Use PlatformInfoHob->FeatureControlValue instead.
OnMpServicesAvailable() will find PlatformInfoHob using
GetFirstGuidHob() and pass a pointer to the WriteFeatureControl
callback.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Stop using the mPlatformInfoHob global variable. Let
BuildPlatformInfoHob() allocate and return PlatformInfoHob instead.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Stop using the mPlatformInfoHob global variable in S3Verification() and
Q35BoardVerification() functions. Pass a pointer to the PlatformInfoHob
instead.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Stop using the mPlatformInfoHob global variable in NoexecDxeInitialization()
function. Pass a pointer to the PlatformInfoHob instead.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Stop using the mPlatformInfoHob global variable in MemTypeInfoInitialization()
function. Pass a pointer to the PlatformInfoHob instead.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Stop using the mPlatformInfoHob global variable in PublishPeiMemory()
and GetPeiMemoryCap() functions. Pass a pointer to the PlatformInfoHob
instead.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Stop using the mPlatformInfoHob global variable in
Q35TsegMbytesInitialization() and
Q35SmramAtDefaultSmbaseInitialization() ) functions.
Pass a pointer to the PlatformInfoHob instead.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Stop using the mPlatformInfoHob global variable in PeiFvInitialization()
function. Pass a pointer to the PlatformInfoHob instead.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Stop using the mPlatformInfoHob global variable in AmdSevInitialize()
and AmdSevEsInitialize() functions. Pass a pointer to the
PlatformInfoHob instead.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Instead of using hard-coded strings ("0.0.0" for BiosVersion etc)
which is mostly useless read the PCDs (PcdFirmwareVendor,
PcdFirmwareVersionString and PcdFirmwareReleaseDateString) and
build the string table dynamuically at runtime.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
According to the Intel GHCI specification document section 2.4.1, the
goal for instructions that do not have a corresponding TDCALL is for the
handler to treat the instruction as a NOP.
INVD does not have a corresponding TDCALL. This patch makes the #VE
handler treat INVD as a NOP.
Signed-off-by: Ryan Afranji <afranji@google.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
There should be a check that the FV HeaderLength cannot be an odd
number. Otherwise in the following CalculateSum16 there would be an
ASSERT.
In ValidateFvHeader@QemuFlashFvbServicesRuntimeDxe/FwBlockServices.c
there a is similar check to the FwVolHeader->HeaderLength.
Cc: Erdem Aktas <erdemaktas@google.com>
Cc: James Bottomley <jejb@linux.ibm.com>
Cc: Jiewen Yao <jiewen.yao@intel.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Min Xu <min.m.xu@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
NOR flash emulation under KVM involves switching between two modes,
where array mode is backed by a read-only memslot, and programming mode
is fully emulated, i.e., the memory region is not backed by anything,
and the faulting accesses are forwarded to the VMM by the hypervisor,
which translates them into NOR flash programming commands.
Normally, we are limited to the use of device attributes when mapping
such regions, given that the programming mode has MMIO semantics.
However, when running under KVM, the chosen memory attributes only take
effect when in array mode, since no memory mapping exists otherwise.
This means we can tune the memory mapping so it behaves a bit more like
a ROM, by switching to EFI_MEMORY_WC attributes. This means we no longer
need a special CopyMem() implementation that avoids unaligned accesses
at all cost.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Sunil V L <sunilvl@ventanamicro.com>
Currently, when dealing with small updates that can be written out
directly (i.e., if they only involve clearing bits and not setting bits,
as the latter requires a block level erase), we iterate over the data
one word at a time, read the old value, compare it, write the new value,
and repeat, unless we encountered a value that we cannot write (0->1
transition), in which case we fall back to a block level operation.
This is inefficient for two reasons:
- reading and writing a word at a time involves switching between array
and programming mode for every word of data, which is
disproportionately costly when running under KVM;
- we end up writing some data twice, as we may not notice that a block
erase is needed until after some data has been written to flash.
