot/+/64078
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: ritul guru <ritul.bits@gmail.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The eSPI decode range defines aren't and shouldn't be used directly from
outside of the common AMD eSPI code which provides functions to abstract
the register access, so move the defines from amdblocks/espi.h to
espi_def.h inside the common AMD LPC/eSPI support directory to limit the
visibility. The special I/O range decode bits need to stay in
amdblocks/espi.h since those are used in the devicetree. Also update the
indentation in espi_def.h so that the defines line up properly.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ic4ea30a1a6f10e94d88bf3b29f86dee2da6b39b5
Reviewed-on: https://review.coreboot.org/c/coreboot/+/64053
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Sabrina has more eSPI decode ranges than Picasso or Cezanne. Those
registers are however not in one block where it's easy to calculate the
addresses of a register from the index of the decode range. Within one
group of decode range registers it's still easy to calculate the
register address, so move the base address from within the macro to the
instantiation of the macro as a preparation for adding the support for
the additional ranges.
TEST=Timeless build results in identical binary for Mandolin
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Id309d955fa3558d660db37a2075240f938361e83
Reviewed-on: https://review.coreboot.org/c/coreboot/+/64052
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Since we can't rely on the MMIO base address in the PSP_ADDR_MSR MSR to
access the PSP mailbox registers, switch to using the SMN mapping of the
PSP mailbox registers. The PSP SMN base address is taken from the amdgpu
driver in the Linux kernel.
BUG=b:229779018
TEST=Mandolin still boots successfully and there are no errors/warnings
about possibly PSP-related things.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I9d17e523e9ae8d8e14ecedc37131a81f82351487
Reviewed-on: https://review.coreboot.org/c/coreboot/+/64034
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
CORE_2_PSP_MSG_38_OFFSET and CORE_2_PSP_MSG_38_FUSE_SPL are only used in
psp_gen2.c, so move them into this file.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I67cc2ff63d1c0322b514521975f3ce0f9b1cf5b1
Reviewed-on: https://review.coreboot.org/c/coreboot/+/64011
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
I find it difficult to constantly decode the registers when reading
them. Let's print out something that's easier to parse.
BUG=b:228289365
TEST=boot guybrush and see status codes printed
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I6c9d98cf43f340cf50e12c93b4c35187de9bb750
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63938
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Rob Barnes <robbarnes@google.com>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Silently failing makes it hard to debug when something goes wrong.
BUG=b:228289365
TEST=build guybrush
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I7423a7011e7656414155386c014a9a0f2fad4abf
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63937
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Drop struct pspv2_mbox and access the PSP mailbox via their offsets into
PSP MMIO region.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ib665d7ae19deae07d6a69c11ba8cf44e45ea4e70
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63966
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Don't use unnamed redefinitions of the pspv2_mbox_command union when the
union definition can be used instead.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I3757db45272f11bb47e5106ad9054c0a9ca0cd52
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63964
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
The pspv2_mbox struct contained an unnamed union that covered the 32
bits of the command register of the PSP v2 mailbox. Since the pspv2_mbox
struct is mainly used for hardware register accesses and the union part
is mostly used to access the different bits before/after writing/reading
the command register, split this functionality. For the register access
a command field is added to the pspv2_mbox struct instead of the unnamed
union and for accessing the separate bits of the command register a new
named union is added.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: If3f00b6fd73c3f749154b77b940e6d5aa385ec49
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63963
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The cmd_response field in the pspv2_mbox struct is the buffer used to
pass data to the PSP and back to the x86 side, so rename it to buffer.
This also aligns the code a bit more with the reference code. Also
rename the wr_mbox_cmd_resp function to wr_mbox_buffer_ptr.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I22c8971b07b3dedcc2e6e50e93c98d69ec7379e8
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63962
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The pspv[1,2]_mbox struct is only used in psp_gen[1,2].c, so move those
definitions from the common psp_def.h to the specific psp_gen[1,2].c
files. Also fix the struct name in the comment about pspv1_mbox.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I0c95e9a6e292b90e0d147c57f59828a9b41e4b82
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63960
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This bit is new on sabrina. We need to enable it after initialization
has completed.
BUG=b:227282870
TEST=Boot skyrim to OS and verify keyboard works
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I795275993589e20c1d09674232ecff782c491335
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63842
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Jon Murphy <jpmurphy@google.com>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Use BHD instead of BDT as the name of cookie macro. Use L2 to make it
clear it is for level 2. The 'BHD2' is misleading, which is going to
be used for combo entry. The definition in psp_verstage is also changed.
