Pass the PSP NVRAM base and size to amdfwtool for all SoCs except Genoa
and Stoneyridge which don't use/support this.
If a mainboard has an section named 'PSP_RPMC_NVRAM' in its FMAP file,
the start and length of it in the flash will be passed to amdfwtool
which then adds the base and length to the corresponding type 0x54 PSP
directory table entry.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Id9f8a7eec68a5222be63e46173132f1c4a461b4f
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83815
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Pass the PSP NVRAM base and size to amdfwtool for all SoCs except Genoa
which doesn't use/support this. This was previously only implemented for
Picasso, but not for the SoCs that support this, so add the support to
those other SoCs as well.
If a mainboard has an section named 'PSP_NVRAM' in its FMAP file, the
start and length of it in the flash will be passed to amdfwtool which
then adds the base and length to the corresponding type 0x04 PSP
directory table entry.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I785ede8eb0df2473a4390b2c305add20f38d7ede
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83814
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Move PSP_NVRAM_BASE and PSP_NVRAM_SIZE from the BIOS directory table
items to the PSP Directory Table items, since the corresponding region
will be referenced by the PSP directory table and not the BIOS directory
table.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Iff7568ea05c701ecd346cc7590cf93b091ff31a2
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83813
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Before 'handle_psp_command' calls any of the functions in this file, it
make sure that the 'size' field in the command buffer's header doesn't
indicate that the command buffer is larger than the SMM memory region
reserved for it.
The read/write command buffer has a 'num_bytes' field to indicate how
many bytes should be read from the SPI flash and put into the data
buffer within the command buffer or how many bytes from this buffer
should be written to the flash. While we should be able to assume that
the PSP won't send us malformed command buffer, we should still better
check this just to be sure.
Test=When selecting SOC_AMD_COMMON_BLOCK_PSP_SMI, Mandolin still builds
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ib4e8514eedc3ad154a705c8a1e85d367e452dbed
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83778
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Use coreboot's SPI flash access infrastructure to do the flash read,
write, or erase operations as requested from the PSP.
This patch is a modified version of parts of CB:65523.
Document #55758 Rev. 2.04 was used as a reference.
Test=When selecting SOC_AMD_COMMON_BLOCK_PSP_SMI, Mandolin still builds
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Signed-off-by: Ritul Guru <ritul.bits@gmail.com>
Change-Id: I4957a6d316015cc7037acf52facb6cc69188d446
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83777
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Detect the block size of the SPI flash and number of flash blocks
reserved for the flash region corresponding to the 'target_nv_id' field
in the command buffer. This information is then written to the
corresponding fields in the command buffer. Since detecting the flash
chip still might result in accesses to it, make sure that it's available
for use and not currently used by an OS driver. Since this code is
inside the SMI handler, we don't have to worry about this code to be
interrupted, so we don't need to set some bit to tell other code that
we're currently using the SPI controller in the SMI handler.
This patch is a modified version of parts of CB:65523.
Document #55758 Rev. 2.04 was used as a reference.
Test=When selecting SOC_AMD_COMMON_BLOCK_PSP_SMI, Mandolin still builds
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Signed-off-by: Ritul Guru <ritul.bits@gmail.com>
Change-Id: I19041a27a9e8f901d42c3f60af834df625455ea6
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83776
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The SPI_SEMAPHORE_DRIVER_LOCKED bit in the SPI_MISC_CNTRL register
doesn't affect the hardware, but it re-used by AMD as a semaphore to
synchronize the access to the SPI controller between SMM and non-SMM
software like an OS-level driver. Since it doesn't affect the hardware,
it's marked as reserved in the PPRs. Add the 'spi_controller_available'
helper function to check this bit to see if some software or driver
outside of SMM is currently using the SPI flash controller to avoid
interfering with that operation.
This patch is a slightly reworked version of parts of CB:65523.
Test=When selecting SOC_AMD_COMMON_BLOCK_PSP_SMI, Mandolin still builds
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Signed-off-by: Ritul Guru <ritul.bits@gmail.com>
Change-Id: I49218e03a5dd555b2b2d34eaad86673e9fc908c3
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83775
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add 'find_psp_spi_flash_device_region' to get a pointer to the spi_flash
struct of the SPI flash used in the system and the region_device struct
for the target FMAP region specified by the target NV ID from the PSP
to x86 mailbox command. In order to have small patches, the newly added
static 'find_psp_spi_flash_device_region' function is marked as inline;
that inline will be removed in a following patch that calls this new
function.
This patch is a slightly reworked version of parts of CB:65523.
Document #55758 Rev. 2.04 was used as a reference.
Test=When selecting SOC_AMD_COMMON_BLOCK_PSP_SMI, Mandolin still builds
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Signed-off-by: Ritul Guru <ritul.bits@gmail.com>
Change-Id: I64b8fba2392de46ecd4c786cef0d5b6acdbd865a
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83774
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add and use functions to validate the target non-volatile storage ID in
the different command buffer structs.
This patch is a slightly reworked version of parts of CB:65523.
Document #55758 Rev. 2.04 was used as a reference.
Test=When selecting SOC_AMD_COMMON_BLOCK_PSP_SMI, Mandolin still builds
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Signed-off-by: Ritul Guru <ritul.bits@gmail.com>
Change-Id: Idda0166c862d41d380b2ed21345eead5e0a1c135
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83758
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This patch is a slightly modified version of parts of CB:65523.
Document #55758 Rev. 2.04 was used as a reference.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Signed-off-by: Ritul Guru <ritul.bits@gmail.com>
Change-Id: I41efeecf9243ddbbd8dc3f842c5ce11058bb7999
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83757
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add stub functions for the SPI flash access from the PSP SMI handler
and call them for the corresponding P2C mailbox commands.
Parts of this patch are taken from CB:65523.
Document #55758 Rev. 2.04 was used as a reference.
Test=When selecting SOC_AMD_COMMON_BLOCK_PSP_SMI, Mandolin still builds
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Signed-off-by: Ritul Guru <ritul.bits@gmail.com>
Change-Id: Iedbc9d41eb0d4e8d81eeba9c01281161eb839991
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83756
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
When the PSP wants to access the SPI flash during runtime, but isn't the
owner of the SPI flash controller, it sends an SMI to the x86 side. The
corresponding SMI handler then checks the P2C (PSP to core) mailbox for
the command and data, processes the command, and if needed puts the
requested data into the P2C buffer.
The P2C mailbox is a memory region in TSEG aka SMM memory. Both location
and size are communicated to the PSP via the PSP SMM info mailbox
command which is sent right after mpinit is done.
This commit adds the code to access the P2C mailbox to the PSP SMI
handler code, but the handling of the actual mailbox commands the PSP
sends to the SMI handler is added in later patches to keep the patch
size manageable.
This patch is a heavily reworked version of parts of CB:65523.
Document #55758 Rev. 2.04 was used as a reference.
Test=When selecting SOC_AMD_COMMON_BLOCK_PSP_SMI, Mandolin still builds
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Signed-off-by: Ritul Guru <ritul.bits@gmail.com>
Change-Id: I50479bed2332addae652026c6818460eeb6403af
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83740
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
This register is currently used by the SPI DMA code that sets an
undocumented bit. A later patch will add and use some other bit in this
register.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I48447dcfb3cee07619a9b42434731f0b21458021
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83773
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The PSP can send SMIs to the x86 side to have the SMI handler service
requests from the PSP. This commit adds an empty PSP SMI handler; the
actual implementation is added in later patches to keep the patches
relatively small.
This patch is a slightly modified version of parts of CB:65523.
Test=When selecting SOC_AMD_COMMON_BLOCK_PSP_SMI, Mandolin still builds
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Signed-off-by: Ritul Guru <ritul.bits@gmail.com>
Change-Id: I65989ff529d728cd9d2cd60b384295417bef77ad
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83739
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The PSP can send SMIs to the x86 side of the system. Add helper
functions to configure and to reset the PSP SMI generation. Since
Stoneyridge also selects SOC_AMD_COMMON_BLOCK_SMI, add the SMITRIG0_PSP
define and rename SMITYPE_FCH_FAKE0 to SMITYPE_PSP in its SoC-specific
smi.h to bring it in line with the newer SoCs.