So replace this sequence with a single read of up to twice the buffered
write maximum size, followed by one or two buffered writes if the data
can be written directly. Otherwise, fall back to the existing block
level sequence, but without writing out part of the data twice.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Sunil V L <sunilvl@ventanamicro.com>
NorFlashWriteSingleWord() switches into programming mode and back into
array mode for every single word that it writes. Under KVM, this
involves tearing down the read-only memslot, and setting it up again,
which is costly and unnecessary.
Instead, move the array mode switch into the callers, and only make the
switch when the writing is done.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Sunil V L <sunilvl@ventanamicro.com>
We never boot from NOR flash, and generally rely on the firmware volume
PI protocols to expose the contents. So drop the block I/O protocol
implementation from VirtNorFlashDxe.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Sunil V L <sunilvl@ventanamicro.com>
We only use NOR flash for firmware volumes, either for executable images
or for the variable store. So we have no need for exposing disk I/O on
top of the NOR flash partitions so let's remove it.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Sunil V L <sunilvl@ventanamicro.com>
We inherited a feature from the ArmPlatformPkg version of this driver
that never gets enabled. Let's remove it.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Sunil V L <sunilvl@ventanamicro.com>
QEMU's mach-virt is loosely based on ARM Versatile Express, and inherits
its NOR flash driver, which is now being used on other QEMU emulated
architectures as well.
In order to permit ourselves the freedom to optimize this driver for
use under KVM emulation, let's clone it into OvmfPkg, so we have a
version we can hack without the risk of regressing bare metal platforms.
The cloned version is mostly identical to the original, but it depends
on the newly added VirtNorFlashPlatformLib library class instead of the
original one from ArmPlatformPkg. Beyond that, only cosmetic changes
related to #include order etc were made.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Sunil V L <sunilvl@ventanamicro.com>
Create a new library class in Ovmf that duplicates the existing
NorFlashPlatformLib, but which will be tied to the VirtNorFlashDxe
driver that will be introduced in a subsequent patch. This allows us to
retire the original from ArmPlatformPkg.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Sunil V L <sunilvl@ventanamicro.com>
Per the UEFI specification, a device driver implementation should return
EFI_UNSUPPORTED if the ChildHandle argument in
EFI_COMPONENT_NAME2_PROTOCOL.GetControllerName() is not NULL.
Signed-off-by: Dimitrije Pavlov <Dimitrije.Pavlov@arm.com>
Reviewed-by: Samer El-Haj-Mahmoud <Samer.El-Haj-Mahmoud@arm.com>
Reviewed-by: Sunny Wang <sunny.wang@arm.com>
While the actual implementation (using qemu fw_cfg) is qemu-specific,
the idea to store the boot order as configured by the VMM in EFI
variables is not. So lets give the variables a more neutral name while
we still can (i.e. no stable tag yet with the new feature).
While being at it also fix the NNNN format (use %x instead of %d for
consistency with BootNNNN).
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
This reverts commit 60d55c4156.
Now that we have stateless secure boot support (which doesn't
need SMM) in OVMF we can enable the build option for MicroVM.
Bring it back by reverting the commit removing it.
Also add the new PlatformPKProtectionLib.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Compiler flag is needed to make (stateless) secure boot be actually
secure, i.e. restore EFI variables from ROM on reset.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
In case the 64-bit pci mmio window is larger than the default size
of 32G be generous and hand out larger chunks of address space for
prefetchable mmio bridge windows.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
In case we have a reliable PhysMemAddressWidth use that to dynamically
size the 64bit address window. Allocate 1/8 of the physical address
space and place the window at the upper end of the address space.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Try detect physical address space, when successful use it.
Otherwise go continue using the current guesswork code path.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Add some qemu specific quirks to PlatformAddressWidthFromCpuid()
to figure whenever the PhysBits value returned by CPUID is
something real we can work with or not.
See the source code comment for details on the logic.
Also apply some limits to the address space we are going to use:
* Place a hard cap at 47 PhysBits (128 TB) to avoid using addresses
which require 5-level paging support.