Change-Id: Ia10ac5e873dab6db7d66e63773a7c63f504950b2
Signed-off-by: Zheng Bao <fishbaozi@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/55411
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
The fch_i23c_pad_init implementation was written without looking at any
reference code and turned out to not work properly on hardware. Before
this function writes to the MISC_I23C_PAD_CTRL registers, the value read
back is 0x3000003c which results in the I2C bus communication to work
while the 0x300003fc the code writes to the register breaks the I2C
communication. Removing the code that sets bits 6..9 fixes the I2C bus
communication.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Tested-by: Karthikeyan Ramasubramanian <kramasub@google.com>
Change-Id: Ie6758b3d13c59b20ce810225fca8a365713b7a2b
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63234
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
When masking out bits with an and mask, the bit mask needs to be
inverted.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I9739d7150e230fbbe6523413de9c07d7340f3c61
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63222
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
I23C_PAD_CTRL_SLEW_N_SHIFT is 6 and not 7 which matches both with the
PPR #57243 revision 1.53 and with I23C_PAD_CTRL_SLEW_N_MASK which covers
both bits 6 and 7.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I622717bebaffe34b6df5e578b082dc10e2a98256
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63216
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
On x86 ramstage is always relocated at runtime in cbmem so there is no
need to have this configurable in Kconfig.
Change-Id: I01b2335d0b82bea8f885ee5ca9814351bbf2aa3c
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63215
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
SMU locks up sometimes if the port80 enable bit is cleared in the ESPI
Decode register. Add a config to choose between clearing the entire ESPI
Decode Register vs retaining the port80 enable bit.
BUG=None
TEST=Build and boot to OS in Skyrim.
Signed-off-by: Karthikeyan Ramasubramanian <kramasub@google.com>
Change-Id: Ia5ee012ac4858d6dd43827274169edf622a70489
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63118
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Fixes building with clang.
Change-Id: I7027f3681e18b8ca0d2f0c899412806082846463
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/63050
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
On boards where PSP uses ESPI to write postcodes, update the verstage to
do it after ESPI initialization.
BUG=b:224543620
TEST=Build and boot to OS in Nipperkin. Ensure that there are no
attempts to write the post code from PSP verstage before ESPI
initialization.
Change-Id: I1b78931c741c75dc845c9b34e3b2b896221f2364
Signed-off-by: Karthikeyan Ramasubramanian <kramasub@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62880
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Mohan Viswanathan
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Despite the SMBus device being function 0 of the FCH PCI device, the
MMIO resource of the FCH IOAPIC is on the LPC device which is function 3
of the same PCI device, so move the FCH IOAPIC initialization code to
the LPC device. Since the HPET was enabled in the same function, also
move it to the LPC device initialization.
TEST=On Mandolin both IOAPICs are still correctly detected by Linux.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I585afd463c1c00cd87ced0617e7802503c5deba5
Reviewed-on: https://review.coreboot.org/c/coreboot/+/58334
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The current SMM framework only allows the mainboard code to handle GPEs.
i.e., Events 0 - 23. This change allows the mainboard code to handle any
SMI events not handled by the SoC code. This will allow the mainboard
code to handle `SMITYPE_ESPI_SMI`.
BUG=b:222694093
TEST=Build guybrush
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I81943e8cb31e998f29cc60b565d3ca0a8dfe9cb2
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62598
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Now that SMM can write to CBMEM we can simply replay the transfer buffer
cbmem console to move it into the main cbmem console.
replay_transfer_buffer_cbmemc() relies on the EARLY_RAM linker symbols.
Since the SMM rmodule get linked with a different linker script than
bootblock/romstage it doesn't have access to these symbols. In order to
pass these symbols into SMM, we parse the bootblock.map file and
generate an early_ram.ld script. This script is then used when linking
SMM.
I replay the buffer in `smm_soc_early_init` because this call happens
before `console_init()`. `console_init()` prints the SMM header and we
want to append the verstage contents before printing the header to avoid
confusion.
BUG=b:221231786
TEST=Perform S0i3 cycles and verify PSP verstage logs now show up when
doing `cbmem -c`.
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I64d33ccdee9863270cfbcaef5d7c614349bd895c
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62402
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Rename SPIROM_BASE_ADDRESS_REGISTER to SPI_BASE_ADDRESS_REGISTER to
clarify that this isn't the address the SPI flash gets mapped, but the
address of the SPI controller MMIO region. This also aligns the register
name with the PPR.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ifd9f98bd01b1c7197b80d642a45657c97f708bcd
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62578
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Jason Glenesk <jason.glenesk@gmail.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
This patch aims to make timestamps more consistent in naming,
to follow one pattern. Until now there were many naming patterns:
- TS_START_*/TS_END_*
- TS_BEFORE_*/TS_AFTER_*
- TS_*_START/TS_*_END
This change also aims to indicate, that these timestamps can be used
to create time-ranges, e.g. from TS_BOOTBLOCK_START to TS_BOOTBLOCK_END.