This patch is split out from CB:65523.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Signed-off-by: Ritul Guru <ritul.bits@gmail.com>
Change-Id: I525a447c9a75fdb95b9750e85a02896056315edf
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83702
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Since the P2C_BUFFER_MAXSIZE value will be needed in another compilation
unit, move the define to the common psp_def.h. P2C_BUFFER_MAXSIZE is
moved there too for consistency reasons.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I8d4d93760c90ad6e0ecadf70600b1d697a02fa82
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83701
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
When sending mailbox commands to the PSP from SMM, the SMM flag needs to
be set right before sending the mailbox command and cleared right after
the command is sent. In order to not have this code duplicated, factor
it out into a function.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I3628463dece9d11703d5a068fe7c604108b69c1f
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83700
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
The reasoning behind this and the positive side effects of this aren't
too clear from the code, so point those out in a comment.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I4f4121031fc1ef600cdf5551f61f1ef4e03b56a5
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83699
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Since it's not exactly obvious what 'c2p_buffer', 'p2c_buffer' and
'smm_flag' are used for, add comments to those.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I4ec092a92fe9f0686ffb7103e441802fc05381f4
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83698
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
To avoid having constructs like 'dev->path.domain.domain' in the SoC
code, create the 'dev_get_domain_id' helper function that returns the
domain ID of either that device if it's a domain device or the
corresponding domain device's domain ID, and use it in the code.
If this function is called with a device other than PCI or domain type,
it won't have a domain number. In order to not need to call 'die',
'dev_get_domain_id' will print an error and return 0 which is a valid
domain number. In that case, the calling code should be fixed.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I3d79f19846cea49609f848a4c42747ac1052c288
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83644
Reviewed-by: Shuo Liu <shuo.liu@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Now that we have a get_psp_mmio_base function that will work on all SoCs
that use the psp_gen2 code, we can move back to accessing the PSP
registers via their MMIO mapping. This sort-of reverts
commit 198cc26e49 ("soc/amd/common/block/psp/psp_gen2: use SMN access
to PSP").
When doing SMN accesses from the SMI handler after the OS has taken over
ownership of the platform, there's the possibility to cause trouble by
clobbering the SMN access index register from SMM. So that should be
either avoided completely or the SMI code needs to save and restore the
original contents of the SMN index register.
The PSP MMIO base will be set up by the FSP before the resource
allocation in coreboot and be treated like a fixed resource by the
allocator. The first SMI where corresponding handler calls
'get_psp_mmio_base' happens when ramstage triggers the APM_CNT_SMMINFO
SMI right after mpinit which happens after the resource allocation. So
the PSP MMIO base address is expected to be configured and so the
'get_psp_mmio_base' function will cache the base address and won't need
to do any SMN access in subsequent calls that might happen after the OS
has take over control.
This isn't currently an issue, since the only PSP mailbox command from
the SMI handler after coreboot is done and the OS has taken over will
be during the S3/S4/S5 entry, and this will be triggered by the OS as
the last step after it is done with all its preparations for suspend/
shutdown. There will however be future patches that add SMI-handlers
which can send PSP mailbox commands during OS runtime, and so we have
to make sure we don't clobber the SMN index register.
TEST=PSP mailbox commands are still sent correctly on Mandolin.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I25f16d575991021d65b7b578956d9f90bfd15f6c
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83448
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This sort-of reverts commit 00ec1b9fc7 ("soc/amd/common/block/psp/
psp_gen2: simplify soc_read_c2p38") and is done as a preparation to
switch back to using the MMIO access to the PSP mailbox registers.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Icca3c7832295ae9932778f6a64c493e474dad507
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83447
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Add get_psp_mmio_base which reads the PSP MMIO base address from the
hardware registers. Since this function will not only be called in
ramstage, but also in SMM, we can't just look for the specific domain
resource consumer like it is done for the IOAPICs in the northbridge,
but have to get this base address from the registers. In order to limit
the performance impact of this, the base address gets cached in a static
variable if an enabled PSP MMIO base register is found. We expect that
this register is locked when it was configured and enabled; if we run
into the unexpected case that the PSP MMIO register is enabled, but not
locked, set the lock bit of the corresponding base address register to
be sure that it won't change until the next reset and that the hardware
value can't be different than the cached value.
