| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Add missing EFIAPI to MMBI IPMI infrastructure functions to bypass GCC build errors.
Signed-off-by: Darren Dong <darren.dong@intel.com>
|
|
Add missing EFIAPI in implementation of UEFI event
notification functions.
Signed-off-by: Michael D Kinney <michael.d.kinney@intel.com>
|
|
Fix build failure of JunoPkg:
- ArmFfaLib is located from ArmPkg to MdeModulePkg and it was included
in ArmVExpress.dsc.inc file So remove it.
- While integrating the PlatformStandaloneMm.dsc.inc, there was a
mistaken for missing !endif
Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
|
|
Commonise dsc file for StandaloneMm to PlatformStandaloneMm.dsc.inc
so that reduces duplication of code.
Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
|
|
On Juno, the StandaloneMM Boot Firmware Volume (BFV) is not in
the Flash area. Instead it is loaded in to the RAM by TF-A.
Therefore, disable the shadow copying of the BFV in StandaloneMM.
Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
|
|
add StandaloneMm build file for Juno platform
also, supports secure variable service with StandaloneMm.
Since the "Normal world shared area with Secure" is allocated at
(2GB - 17MB) as much as 1MB, change the PcdSystemMemorySize too.
Note: Although we are saying secure variable storage here,
the Flash on Juno hardware is accessible from Normal world.
The purpose of enabling StandaloneMM support for Juno is to
demonstrate the functionality on real hardware.
Signed-off-by: Yeoruem Yun <yeoreum.yun@arm.com>
|
|
When Juno uses StandaloneMm,
last 2MB NorFlash0 area is used for secure storage for
- secure variable service
- fTPM storage
fTPM's usage will be coming on future patch.
Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
|
|
The VanGogh platform includes some override code as the patch for EDK2. Due to the latest EDK2 release fixing some issues on AMD platforms, using override is no longer suitable. The VanGogh platform will follow the new rules in the master branch and remove the override folder. For VanGogh platform development, please use the previous commit ID to ensure proper operation.
Cc: Chen, Chen <Chen.Chen@amd.com>
Cc: Igniculus Fu <igniculus.fu@amd.com>
Reviewed-by: Eric Xing <eric.xing@amd.com>
Signed-off-by:Duke Zhai <duke.zhai@amd.com>
|
|
The "build" command can provide user with more information during build.
It can also be more quiet during operation.
This change adds "--silent" option to limit amount of output from build
process.
It also adds "--verbose" option to give user more details during build.
Signed-off-by: Marcin Juszkiewicz <marcin@juszkiewicz.com.pl>
|
|
RealTimeClockLib.h moved from EmbeddedPkg to MdeModulePkg some time ago.
Signed-off-by: Marcin Juszkiewicz <marcin@juszkiewicz.com.pl>
|
|
edk2 is removing support for ARM32 platforms so this
commit removes Omap35xx from edk2-platforms.
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
|
|
edk2 is removing support for ARM32 platforms. As such
this commit removes BeagleBoard from edk2-platforms.
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
|
|
edk2 is removing support for ARM32, so drop all
ARM32 code from edk2-platforms. This does not
change code copied from edk2 as-is.
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
|
|
PCD defined by *.dsc.inc works as default and should not override
platform specific PCD settings.
Move MinPlatform common *.dsc.inc to early place of BoardPkg DSC file
so Platform specific PCD settings will take effect (Override).
This is for resolving potential build and functionality issues due to
wrong PCD configuration.
Signed-off-by: Chasel Chiu <chasel.chiu@intel.com>
|
|
NULL library should not need to include BoardInitLib.h which requires
MinPlatformPkg.dec in INF
Signed-off-by: Chasel Chiu <chasel.chiu@intel.com>
|
|
Removing edk2 override files from code for AMD server "Genoa" platforms. Going forward, any required overrides to tianocore/edk2 will be implemented in AMD's NDA AGESA code and will be considered temporary. Temporary overrides will be removed from AMD's NDA AGESA code once changes that eliminate the need for an override are upstreamed to tianocore/edk2.