Signed-off-by: Jakub Czapiga <jacz@semihalf.com>
Change-Id: I533e32392224d9b67c37e6a67987b09bf1cf51c6
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62019
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Shorten define names containing PCI_{DEVICE,VENDOR}_ID_ with
PCI_{DID,VID}_ using the commands below, which also take care of some
spacing issues. An additional clean up of pci_ids.h is done in
CB:61531.
Used commands:
* find -type f -exec sed -i 's/PCI_\([DV]\)\(EVICE\|ENDOR\)_ID_\([_0-9A-Za-z]\{2\}\([_0-9A-Za-z]\{8\}\)*[_0-9A-Za-z]\{0,5\}\)\t/PCI_\1ID_\3\t\t/g'
* find -type f -exec sed -i 's/PCI_\([DV]\)\(EVICE\|ENDOR\)_ID_\([_0-9A-Za-z]*\)/PCI_\1ID_\3/g'
Change-Id: If9027700f53b6d0d3964c26a41a1f9b8f62be178
Signed-off-by: Felix Singer <felixsinger@posteo.net>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/39331
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Michael Niewöhner <foss@mniewoehner.de>
Since we want to read the non-x86 CBMEMC from SMM we need to be stricter
on where we read from. This change forces the verstage binary and x86
code to agree on the CBMEMC transfer buffer location and size.
BUG=b:221231786
TEST=Boot guybrush and verify verstage transfer buffer still ends up in
cbmem
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: Ida7d50bef46f280be0db1e1f185b46abb0ae5c8f
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62501
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
We don't need to force the caller to look up and cast the transfer
region. We can do it in the function.
BUG=b:221231786
TEST=Build guybrush
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: Ib46a673ef5a43deb56a6d522152085036a47ab66
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62401
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
I want to reuse the transfer buffer methods in SMM, so I need to add
them into their own file. I renamed `setup_cbmem_console` to
`replay_transfer_buffer_cbmemc` so it has a more descriptive name. I
also fixed the comment on `verify_psp_transfer_buf`.
BUG=b:221231786
TEST=Boot guybrush to OS
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I4f3a8b414b91f601c3a9c3dc7af8f388286fe4da
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62348
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
We need to save the transfer buffer so we can transfer the cbmem
console and timestamps into x86 DRAM.
BUG=b:221231786
TEST=Boot guybrush and verify S0i3 resume works. Also dumped the
transfer buffer from the OS and verified the console contents got
transferred.
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I1d3b34c90e0e18609b0c6a0cdedab35aeefbd84b
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62347
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
If an AMD FSP binary has no valid image revision information, print a
warning instead of an error.
Change-Id: Ie9c5a387b81205fe93382778090260e41e261776
Signed-off-by: Julian Schroeder <julianmarcusschroeder@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62349
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Jason Glenesk <jason.glenesk@gmail.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
We print these out in the normal flow, so lets add them for S0i3 resume
as well.
BUG=b:221231786
TEST=Perform suspend/resume cycle on guybrush and verify we get the new
POST codes.
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: Ia7d607453d58084868cfa50770fd0f370b2ea2bd
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62346
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Rob Barnes <robbarnes@google.com>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
This feature was never used. Let's remove it to keep things simple.
BUG=221231786
TEST=Boot test guybrush and morphius and verify transfer buffer is
correctly passed.
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I93a284db919f82763dcd31cec76af4b773eb3f80
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62345
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Rob Barnes <robbarnes@google.com>
Using __fallthrough instead of a comment about the fall-through being
intentional should make clang stop complaining about intended fall-
through statements.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I940529be02e20c72f6e97b2cfa10f0dd8f7020b6
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62216
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-by: Jason Glenesk <jason.glenesk@gmail.com>
Compiling efs_fmap_check.c depends on fmap_config.h already being
generated, so add this dependency.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I85e0900574f928d1594f8d1831ba58f959b75d27
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62132
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Also add the Makefile dependency on the fmap_config.h file to make sure
that this file already exists when it's included.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I540ea2c14fd187845efd3c0c8c1e4b8f82c8cac3
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62130
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
There are platforms equipped with AMD SoC where I2C3 controller
connected to TPM device is shared between X86 and PSP. In order to
handle this, PSP acts as an I2C-arbitrator, where x86 (kernel) sends
acquire and release requests to be accepted by PSP. An example of
implementation within Linux kernel is available [1].
There is a need to introduce new ACPI_ID ("AMDI0019") so that dedicated
driver on OS side can bind to it and handle this special setup. Since
PSP takes care of I2C controller power management, we need to remove
PowerResource object from DSDT.
BUG=b:204508404
BRANCH=guybrush
[1]: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?id=78d5e9e299e31bc2deaaa94a45bf8ea024f27e8c
Signed-off-by: Jan Dabros <jsd@semihalf.com>
Change-Id: Iccfc09d8c580d7ab2acb69d26b9c293cf625fb34
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61863
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The and-mask passed to the gpio_update32 call needs all 32 bits to be
set to ones. When building as 32 bit binary the -1UL will result in the
needed bit mask, but for a 64 bit build the constant would have 64 bits
set to ones which then gets truncated to 32 bits causing a compiler
error. Use 0xffffffff as bit mask instead which behaves correctly in
both cases and also clarifies what this is doing.
TEST=Timeless build for Chausie results in identical image.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I0b6a50bd914fdbb7a78885efb6c610715e2d26c1
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62053
Reviewed-by: Elyes Haouas <ehaouas@noos.fr>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Aamir Bohra <aamirbohra@gmail.com>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-by: Jason Glenesk <jason.glenesk@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This fixes a build failure when trying to build the code in 64 bit mode.
TEST=Timeless build for Chausie results in identical image.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: If8fe7b626d9d72c0b8ed07ced93e46f795e36848
Reviewed-on: https://review.coreboot.org/c/coreboot/+/62052
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aamir Bohra <aamirbohra@gmail.com>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-by: Jason Glenesk <jason.glenesk@gmail.com>
Reviewed-by: Elyes Haouas <ehaouas@noos.fr>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Add PSP command to send SPL fuse command if PSP indicates SPL fusing
is required. Also add Kconfig option to enable sending message.
BUG=b:180701885
TEST=On a platform that supports SPL fusing. Build an image with an SPL
table indicating fusing is required, confirm that PSP indicates fusing
required and coreboot sends the appropriate command. A message indicating
PSP requested fusing will appear in the log: "PSP: Fuse SPL requested"
Signed-off-by: Jason Glenesk <jason.glenesk@amd.corp-partner.google.com>
Change-Id: If0575356a7c6172e2e0f2eaf9d1a6706468fe92d
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61462
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-by: ritul guru <ritul.bits@gmail.com>
The gen2 ACP register definitions and locations are different from
previous models. Specific code is refactored into acp_gen1 and acp_gen2.
Update ACP register locations and definitions for gen2.
Change-Id: If665b93cddf22435512f1276fcfee2f497dc6ef5
Signed-off-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61832
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Since the APIC bus isn't used since a long time and the IOAPIC and LAPIC
talk to each other via the system bus, there is no longer the
requirement that the IOAPIC IDs mustn't overlap with the LAPIC IDs that
start at 0 and end at CONFIG_MAX_CPUS - 1. The current Intel code uses 2
as the IOAPIC ID while most of their CPUs have more than 2 logical cores
resulting in the IOAPIC having the same ID as one of the LAPICs.
All chipsets in soc/amd use the defines for FCH_IOAPIC_ID and
GNB_IOAPIC_ID for initializing the IOAPIC register, writing both MADT
and IVRS ACPI tables and there's no MPTable support for those SoCs that
might also rely on those IDs being consistent.
This patch changes the definitions for FCH_IOAPIC_ID and GNB_IOAPIC_ID
from CONFIG_MAX_CPUS and CONFIG_MAX_CPUS + 1 to 0 and 1. This also makes
sure that the IOAPIC IDs still fit in 4 bits despite Cezanne having a
CONFIG_MAX_CPUS of 16 resulting in the IOAPIC IDs being larger than 4
bits with the old code. While the Cezanne FCH IOAPIC supports 8 bits of
IOAPIC IDs, this is non-standard.
TEST=AMD Mandolin and Google Liara still work.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Suggested-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Change-Id: Id3a356480bb8407e0347cb5cef691fde7edc8deb
Reviewed-on: https://review.coreboot.org/c/coreboot/+/55430
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Refactor existing acp code into acp_gen1 variant as preparation for gen2
variant in sabrina.
Change-Id: Id9248584237196b5404b79d3a8552cb90fe4491e
Signed-off-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61831
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Check if FSP binary and coreboot FSP structures (fspmupd.h) match
sufficiently.
A change in minor number denotes less critical changes or additions
to the FSP API that still allow for the boot process to proceed.
A change of the AMD image revision major number will halt boot.
The Fspmupd.h header now defines IMAGE_REVISION_ macros for AMD
Picasso, Cezanne and Sabrina APUs.