This is a preparation to move back to using MMIO access to the PSP
registers and will also enable cases that require the use of the MMIO
mapping of the PSP registers.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I1d51e30f186508b0fe1ab5eb79c73e6d4b9d1a4a
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83446
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
The PSP code introduced in a following patch needs both SoC-specific
functions get_iohc_info and get_iohc_non_pci_mmio_regs to also be
available in SMM, so add those compilation units to the corresponding
target.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I4e32084b45f07131c80b642bc73d865fc57688a8
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83445
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Reviewed-by: Varshit Pandya <pandyavarshit@gmail.com>
Instead of implementing the functions get_iohc_misc_smn_base and
get_iohc_fabric_id in the SoC code, move those functions to the common
AMD code, and implement get_iohc_info in the SoC code that returns a
pointer to and the size of a SoC-specific array of domain_iohc_info
structs that contains the info needed by the common code instead. This
allows to iterate over the domain_iohc_info structs which will be used
in a later patch to find the PSP MMIO base address in both ramstage and
smm.
TEST=Mandolin still boots and all non-PCI MIO resources are still
reported to the resource allocator
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ifce3d2b540d14ba3cba36f7cbf248fb7c63483fe
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83443
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Reviewed-by: Varshit Pandya <pandyavarshit@gmail.com>
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
No need to open-code this when we have a function for this.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Iae570ba750cb29456436349b4263808e2e410e2e
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83643
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Shuo Liu <shuo.liu@intel.com>
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
In AMD platforms, the bit 4 of CMOS's Register A (0x0a) is DV0 bank
selection (0 for Bank 0; 1 for Bank 1) [1]. Since the MC146818 driver
accesses VBNV via Bank 0, the bit must be cleared before we can save
VBNV to CMOS in verstage.
Usually there's no problem with that, because the Register A is
configured in cmos_init() in ramstage. However, if CMOS has lost power,
then in the first boot after that, the bit may contain arbitrary data in
verstage. If that bit happens to be 1, then CMOS writes in verstage will
fail.
To fix the problem, define vbnv_platform_init_cmos() to call
cmos_init(0), which will configure the Register A and therefore allow
saving VBNV to CMOS in verstage.
[1] 48751_16h_bkdg.pdf
BUG=b:346716300
TEST=CMOS writes succeeded in verstage after battery cutoff
BRANCH=skyrim
Change-Id: Idf167387b403be1977ebc08daa1f40646dd8c83f
Signed-off-by: Yu-Ping Wu <yupingso@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83495
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
When we tried to add CMOS support to PSP verstage (CB:83495), the clang
builds failed on boards with cezanne SoC (such as Guybrush), due to
over-sized verstage. On the other hand, there is no such problem for gcc
builds on the same boards.
Building PSP verstage by clang generates much larger verstage size (81K)
compared with using gcc (67K). To unblock adding features to verstage,
temporarily enable -Oz for clang builds.
Change-Id: I033458556986ade88fb8e68499b632deae4dd419
Signed-off-by: Yu-Ping Wu <yupingso@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83594
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Nico Huber <nico.h@gmx.de>
To be able to use the IOHC_MMIO_EN define in other compilation units,
move the define to the corresponding header file.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: If88950418406d1709ed95b3d05f7e6ad66438f95
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83444
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Reviewed-by: Varshit Pandya <pandyavarshit@gmail.com>
The UEFI reference firmware uses AMDI0030 instead of AMD0030 as HID for
the GPIO controller.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I4a6fa1acdca0ee5b6e1358b6279b7c501d3dfd16
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83439
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
The UEFI reference firmware uses AMDI0030 instead of AMD0030 as HID for
the GPIO controller.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I8dd48d7d9cf3f6d75853bb825e5ddc32bba430b8
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83438
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
The table "IOMUX Functional Table" in PPR #57254 rev. 1.60 was used as a
reference. This should fix the ESPI_ALERT_D1 IOMUX setting for the
boards using the Glinda SoC which previously didn't match the hardware.