Temporary overrides for Turin can be found in the 1.0.0.7 Turin AGESA PI Package.
Note: A similar change and policy will be implemented for AMD embedded platforms.
Signed-off-by: Paul Grimes <paul.grimes@amd.com>
|
|
The developer building edk2 for RB3Gen2 has to provide DTB corresponding
to RB3Gen2 from the suitable Linux kernel build path where DT sources are
maintianed. Following additional build instructions should help:
- Provide devicetree blob (DTB) corresponding to the Qualcomm target like for
RB3Gen2:
$ mkdir -p $(WORKSPACE)/devicetree
$ cp <kernel-build-path-to-dtb>/qcs6490-rb3gen2.dtb $(WORKSPACE)/devicetree/
- Add DTB path to PACKAGES_PATH:
$ export PACKAGES_PATH=$(WORKSPACE)/edk2:$(WORKSPACE)/edk2-platforms:$(WORKSPACE)/devicetree
- Enable build option: -D USER_PROVIDED_DTB=true
$ build -b RELEASE -a AARCH64 -t GCC -D USER_PROVIDED_DTB=true -p Platform/Qualcomm/RB3Gen2/RB3Gen2.dsc
With DTB support, one should be able to boot a Linux distro in EL2 mode with
KVM enabled.
Signed-off-by: Sumit Garg <sumit.garg@oss.qualcomm.com>
|
|
Enable support for UFS driver on RB3Gen2 platforms. This driver is
registered as non-discoverable device over PCIe I/O protocol supported
by UFS pass through host controller driver framework.
Signed-off-by: Sumit Garg <sumit.garg@oss.qualcomm.com>
|
|
UFS controller on Qcom SoCs is attached to a non-discoverable bus, so it
is registered as a non-discoverable device with DMA being non-cache
coherent.
This UFS driver assumes that the prior stage XBL on Qcom SoCs has fully
initialized the UFS controller where the primary boot medium is UFS flash
on Qcom SoCs. Given that the initialization is skipped here and only the
UFS controller MMIO address space is registered for UFS I/O to take place.
Signed-off-by: Sumit Garg <sumit.garg@oss.qualcomm.com>
|
|
RealTimeClockLib header has been moved to MdeModulePkg.
So the package has been added to the relevant INF files.
Signed-off-by: Sachin Ganesh <sachinganesh@ami.com>
|
|
RealTimeClockLib header has been moved to MdeModulePkg.
So the package has been added to the relevant INF files.
Signed-off-by: Sachin Ganesh <sachinganesh@ami.com>
|
|
RealTimeClockLib header has been moved to MdeModulePkg.
So the package has been added to the relevant INF files.
Signed-off-by: Sachin Ganesh <sachinganesh@ami.com>
|
|
RealTimeClockLib header has been moved to MdeModulePkg.
So the package has been added to the relevant INF files.
Signed-off-by: Sachin Ganesh <sachinganesh@ami.com>
|
|
RealTimeClockLib header has been moved to MdeModulePkg.
So the package has been added to the relevant INF files.
Signed-off-by: Sachin Ganesh <sachinganesh@ami.com>
|
|
gEfiUnicodeCollationProtocolGuid was deprecated in favor of
a newer version gEfiUnicodeCollation2ProtocolGuid which
provides improved language support and better extensibility.
It was officially removed from UEFI specifications
2.10A and 2.11 and should no longer be used in
current implementations.
Cc: Sachin Ganesh <sachinganesh@ami.com>
Signed-off-by: Gowtham M <gowthamm@ami.com>
|
|
Update of AMD EDKII Platform Code for 5th Generation EPYC Processors (formerly Turin).
This sample platform code was tested to work with the Turin 1.0.0.7 AGESA PI package, which is available from AMD under license and NDA.
Signed-off-by: Paul Grimes <paul.grimes@amd.com>
|
|
Since it's been almost 10 years since the Trusted Firmware project
rebranded from ATF to TF-A, go through and change all references
in Ampere code from 'ATF' to 'TF-A' or 'TFA' depending on the
context.