BUG=b:184650244
TEST=build, boot and check fsp image revision info. Example:
FSP major = 1
FSP minor = 0
FSP revision = 5
FSP build = 0
Signed-off-by: Julian Schroeder <julianmarcusschroeder@gmail.com>
Change-Id: I0fbf9413b0cf3e6093ee9c61ff692ff78ebefebc
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61281
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Introduce and use functions to translate eSPI IO/MMIO decode range IDs
into the corresponding register bits and the IO/MMIO range and size
register IDs into register offsets. This is a preparation to support the
additional eSPI decode ranges on Sabrina where not all enable bits and
base/size registers for one type of decode ranges are consecutive.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Id91fe32447a06b049e33dfdacc8edfa2ebb2df39
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61778
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
This aligns the register names more with the PPR.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I4e7dc8dfc0fa5e86b9d4425f2496be86e039b686
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61777
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
On systems that use the first 128kByte of the SPI flash for the EC
firmware, it is not possible to place the EFS/amdfw part at the lowest
location in flash where the on-chip PSP firmware will look for the EFS,
since this is at an offset of 128kByte into the flash which is where the
cbfs master header resides when the main CBFS is placed right after the
EC firmware. This patch introduces the AMD_SOC_SEPARATE_EFS_SECTION
option that allows putting the EFS in a separate FMAP section that can
be located right after the EC firmware FMAP section. The EFS FMAP
partition is checked to ensure it begins at the expected location.
Change-Id: I5ed0f76c9c9c9c180ee5f1b96f88689d0979bb5e
Signed-off-by: Fred Reitberger <reitbergerfred@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61558
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
This will allow PSP verstage to write logs to the serial console. We
are no longer dependent on using a serial enabled PSP boot loader.
Ideally we would delete this psp printk and use the standard printk.
Since picasso doesn't currently support mapping the UART though, I'll
keep it for now.
BUG=b:215599230
TEST=Boot guybrush and verify PSP logs are output on serial console
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: Ibd77cc754fae5baccebe7adc5ae0790c79236d26
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61610
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Now that the console system itself will clearly differentiate loglevels,
it is no longer necessary to explicitly add "ERROR: " in front of every
BIOS_ERR message to help it stand out more (and allow automated tooling
to grep for it). Removing all these extra .rodata characters should save
us a nice little amount of binary size.
This patch was created by running
find src/ -type f -exec perl -0777 -pi -e 's/printk\(\s*BIOS_ERR,\s*"ERROR: /printk\(BIOS_ERR, "/gi' '{}' ';'
and doing some cursory review/cleanup on the result. Then doing the same
thing for BIOS_WARN with
's/printk\(\s*BIOS_WARNING,\s*"WARN(ING)?: /printk\(BIOS_WARNING, "/gi'
Signed-off-by: Julius Werner <jwerner@chromium.org>
Change-Id: I3d0573acb23d2df53db6813cb1a5fc31b5357db8
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61309
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: HAOUAS Elyes <ehaouas@noos.fr>
Reviewed-by: Lance Zhao
Reviewed-by: Jason Glenesk <jason.glenesk@gmail.com>
This comment was added with the AMD family 15h Trinity IOMMU support in
commit 88ebbeb7e2 and looks like a copy of
the comment about the subtractive decode ranges in the LPC device. The
IOMMU doesn't have any subtractively decoded I/O or MMIO ranges and this
is also not what the code does. This resource is the MMIO region to
configure the IOMMU instead, so fix the comment in all copies of the
IOMMU support code.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I2e1e3a46b839b9e58b836932c1bc9b41b1b1dc02
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61631
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The PM2 ACPIMMIO region should only be accessed with 8 bit accesses.
Using 16 or 32 bit read accesses will return the data from the first
byte for all 2 or 4 bytes and 16 or 32 bit write accesses will result in
only the first byte being written which is both unexpected behavior.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I5ace50d3b81b5bf3ea3b10aa02f25c58a6ea99b9
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61625
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Bit 23 in the PM_RST_STATUS register is called LtReset on Stoneyridge
and ShutdownMsg on Picasso/Cezanne/Sabrina. Bit 30 is reserved on
Stoneyridge and defined as SdpParityErr on the newer SoCs. Bit 31 is
only defined for Sabrina. Since the default value of undefined bits is 0
it isn't a problem to have descriptions for reserved reset status bits
on some SoCs.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I0782116d327fcad3817a10eb237ac6c8294846b3
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61624
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
I2C bus 0..2 on Sabrina uses a different pad type which supports 1.1V
and 1.8V levels, but doesn't support 3.3V I2C levels. Compared to the
existing I2C pad control registers the bit definitions are different, so
add a separate function to configure those pads which however still has
the same function signature and is compatible with same data structs
used for the devicetree settings. PPR #57243 Rev 1.50 was used as a
reference.
TEST=None
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ie210c3437f2608d1e9fb99dcb151fc4190721375
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61570
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The I2C pad control registers of Picasso and Cezanne are identical and
the one of Sabrina is a superset of it, so factor out the functionality.
To avoid having devicetree settings that contain raw register bits, the
i2c_pad_control struct is introduced and used. The old Picasso code for
this had the RX level hard-coded for 3.3V I2C interfaces, so keep it
this way in this patch but add a TODO for future improvements.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I1d70329644b68be3c4a1602f748e09db20cf6de1
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61568
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Add platform_report_mode function which report current developer mode
status to the PSP. L1 widevine app in the PSP will use this information
to select key box.
BUG=b:211058864
TEST=build and boot guybrush
TEST=build picasso chrome os boards
Signed-off-by: Kangheui Won <khwon@chromium.org>
Change-Id: I04b5fcfa338b485b36f1b946203f32823385c0b1
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61369
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Jason Glenesk <jason.glenesk@gmail.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Using enum cb_err as return type instead of int improves the readability
of the code.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I55e6d93ca141b687871ceaa763bbbbe966c4b4a3
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61511
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Jason Glenesk <jason.glenesk@gmail.com>
This code is common to at least all Zen-based APUs (Picasso, Cezanne,
Sabrina) and is also useful outside of the SoC-specific dynamic ACPI
table generation code.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ie96d4429fb6ed9223efed9b3c754e04052d7ca7c
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61357
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Rob Barnes <robbarnes@google.com>
Reviewed-by: Chris Wang <chris.wang@amd.corp-partner.google.com>
Reviewed-by: Eric Peers <epeers@google.com>
The VERSTAGE region is only needed when running verstage in the x86.
This change reduces the early ram size by 512 KiB when using PSP
verstage.
BUG=none
TEST=Boot guybrush to OS
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I45ce421397807dbb1eb48aedd05209b91e89aa4f
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61190
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Document #55758 Rev. 1.13 says that family 17h models 30h-3Fh and later
use the spi_readmode_f17_mod_30_3f struct element for SPI_MODE_FIELD and
spi_fastspeed_f17_mod_30_3f for SPI_SPEED_FIELD, so also use this for
The AMD Sabrina SoC which is family 17h models A0h-AFh.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I336f9ea4a0defdf34e1af4b6d568cfe46488f75e
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61078
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
This change splits the size of the console transfer region and size of
the bootblock/romstage Pre-RAM console region. This allows having a
larger Pre-RAM console while not impacting the size of the PSP verstage
console.
Instead of directly using the PRE_X86_CBMEM_CONSOLE_SIZE symbol in
`setup_cbmem_console`, I chose to use the offsets provided in the
transfer buffer. It would be nice to eventually do this for all the
fields in the transfer buffer.
BUG=b:213828947
TEST=Boot guybrush and verify verstage logs are no longer truncated
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I8b8cc46600192a7db00f5c1f24c3c8304c4db31d
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61189
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Kangheui Won <khwon@chromium.org>
This implementation is the same for all SoC that select
SOC_AMD_COMMON_BLOCK_NONCAR, so factor it out to the common AMD non-CAR
CPU support code folder.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I53528f0bb75e9d945740ad5065c75e7de7b5878f
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61257
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This will verify that signed verstage binaries and the bootblock code
executing agree on the transfer buffer struct size.
BUG=b:213828947
TEST=Boot guybrush to OS
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I597e38fe0a37416ffd3bc01fd974fa8f6610a88c
Reviewed-on: https://review.coreboot.org/c/coreboot/+/61187
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Kangheui Won <khwon@chromium.org>
The existing common AMD SoC code supports some of AMD Family 17h Model
A0h SoC's PCI devices that however have different PCI IDs. Add the new
PCI ID defines to the PCI ID lists of the common PCI drivers.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I50960e502c63a2ffcfed35178c5e7c9729ef061e
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60985
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Define the register offsets and bits in a separate header file instead
of in the middle of the .c file.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I814192b2dfeff05877ac857dd89e8cdc7ae5ee25
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60770
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
In espi_wait_channel_ready the return value of espi_get_configuration
didn't get checked before. In the case of the espi_send_command call in
espi_get_configuration returning CB_ERR, espi_get_configuration didn't
write to the local config variable, so if this happens in the first pass
of the do-while loop, the following espi_slave_is_channel_ready call
would use the uninitialized local config variable as parameter. Fix this
by checking the return value of espi_get_configuration.
TEST=None
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Iff1a0670e17b9d6c6f4daf2ea56badf6c428b8c9
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60209
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Use enum cb_err as return type of all remaining functions that only
return success or failure.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I6cff8480d99641fdfb613bb3e4edc4055ad5efc6
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60208
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Use enum cb_err as return type of all functions that aren't exposed
outside of this compilation unit. The checks if a function has returned
a failure are replaced with checks if the return value isn't CB_SUCCESS
which is equivalent if only those two values are used, but also detects
a failure if any unexpected value would be returned.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: If8c703f62babac31948d0878e91bd31b31bebc01
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60207
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
The intermediate ret variable isn't needed. espi_open_generic_io_window
only returns 0 or -1, so if ret is != 0, it has to be -1. This is a
preparation to use the enum cb_err type for the return values.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ia6c7f4cedf8c2defadcf4c4da1697a97c7b401f2
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60206
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Both the Picasso PPR #55570 Rev 3.18 and the Cezanne PPR #56569 Rev 3.03
define bit 9 of the PM_RST_STATUS register as internal Thermal Trip
reset status bit.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ida8b13fe62b16c18fc9924520b83220e73eca624
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60184
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Jason Glenesk <jason.glenesk@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
The MBOX_BIOS_CMD_DRAM_INFO PSP mailbox command is only available on the
first generation of PSP mailbox interface and not on the second
generation. The second generation of the PSP mailbox interface was
introduced with the AMD family 17h SoCs on which the DRAM is already
initialized before the x86 cores are released from reset.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I97b29fdc4a71d6493ec63fa60f580778f026ec0b
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60124
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Replace FCH_SC with FCH SPI in the printk messages to make those a bit
clearer and also remove an unneeded line break in another printk call.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I6ff02163e6a48a2cc8b7fe89b15826e154715d29
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60120
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
When wait_for_ready returned a timeout, execute_command still ended up
returning success. Fix this be returning a failure in this case.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Id012e74e26065c12d003793322dcdd448df758b0
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60119
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Introduce and use enum spi_dump_state_phase to indicate from which phase
of the SPI transfer dump_state gets called to print the relevant debug
information for that phase.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I2f54d4a7eb2f3b9756b77a01533f7c99e8597bfa
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60118
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The Cezanne PPR #56569 Rev 3.03 has one more SPI FIFO bytes defined
compared to the previous generations. It is unclear if adding some
special handling for Cezanne would be worth the effort, since the
current code just doesn't use the last byte which should be safe to do,
since this only affects the maximum number of bytes that can be used for
one SPI transaction. Having another byte to use on Cezanne wouldn't
reduce the number of SPI transactions to write a 256 byte data block.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ic730f4fe838f59066120c811833995c132c84c1c
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60117
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The last byte of the SPI FIFO SPI_FIFO_LAST_BYTE is at offset 0xc6 of
the SPI controller's MMIO region for Stoneyridge and Picasso. Both
SPI_FIFO_LAST_BYTE and SPI_FIFO_DEPTH had an off-by-one error that ended
up cancelling out each other, so the resulting value for SPI_FIFO_DEPTH
isn't changed.
TEST=Timeless build results in identical image for Mandolin.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I1676be902ccf57e2e9f69d81251b4315866a0628
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60116
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Add PSP_S0I3_RESUME_VERSTAGE Kconfig option. When enabled, verstage will
be run in PSP during S0i3 resume. Setting softfuse bit 40 enables this
in PSP.
BUG=b:200578885, b:202397678
BRANCH=None
TEST=Verstage runs during s0i3 resume on Nipperkin
Change-Id: I2c185f787c1e77bd09f6cbbb1f47deb665ed0c79
Signed-off-by: Rob Barnes <robbarnes@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/60024
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The PSP EFS code to get the SPI mode and speed from the amdfw part of
the firmware image also works for Stoneyridge which is the one SoC that
selects SOC_AMD_COMMON_BLOCK_PSP_GEN1. Also amdblocks/psp_efs.h already
handles the SOC_AMD_STONEYRIDGE case.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ibddd3f9237e561d9f0f6b4ad70f59cce1f956986
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59966
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
According to https://uefi.org/specs/ACPI/6.4/04_ACPI_Hardware_Specification/ACPI_Hardware_Specification.html#pm1-event-grouping
> For ACPI/legacy systems, when transitioning from the legacy to the G0
> working state this register is cleared by platform firmware prior to
> setting the SCI_EN bit (and thus passing control to OSPM). For ACPI
> only platforms (where SCI_EN is always set), when transitioning from
> either the mechanical off (G3) or soft-off state to the G0 working
> state this register is cleared prior to entering the G0 working state.
This means we don't want to clear the PM1 register on resume. By
clearing it the linux kernel can't correctly increment the wake count
when the power button is pressed. The AMD platforms implement the _SWS
ACPI methods, but the linux kernel doesn't actually use these methods.
BUG=b:172021431
TEST=suspend zork and push power button and verify power button
wake_count increments. Verified other wake sources still work.
