Update AllApArrivedWithException if the 1st timeout sync is skipped
but all APs have arrived.
Example:
If IsCpuSyncAlwaysNeeded() returns false, LMCE is enabled and
triggered, and another SMI source combined with LMCE causes all APs
to enter SMI. Then 2nd timeout sync will be also skipped, but all APs
have arrived, so AllApArrivedWithException should be updated before
using.
Signed-off-by: Jiaxin Wu <jiaxin.wu@intel.com>
Avoids openssl-3.5 trying to figure automatically with some macro
magic, which happens to not work with the microsoft compiler.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
SATA CD-ROMS are still conventionally used in many
virtual environments, so it's nice to support them
out of the box.
Tested with QEMU 9.2.3 with the following controller
and drive:
qemu-system-loongarch64 -M virt [...] \
-device ahci,id=ahci0,multifunction=on,bus=pcie.0,addr=0x7 \
-drive if=none,id=drive-sata2,media=cdrom,format=raw,aio=threads,file=/path/to/bootable.iso \
-device ide-cd,bus=ahci0.2,drive=drive-sata2,id=sata2,bootindex=102
Signed-off-by: Ivan A. Melnikov <iv@altlinux.org>
The debug print incorrectly uses "%016x" to print a UINT64 PhysicalAddress.
This can cause truncation since "%x" expects UINT32.
Update the format specifier to "%016llx" to properly handle UINT64 values
and ensure correct output across all architectures.
Signed-off-by: Aaron Li <aaron.li@intel.com>
Stanislaw Grams replaces Min Xu as the reviewer for patches to
OvmfPkg/Confidential Computing
Signed-off-by: Stanislaw Grams <stanislaw.grams@intel.com>
With firmware context APIs removed, store HobList pointer directly in the
scratch register for Peiless booting.
Signed-off-by: Tuan Phan <tphan@ventanamicro.com>
Encapsulate IA32 and X64 code under preprocessor directives to enable
module compatibility with other architectures.
Signed-off-by: Tuan Phan <tphan@ventanamicro.com>
Eliminate the APIs using the scratch register as PI 1.9 specifies
it for the PEI service table pointer.
Signed-off-by: Tuan Phan <tphan@ventanamicro.com>
GICv5 uses a different ID scheme for interrupt numbers to previous GICs; the
top bits indicate the type of interrupt (PPI/SPI/LPI). Update TimerDxe to use
the new ID scheme for GICv5. As the timer PPIs are architected for GICv5 they
should be the same on all GICv5 platforms.
Signed-off-by: Sarah Walker <Sarah.Walker2@arm.com>
Add initial driver support for GICv5. This initial driver supports Private
Peripheral Interrupts (PPIs) and Shared Peripheral Interrupts (SPIs).
Signed-off-by: Sarah Walker <Sarah.Walker2@arm.com>
[ardb: Merge v5 support into GicV3Dxe]
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Before extending the GicV3Dxe driver with support for GICv5, make the
driver AArch64-only to avoid the need for adding support to the 32-bit
build, which would be rather pointless.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Some of code for handling Rx/Tx buffer is duplicate.
This patch commonize some of duplication routine used in
Rx/Tx buffer related functions.
Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
To use Arm-FFA intereface in PeilessSec, implments
ArmFfaSecLib used by PeilessSec.
For example, communicate with TPM service using CRB over ARM-FFA
(via Tpm2DeviceLibFfa), PeilessSec need to use Arm-FFA interface.
Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
The PcdFfaTxRxPageCount can never be changed dynamically
and is configured at build time to specify the size of
the Rx/Tx buffers.
Therefore, make PcdFfaTxRxPageCount a PcdsFixedAtBuild PCD.
Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
The commit e15fe06603
("MdeModulePkg/Library: make ArmFfaPeiLib available early PEIM stage")
uses ArmFfaPeiLib in the early PEIM stage.
However, the Rx/Tx buffer allocated in the early PEIM stage uses
temporary memory. This results in a memory leak when the temporary
memory's heap is relocated to permanent memory.
For example, if the Rx/Tx buffer memory is allocated at 0x20006000
in temporary memory, and if offset between temporary memory and
permanent is 0x40000000, then:
- Once permanent memory installed the temporary memory at 0x20006000
is migrated to 0x60006000.
- However, ArmFfaPeiLib allocates new Rx/Tx buffer without freeing
the migrated Rx/Tx buffers, i.e. the buffers at 0x60006000.
This results in a memory leak as the migrated Rx/Tx buffer area is
lost.
To address this memory leak, use the MemoryAllocationHob's name, so
that the migrated memory area will be reused as Rx/Tx buffer.
This patch also includes rename ArmFfaRxTxStmm.c to
ArmFfaStandaloneMmRxTxMap.c to keep the file name convention in
ArmFfaLib with ArmFfa{Phase}{...}.c
Fixes: e15fe06603 ("MdeModulePkg/Library: ...")
Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
Continuous-integration-options: PatchCheck.ignore-multi-package
This commit adds additional dumping logic to Tpm2DeviceLibDTpm,
print at DEBUG_SECURITY to aid in TPM debugging.
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
Co-authored-by: Bret Barkelew <brbarkel@microsoft.com>
The TPM code is currently very noisy (e.g. in a sample platform,
4,000 of the 5,700 lines printed to the serial port at DEBUG_INFO
level were from the TPM code). For TPM debugging, this is very
critical information, but for most builds it simply spams the logs
and slows down the build.
This commit moves the event log and PCR dumping to log at
DEBUG_SECURITY level.
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
The TCG code is very noisy when a TPM is connected. This
commit downgrades some prints to verbose and removes some
others that do not have value (such as function enter and
exit prints).
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
Tcg2Dxe and its libraries are currently the noisiest modules in
edk2. For a sample platform printing at INFO level, Tcg2Dxe
printed 4,000 lines out of 5,700 total lines printed.
This commit defines a DEBUG_SECURITY bit to control the debug output
of Tcg2Dxe and other security related components. Most of the output is
not useful except for deep debugging of TPM transactions, so it is
appropriate to only print when the DEBUG_SECURITY bit is present.
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
Implements the _aulldiv() intrinsic function necessary to compile CryptoPkg
using the CLANGPDB toolchain. The existing MASM assembly implementation of
this function has been converted to NASM to enable it to be used with both
the Visual Studio and Clang compilers.
Signed-off-by: Nate DeSimone <nathaniel.l.desimone@intel.com>
During USB device enumeration, issuing a hot reset on a port is skipped if
there is a reset change status already detected on the port. This can
happen when enumerating devices after a host controller soft reset (which
drives a hot reset down the ports).
However, in certain cases an attached device may not be responsive even if
the reset change and connection status bits are set. For e.g., according
to xHCI spec section 4.19.5.1 the port reset change bits can be set when
a hot reset driven on the port transitions to a warm reset and completes
with errors. For such instances it is worthwhile to force a hot reset
during enumeration to try and recover unresponsive devices.
During enumeration check whether querying port status returns
EFI_DEVICE_ERROR and try a port reset if there is a device attached to
the port.
Signed-off-by: Aaron Pop <aaronpop@microsoft.com>
If IASL_PREFIX is not already defined, then edksetup.bat should set it.
This unifies the behavior between Visual Studio and Mingw-w64.
Signed-off-by: Nate DeSimone <nathaniel.l.desimone@intel.com>