Signed-off-by: Rebecca Cran <rebecca@bsdio.com>
|
|
Some builds failed with below error:
PcdFspWrapperBootMode is not defined in DSC file.
Fixing this issue by defining this PCD in MinPlatform common include
DSC file.
Signed-off-by: Chasel Chiu <chasel.chiu@intel.com>
|
|
Modern graphics cards benefit from being able to negotiate a
larger BAR size than they advertise by default. In testing,
the Linux amdgpu driver crashed without it enabled. Since
I'm not aware of any problems with it, enable it for all Ampere
Altra platforms.
Signed-off-by: Rebecca Cran <rebecca@bsdio.com>
|
|
edk2 commit 066275413411 ("MdePkg/Library: Remove MM_STANDALONE LibraryClass in UefiDevicePathLib.inf")
remove supports UefiDevicePathLib.inf in StandaloneMm but
make it use UefiDevicePathLibBase.inf in MdePkg.
Change the UefiDevicePathLib as UefiDevicePathLibBase.inf to fix
build failure.
Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
|
|
UefiDevicePathLibBase
Since commit 06627541341149fb9f22a5f45fd90c8a60d9e35a in edk2
UefiDevicePathLibBase should be used. Fixes build failure.
Signed-off-by: Mikko Rapeli <mikko.rapeli@linaro.org>
|
|
Fixes:
edk2/edk2-platforms/Platform/StandaloneMm/PlatformStandaloneMmPkg/PlatformStandaloneMmRpmb.dsc(...): error 4000: Instance of library class [MmServicesTableLib] is not found for module [/home/builder/src/base/repo/meta-ts/build/tmp/work/rockpi4b-poky-linux/edk2-firmware/master+git/sources/edk2/StandaloneMmPkg/Core/StandaloneMmCore.inf], [MmServicesTableLib] is:
consumed by /home/builder/src/base/repo/meta-ts/build/tmp/work/rockpi4b-poky-linux/edk2-firmware/master+git/sources/edk2/MdeModulePkg/Library/ArmFfaLib/ArmFfaStandaloneMmCoreLib.inf
consumed by /home/builder/src/base/repo/meta-ts/build/tmp/work/rockpi4b-poky-linux/edk2-firmware/master+git/sources/edk2/ArmPkg/Library/ArmStandaloneMmCoreEntryPoint/ArmStandaloneMmCoreEntryPoint.inf
Signed-off-by: Mikko Rapeli <mikko.rapeli@linaro.org>
|
|
Added compatibility for Firmware Performance Data Table, an ACPI table
, to store firmware boot stage timings.
Signed-off-by: Prachotan Bathi <prachotan.bathi@arm.com>
|
|
The CPU speeds returned by AmpereCpuLib are in Hz. Divide them by
1000000 to get the values to use in the SMBIOS tables, which are in
MHz.
Signed-off-by: Rebecca Cran <rebecca@bsdio.com>
|
|
Improve some HII strings related to ANC and SLC cache.
Clarify in the help string for ANC mode that 'ANC' means
Ampere NUMA Control. And simplify the rest of the string
to avoid repetition.
Replace '$' with the word 'Cache'. While '$' might be a common
shorthand in the industry, it confuses other people.
In the platform board information page, the help string was wrong:
it doesn't indicate SLC as L3 Cache, but simply the amount of
System Level Cache.
Signed-off-by: Rebecca Cran <rebecca@bsdio.com>
|
|
SecBoardInitLibNull does not requires any package except MdePkg.dec
Signed-off-by: PaddyDeng <paddydeng@ami.com>
|
|
Add LINUXBOOT_FILE_IN_UEFI_EXTRA environment variable to control the generation
of a separate LinuxBoot FD file stored in UEFI Extra region.
Signed-off-by: Tam Chi Nguyen <tamchinguyen@os.amperecomputing.com>
|
|
Some platforms consuming MinPlatformPkg does not use FSP architecture.
Therefore does not have IntelFsp2WrapperPkg presents.
Signed-off-by: PaddyDeng <paddydeng@ami.com>
|
|
When register a root SMI handler, it is expected to use windows calling
convention for the SMI handler. This patch adds missing EFIAPI for the
SMI handler SmmSwDispatcher to avoid potential issue from Linux build.