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: Iaa886540d90f4751d14837c1485ef50ceca48561
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59929
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Use a read modify write sequence when setting the SPI_USE_SPI100 bit in
the SPI100_ENABLE register. This avoids clearing other bits in the
register which might cause instabilities of the SPI interface. The
reference code for both Picasso and Cezanne also only sets the
SPI_USE_SPI100 bit and doesn't zero out the other bits.
TEST=Verified that Mandolin still boots. It didn't show any signs of
possibly related instabilities before though, so this test doesn't say
much.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I71c2ec1729d5cb4cdff6444b637af29caaa6f1c0
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59932
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
commit 90ac882a32 (soc/amd/common/block/
spi: introduce SOC_AMD_COMMON_BLOCK_SPI_4DW_BURST) introduced a Kconfig
option to enable/disable the 4DW burst support in the SPI flash data
prefetcher, but missed to update the documentation above the
fch_spi_early_init prototype, so update the outdated documentation now.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I07c4b0b02251da63d34a172e2636894e99845d6b
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59931
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
These functions can't be weak, because they actually need to configure
the GPIOs for eSPI and the TPM. With this change zork boots again.
I also noticed that zork doesn't use the early table in bootblock. This
means that zork will only boot if psp_verstage is enabled.
BUG=b:209465425
TEST=boot zork to ramstage
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Change-Id: I384fd578efe7da0a3d74829cccf38c3ed524f130
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59922
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Rob Barnes <robbarnes@google.com>
Use KiB and MiB instead of multiplying/dividing with/by the numeric
value when doing region size calculations.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I56c380190b11aa3214cce31b82974327e3d15000
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59936
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Despite Stoneyridge being one only SoC in soc/amd that uses the first
generation of the PSP mailblox interface, this code is common for all
SoCs that use the first PSP mailbox interface generation, so move it to
the common PSP generation 1 code.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I78126cb710a6ee674b58b35c8294685a5965ecd6
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59701
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Both SPI_ROM_BIOS_SEMAPHORE and SPI_ROM_EC_SEMAPHORE bits in the
LPC_PCI_CONTROL are defined in the Stoneyridge BKDG #55072 Rev 3.04,
Raven1 and Picasso PPR #55570 Rev 3.18, Raven2 PPR #55772 Rev 3.08 and
Cezanne PPR #56569 Rev 3.03 which are all platforms that use this code.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I855e640d020daf21c9f5b2f62a2ad0fd0274a575
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59674
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
When using 32 bit PCI accesses in lpc_enable_port80, we can use the
LPC_IO_OR_MEM_DECODE_ENABLE and DECODE_IO_PORT_ENABLE4 defines and don't
need to re-define bits with offsets from the beginning of the third byte
within this 32 bit register. This allows to drop the
LPC_IO_OR_MEM_DEC_EN_HIGH register definition which points to
LPC_IO_OR_MEM_DECODE_ENABLE + 2 and to drop the re-definitions of the
bit re-definitions with a different offset.
The code in lpc_enable_port80 was originally copied from sb/amd/agesa/
hudson/early_setup.c which might be sort-of a copy from what the AGESA
reference code does.
TEST=When commenting out SOC_AMD_COMMON_BLOCK_USE_ESPI in the Kconfig of
Mandolin and selecting AMD_LPC_DEBUG_CARD, all POST codes still get
shown on the POST code LED display when this patch is applied.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I001bb1c2ccf99e36d4fbd73d3bf96b78ddb87d67
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59676
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
This function is unused and none of the SoCs using this code has a
physical PCI interface any more, so drop this function.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ia5c5a8ec29264a075fefe75038ef2a84684d6427
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59675
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
PSPV2_STATUS_ERROR and PSPV2_STATUS_RECOVERY aren't used and the bit
definitions are also wrong, so drop those defines. For the PSP mailbox
interface version 2, struct pspv2_mbox is used to access the correct
status bits.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I8e2aadfde00e2f7b0f99b462b8e3d6954959a584
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59699
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Adding the DP_ prefix to the defines for MMIO_NP, MMIO_WE and MMIO_RE
clarifies the scope of those definitions. For consistency also add this
prefix to MMIO_DST_FABRIC_ID_SHIFT.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I3a509ccc071aa51a67552fb9e7195358a76fe4dc
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59627
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
On Picasso the MMIO_NP bit in the D18F0_MMIO_CTRL0 data fabric register
is bit 12, but that has changed to bit 16 in Cezanne.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I64c06b84e2c0737b259077e7932f418306638e19
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59626
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
The bit is called REF_CLK_OK_STATE and not RST_CLK_OK_STATE, so change
the name of the define to FCH_AOAC_REF_CLK_OK_STATE.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Iae26db94d83ebb2cb799f6d3e0bec37c8e849219
Reviewed-on: https://review.coreboot.org/c/coreboot/+/59628
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