Compared to Phoenix, Glinda has two more chip select outputs for the
SPI2 controller and an additional ZST_STUTTER_RAIL IOMUX function.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Id9adfbe0c7aee90d6fe990f239d82a1d013e7f5f
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83437
Reviewed-by: Anand Vaikar <a.vaikar2021@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
The APOB NV size/base are embedded into the amdfw binary and read by
the PSP. These need to be synchronized with the FMAP region used by
coreboot to store the APOB data. soc_update_apob_cache() will only
use RECOVERY_MRC_CACHE if supported and if vboot is enabled, so the
NV base passed to the PSP needs to reflect that as well.
This fixes the issue of RAM training running on every boot on
non-vboot builds for Myst boards.
TEST=untested, but same change as made for Mendocino
Change-Id: Ib4a78a39badf0a067e22eebe5869e5ea51723f35
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83401
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
The APOB NV size/base are embedded into the amdfw binary and read by
the PSP. These need to be synchronized with the FMAP region used by
coreboot to store the APOB data. soc_update_apob_cache() will only
use RECOVERY_MRC_CACHE if supported and if vboot is enabled, so the
NV base passed to the PSP needs to reflect that as well.
This fixes the issue of RAM training running on every boot on
non-vboot builds for Skyrim boards.
TEST=build/boot Skyrim (Frostflow), verify RAM training only
run on first boot after flashing.
Change-Id: I9be1699d675331b46ee9c42570700c2b72588025
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83400
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Use the PCI_DEVFN macro to make the calculation of the ivhd->device_id
value a bit clearer.
TEST=Timeless build results in identical binary for Mandolin
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I7b7949ad3524790e7d7d527c488a32e785f55bc0
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83343
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add AMD SOC Family 17h Renoir CPUIDs per PPR doc #55922
Renoir is similar to Cezanne with only differences in CCX count.
Cezanne has one Zen3 CCX with 8 cores per CCX compared to
the two Zen2 CCX with 4 cores per CCX. Hence, coreboot side
Cezanne SOC code should be mostly compatible with Renoir and
can be leveraged.
Change-Id: I6b43eb782527351c79b835d094a5b61103cd6642
Signed-off-by: Anand Vaikar <a.vaikar2021@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83099
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
<stdarg.h> header is used to define macros for handling variable
argument lists in functions like printf. It does not depend on the string
or memory manipulation functions provided by <string.h>.
So let follow conventions and include only the necessary headers in each
header file.
Change-Id: I07ffc65b7feefb8ec4ab8dd268113f9ed8d24685
Signed-off-by: Elyes Haouas <ehaouas@noos.fr>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/82664
Reviewed-by: Maximilian Brune <maximilian.brune@9elements.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The CHECK_REV_IN_OPROM_NAME Kconfig option was introduced to solve the
problem of the PCI VID/DID combination of the Picasso iGPU not being
sufficient information to know which VGA BIOS file to run, so a new
function that additionally checks the PCI revision of that device was
introduced. Later it turned out that there might be a case where even
that isn't sufficient, so the soc_is_raven2() function is used in the
remap function to always use the correct VBIOS file.
Picasso is the only SoC that selected the CHECK_REV_IN_OPROM_NAME
Kconfig option, so all other SoCs are unaffected by this change.
Now that we use the VBIOS images with only the PCI VID and DID in the
CBFS file name for Picasso, SeaBIOS will find the VBIOS with the same ID
as the iGPU in CBFS and we don't need the workaround to add a third
VBIOS image via VGA_BIOS_DGPU_* that has the name that SeaBIOS expects.
This will result in SeaBIOS now running the VBIOS that has the same PCI
VID/DID as the hardware which will be the wrong one in the RV2 silicon
showing the PCO silicon PCI VID/DID, but that was also the case with the
VGA_BIOS_DGPU_* workaround where the board's Kconfig just selected one
of the two possible images during build time and hoped that it was the
correct one for that actual hardware. The only board where this patch
might cause a regression compared to the old behavior is the AMD Cereme
reference board with Pollock APU, but I'm not even sure if any coreboot
developer still has one of those boards, so I'm willing to accept that.
To properly solve the problem with SeaBIOS using the correct VBIOS file
in all cases, we'd need to generate that info during coreboot runtime
and somehow pass it to SeaBIOS, but that's out of scope for this patch.
TEST=On Mandolin with PCO silicon, the display output in both SeaBIOS
and Ubuntu still works. Booting Windows 10 via the pre-built EDK2
payload that I'm using also resulted in the display output working.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ia6de533c536044698d85404427719b8f534870fa
Reviewed-on: https://review.coreboot.org/c/coreboot/+/82598
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add the USB PHY configuration structs for the openSIL case, so that
those can be configured in the devicetree like in the FSP case.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Ied25e90859c4b1bc9b876bed3f3c46358ca36d32
Reviewed-on: https://review.coreboot.org/c/coreboot/+/82584
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
In the FSP case, the DDI descriptors aren't part of the devicetree and
are instead retrieved in romstage by calling the mainboard's
mainboard_get_dxio_ddi_descriptors function which allows updating the
descriptors during romstage where the devicetree is static. In the
openSIL case, the DDI configuration is first needed in ramstage, so we
can put this info into the devicetree and update it if needed in
ramstage.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I3de12ff6af42e38751a3016efa313613677fa87a
Reviewed-on: https://review.coreboot.org/c/coreboot/+/82580
Reviewed-by: Varshit Pandya <pandyavarshit@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Remove the TODO to update the chipset devicetree for Phoenix, since this
has already been done.
When re-checking the chipset devicetree, I found conflicting information
about the existence of the PCI bridge to an external PCIe port on bus 0
device 1 function 5, but after looking into this, I'm reasonably certain
that it either doesn't exist or at least wouldn't be usable, so I won't
add that one to the chipset devicetree.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I8f0e1540ed45408e86186253d3982a7ba0065ac6
Reviewed-on: https://review.coreboot.org/c/coreboot/+/82578
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Reviewed-by: Varshit Pandya <pandyavarshit@gmail.com>
Add the stub MPIO chips that contain the PCIe engine configuration for
the external PCIe interfaces to the devicetree. Birman's
port_descriptors_phoenix.c was used as a reference. The static
configuration in the devicetree assumes that the default WLAN0_WWAN0 is
selected; for the other cases we'll still need to fix up things
accordingly in the mutable devicetree. The WLAN01 and WWAN01 cases still
need to be handled in a follow-up patch. Since openSIL currently doesn't
use the info from the gpio_group struct element, but deasserts both PCIe
reset pins GPIO 26 and 27, the gpio_group isn't specified in the chip
configuration in the devicetree.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Icabe60322d46c1195284dd77ec39f9d143e3d2cb
Reviewed-on: https://review.coreboot.org/c/coreboot/+/81101
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
This adds a comment for unused AMD_FWM_POSITION_20000_DEFAULT.
Change-Id: Id8369f488893e7e5b2e7e7126d1b53199ed1aa77
Signed-off-by: Elyes Haouas <ehaouas@noos.fr>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/81973
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Reviewed-by: Eric Lai <ericllai@google.com>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
The chip drivers in the devicetree use the path where the corresponding
chip.h file resides both to include this chip.h file in the static.c
generated by util/sconfig from the devicetree and also for the names of
the chip config and chip ops struct. To be able to build a SoC using
either the MPIO chip driver from the openSIL stub or from the actual
openSIL glue code without needing different devicetree files for the
different cases, introduce a common MPIO chip.h file that then includes
the correct MPIO header file. The chip config and ops structures also
need to be renamed to take this change into account.
Thanks to Matt for pointing out how to make the path to the actual MPIO
chip.h file configurable via a Kconfig setting. This allows overriding
this path from site-local without the need to have any reference to
site-local in the upstream code.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: Iead97d1727569ec0d23a2b9c4fd96daff4bebcf6
Reviewed-on: https://review.coreboot.org/c/coreboot/+/82262
Reviewed-by: Varshit Pandya <pandyavarshit@gmail.com>
Reviewed-by: Martin L Roth <gaumless@gmail.com>
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
There's nothing in this header file that needs to be updated for the
Phoenix SoC, so remove the 'Update for Phoenix' TODO.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I9d7b5e8d8d6c8c22c2fae8e89d073481d21d8bdc
Reviewed-on: https://review.coreboot.org/c/coreboot/+/82150
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