* Cap at 40 PhysBits (1 TB) in case the CPU has no support for
gigabyte pages, to avoid excessive amounts of pages being
used for page tables.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
When finding an unsupported entry just skip over and continue
with the next entry instead of stop processing altogether.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
See comment for details. Needed to avoid the parser abort,
so we can continue parsing the bootorder fw_cfg file.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Traditional q35 memory layout is 2.75 GB of low memory, leaving room
for the pcie mmconfig at 0xb0000000 and the 32-bit pci mmio window at
0xc0000000. Because of that OVMF tags the memory range above
0xb0000000 as uncachable via mtrr.
A while ago qemu started to gigabyte-align memory by default (to make
huge pages more effective) and q35 uses only 2G of low memory in that
case. Which effectively makes the 32-bit pci mmio window start at
0x80000000.
This patch updates the mtrr setup code accordingly.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Wire up the newly added UefiDriverEntrypoint in a way that ties dispatch
of the Ip4Dxe and Ip6Dxe drivers to QEMU fw_cfg variables
'opt/org.tianocore/IPv4Support' and 'opt/org.tianocore/IPv6Support'
respectively.
Setting both variables to 'n' disables IP based networking entirely,
without the need for additional code changes at the NIC driver or
network boot protocol level.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
All QEMU based OVMF platforms override the same set of network
components, to specify NULL library class resolutions that modify the
behavior of those components in a QEMU specific way.
Before adding more occurrences of that, let's drop those definitions in
a common include file.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
Add a new library that can be incorporated into any driver built from
source, and which permits loading of the driver to be inhibited based on
the value of a QEMU fw_cfg boolean variable. This will be used in a
subsequent patch to allow dispatch of the IPv4 and IPv6 network protocol
driver to be controlled from the QEMU command line.
This approach is based on the notion that all UEFI and DXE drivers share
a single UefiDriverEntryPoint implementation, which we can easily swap
out at build time with one that will abort execution based on the value
of some QEMU fw_cfg variable.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
Both ACPI shutdown and ACPI PM timer devices has been moved to different
port addresses in the latest version of Cloud Hypervisor. These changes
need to be reflected on the OVMF firmware.
Acked-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
Add VgaInb() helper function to read vga registers. With that in place
fix the unblanking. We need to put the ATT_ADDRESS_REGISTER flip flop
into a known state, which is done by reading the
INPUT_STATUS_1_REGISTER. Reading the INPUT_STATUS_1_REGISTER only works
when the device is in color mode, so make sure that bit (0x01) is set in
MISC_OUTPUT_REGISTER.
Currently the mode setting works more by luck because
ATT_ADDRESS_REGISTER flip flop happens to be in the state we need.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
The function reads the boot order from qemu fw_cfg, translates it into
device paths and stores them in 'QemuBootOrderNNNN' variables. In case
there is no boot ordering configured the function will do nothing.
Use case: Allow applications loaded via 'qemu -kernel bootloader.efi'
obey the boot order.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
The current ACPI Reclaim memory size is set as 0x10 (64KiB). The ACPI
table size will be increased if the memory slots' number of the guest
gets increased. In the guest with more memory slots, the ACPI Reclaim
memory size may not be sufficient for hibernation. This may cause
resume failure of the hibernated guest that was booted up with a fresh
copied writable OVMF_VARS file. However, the failure doesn't happen in
following hibernation/resume cycles.
The ACPI_MAX_RAM_SLOTS is set as 256 in the current QEMU. With
ACPI_MAX_RAM_SLOTS, 18 pages are required to be allocated in ACPI
Reclaim memory. However, due to the 0x10 (16 pages) setting, 2 extra
pages will be allocated in other space. This may break the
hibernation/resume in the above scenario.
This patch increases the ACPI Reclaim memory size to 0x12, i.e.
PcdMemoryTypeEfiACPIReclaimMemory is set as 0x12 (18 pages).
Signed-off-by: Annie Li <annie.li@oracle.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