Signed-off-by: Guo Dong <guo.dong@intel.com>
|
|
Many Qualcomm targets use Gunyah as a firmware-launched hypervisor, making
EL2 unavailable for Non-secure firmware and operating systems, preventing
use of alternative hypervisors.
Quoting Arm Base Boot Requirements here:
6.3.1.1. UEFI boot at EL2
Systems must boot UEFI at EL2, to enable the installation of a hypervisor
or a virtualization-aware operating system.
So lets use a TZ SMC call supported on these Qcom targets to clean up
EL2/Gunyah configuration and transition to booting edk2 (UEFI) at EL2
exiting Gunyah. All the SMMU configuration gets cleaned up and handed
over in EL2 in bypass mode. The edk2 is able to configure EL2 MMU/SMMU as
you would expect on an Arm silicon executing edk2 in EL2.
Signed-off-by: Sumit Garg <sumit.garg@oss.qualcomm.com>
|
|
Since the edk2 commit 6a329eb853d1
("ArmPlatformPkg: Capture TransferList Information for SEC")
Sec requires the ArmTransferListLib.
To resolve build failure, add ArmTransferListLib.
Signed-off-by: Sumit Garg <sumit.garg@oss.qualcomm.com>
|
|
ArmMmuLib is moved from ArmPkg to UefiCpuPkg, update ArmMmuLib path for
RB3Gen2 platform to adapt EDK2 change.
Signed-off-by: Sumit Garg <sumit.garg@oss.qualcomm.com>
|
|
The files added under ManageabilityTransportKcsLib have been cloned from
ManageabilityTransportKcsLib/Dxe because this library can be used in both
PEI and DXE.
The original source files under the /Dxe and /Common folders will be
removed, and DxeManageabilityTransportKcs.inf will be kept for
backwards compatibility.
Signed-off-by: Cassandra Lam <Cassandra.Lam@amd.com>
|
|
Update fw_ver.sh to run `date` with a minus after the % sign
to suppress the leading zero that's normally printed. This
prevents printf from interpreting the values as octal, which
was causing a failure when August (month 8) began.
Signed-off-by: Rebecca Cran <rebecca@bsdio.com>
|
|
edk2 commit db5b28ed9e27 ("MdePkg, EmbeddedPkg: Move Fdt Table Guid to MdePkg")
removes gFdtVariableGuid while it moves the gFdtTableGuid to MdePkg.dec.
But, gFdtVariableGuid is used for development purpose to override
fdt by saving its path to load into the variable with gFdtVariableGuid.
gFdtVariableGuid is only used by FdtPlatformDxe driver and not at
any other place. So move gFdtVariableGuid into FdtPlatformDxe.dec
This patch fixes build failure of ArmJuno Platform.
Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
|
|
Since the ADLINK Ampere Development Kit and ASRock Rack ALTRAD8
boards aren't full systems, we don't know what type of chassis
they'll be installed in. So set the chassis type to unknown.
Signed-off-by: Rebecca Cran <rebecca@bsdio.com>
|
|
Daniel (@JohnAZoidberg) is not active in RISC-V EDK2 for very long time.
So, I am taking over until I find some one from SiFive to actively
maintain it.
Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
|
|
Since we no longer use the where EDK2 starting in M-mode directly, we
don't need the opensbi submodule as well. So, remove the .gitmodules
file since OpenSBI was the only submodule.
Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
|
|
Instead of starting EDK2 in M-mode with an integrated OpenSBI library,
we are moving to the payload design where any previous stage M-mode
firmware can launch EDK2 in S-mode as a payload. Apart from easier
maintenance, it has other advantages like supporting guests etc. With
payload design, several libraries also become redundant. So remove them.
U540 platform itself is a discontinued model. So, there is no point in
keeping U500 which is an FPGA model. Since qemu supports U540, keeping
its support.
Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
|
|
Since SiFive U540 doesn't have a proper storage driver in EDK2, enable
RAM disk so that one can boot linux with minimal rootfs.
Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
|
