Skip to content

B850 support part3#870

Open
miczyg1 wants to merge 24 commits intodasharo-25.12from
b850_support_part3
Open

B850 support part3#870
miczyg1 wants to merge 24 commits intodasharo-25.12from
b850_support_part3

Conversation

@miczyg1
Copy link
Copy Markdown
Contributor

@miczyg1 miczyg1 commented Apr 3, 2026

No description provided.

miczyg1 added 24 commits March 30, 2026 12:34
There are many chanegs to the OpenSIL design since Genoa. Some MPIO
macros are even missing. Ensure evrything is initialized correctly
and the OpneSIL builds.

Upstream-Status: Pending
Change-Id: I388fe0171c07bd423e1344fbe380187052895b35
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
…ntry/exit

Change-Id: I86ebec73c281625eca7376992f3c8d6492eea183
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Change-Id: I6d6d3b73fc840f3cc366c5d221df8c18f27bd425
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
On Phoenix AM5 platform a similar problem to Turin platform occured.
If coreboto set a different ECAM_MMCONF_BASE_ADDRESS than PSP in ABL,
the PCI access did not work. Thankfully the SMN base address and DF
registers are the same for Phoenix and Turin, which suggests that
the synchronization of PCI MMCONF address in DF may be in common code.

TEST=Set an arbitrary PCI MMCONF in coreboot above TOM address and see
bootblock is still giving signs of life on serial console on MSI PRO
B850-P.

Change-Id: Ibf39bdba4dfd1c461bcd835f07118cf6b0fd1bb6
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
If the xHCI MMIO does not respond it may return all FFs. This causes
the code to loop through entire xHCI MMIO resource region slowing down
the boot process. Also fix iterating over the capabilities. The next
capability pointer also denotes the number of words to the next
capability, so the current pointer has to be moved by a next capability
pointer shifted left by two bits. See xHCI specification for the next
capability pointer field description.

Change-Id: I777736d3f4b8493b2d2c3aca7882961eb95d7a77
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Change-Id: Ifaa218ab8e1d799f7e053481175c7375c2cabc3d
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Fix scope of serial redirection options and remove duplicates.

Upstream-Status: Inappropriate [custom build options]
Change-Id: I07379ecdd4f4752cf7eeaa5b3772369406383a34
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
IOMMU VF BARs are available only on server for now. Guard VF BAR init
with new Kconfig option to avoid setting them in unsupported HW.

Upstream-Status: Pending
Change-Id: I365e5d472335d3b79aebf16f2df6a280b8bd87b0
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
On newer platforms there is SPI_ADDR32_CTRL0 with SPI_ROM_ADDR_32BIT
controllign whether the SPI controller takes 32bit flash address
for SPI transactions or not.

Upstream-Status: Pending
Change-Id: Ia655469acea37ae6416ab00f49e64fcb3ba88d75
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
The PSP_BASE_MSR register has to be programmed with the CPP MMIO
base address. Add API to obtain the address, so that SoC code can
program the MSR during CPU init if necessary.

Upstream-Status: Pending
Change-Id: I37bd5177d0d406c0ab6588438ab1d39c728bb4ad
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Program PSP_BASE_MSR on BSP if not yet programmed and if CCP MMIO
has been allocated.

Upstream-Status: Pending
Change-Id: I2f5d9e2f47244b6bb71a472b145969f9b0d4b40a
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Report the root bridge information to EDK2 to avoid problems
with root bridge apertures reporting. Similar problem occured
on Turin system.

Upstream-Status: Pending
Change-Id: I5c0e01033b8afdefc9944690deac368dccaac7e8
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
IVRS should report DMA-capable non-PCI devices using ACPI HID and UIDs.
Send the IVRS table to PSP for future DRTM support.

Change-Id: I4412bc529e2cdff47b460e08c86fd3bce42c0a2b
Upstream-Status: Pending
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Add necessary PSP commands so that the PSP Boot Done message does not
halt the platform.

Upstream-Status: Pending
Change-Id: If73a2b710d2b0cb3f34dd27e3fb952dc8a815106
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Synchronize all ACPI I/O ports with what coreboot expects it to be.
Pass additional coreboot configurtion to openSIL to match the behavior
with Kconfig.

Upstream-Status: Pending
Change-Id: I262406ce33ba842f862f60146841155a1afaf41c
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Add IRQ routing and GPIO configuration. Limit the coreboot
to 16MB, because flashing whole 32MB is time consuming.

Change-Id: I1255076f6d55a5e96a3d43217e2fc4f442d267e8
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
For some reason the call to enable IOAPIC decoding was not present.
Almost all AMD SoCs do that, so add it for Phoenix too.

Upstream-Status: Pending
Change-Id: Ia45c344994dd4cadb9ce46ad3fec97a2aff4074d
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Upstream-Status: Inappropriate [custom config]
Change-Id: I60362b14985fd4cb6ee3b295c5b777e3db406951
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Upstream-Status: Pending
Change-Id: I14824f465a5ec138a1f3b0141774c9a84972845c
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Do not poll for ROM Armor enforcement if we didn't enable RON Armor
in the build at all. It saves precious time spent in SMM.

Upstream-Status: Pending
Change-Id: Idafdb7433ddfa8ff22b80818b4772f801e43b0a3
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
…ties

Follow Turin's example and set bus master on PSP. Update HSTI at the
end of coreboot so that OS PSP driver see all capabilities.

Upstream-Status: Pending
Change-Id: I78358b97b743798e0d99a8952f7de4788b8414a0
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Add SD controller to the device list. Despite it is no longer
available in silicon according to PPR, it may appear on PCI bus
if not disabled by silicon initialization module.

Upstream-Status: Pending
Change-Id: Ia99cdbc355d3714997b039370e049496a5c7acb1
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Upstream-Status: Pending
Change-Id: Ib130ea05deb3a85a62641fbe89369660a9def256
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
The PCI MMCONF can not be moved and has to be in sync with APCB.
Assign I/O address for ACPI so that the resource allocator does
not complain on resource that didn't fit.

Upstream-Status: Pending
Change-Id: I66d4700d56477e06973ba46ae05d0232e5536608
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant