Add missing GDT alignment into mBuffer to prevent possible memory
corruption on ALIGN_POINTER operation on NewGdtTable
in ArchExceptionHandler
Signed-off-by: Savva Mitrofanov <savvamtr@gmail.com>
Missing masks leads to shift out of bounds. Also there is no need to
construct CHAR16 using cast to CHAR8 buffer, better to use native endian
by assigning data directly into Ucs2Char variable
Signed-off-by: Savva Mitrofanov <savvamtr@gmail.com>
Reviewed-by: Marvin Häuser <mhaeuser@posteo.de>
GetSize routines should return zero size when checking Token existence
Signed-off-by: Savva Mitrofanov <savvamtr@gmail.com>
Reviewed-by: Vitaly Cheptsov <vit9696@protonmail.com>
In case PlatformBootManagerLib does not have PciLib dependency,
we need to explicitly depend on UefiBootServicesTableLib.
Otherwise UefiBootServicesTableLib may not be constructed before
DxePciLibI440FxQ35, which uses a constructor-less PcdLib that directly
accesses gBS.
This can be viewed as a bug in the current implementation of BaseTools,
namely GetModuleLibInstances. This function drops all constructor-less
dependencies from the dependency resolution list to avoid dependency
cycles, which at the same time causes issues like above.
To properly fix the issue one should go over each library with
constructors and for each its dependency without constructors add all
the secondary dependencies that do have constructors. While doable,
it may cause considerable performance issues and is thus not done
in this patch.
Signed-off-by: Vitaly Cheptsov <cheptsov@ispras.ru>
Added missing EFIAPI modifier to SmmSwDispatcher function which passed into gSmst->SmiHandlerRegister routine.
Signed-off-by: Savva Mitrofanov <sk.mitrofanov@ispras.ru>
Reviewed-by: Vitaly Cheptsov <cheptsov@ispras.ru>
UefiCpuPkg modules like ExceptionHandlerAsm.nasm require relocations
to .text or PIC as they use instructions like mov or bt against
external symbols:
1. mov rax, ASM_PFX(CommonInterruptEntry)
2. mov rax, HookAfterStubHeaderEnd
3. bt [ASM_PFX(mErrorCodeFlag)], ecx
4. mov [rcx + (@VectorNum - HookAfterStubHeaderBegin)], al
It is not easily possible to rewrite 3/4 cases without involving
more registers or more commands, and EDK II has never targeted
read-only .text at load time, only at runtime.
Change the defaults to let OVMF compile by CLANGDWARF as per:
https://lists.llvm.org/pipermail/llvm-dev/2017-March/111441.html
Signed-off-by: Vitaly Cheptsov <cheptsov@ispras.ru>
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=2510
Some firmwares:
- Report Shift modifier even when they report upper-case unicode letter.
- Report Ctrl modifier with "shifted" UniChar (i.e. X - 'A' + 1).
This change provides support for these firmwares preserving the compatibility
with the previous input handling.
Signed-off-by: Michael Belyaev <usrsse2@icloud.com>
Reviewed-by: Vitaly Cheptsov <vit9696@protonmail.com>
Adds a Pedro Falcato's UEFI EXT4 filesystem driver that implements the
EFI_FILE_PROTOCOL and EFI_SIMPLE_FILE_SYSTEM_PROTOCOL.
Signed-off-by: Savva Mitrofanov <savvamtr@gmail.com>
Remove efivarstore IFR buffer error when string load default.
In the case of varstore type IFR, it will be NULL,
just break into next opcode process.
Resolve issue from commit cbe8a09dbac5c39d24ce5c4b9e7d44607f6c23da
varstore type IFR string might return status error
while execute ExtractConfig function.
Signed-off-by: Longhao Lee <longhaox.lee@intel.com>
Unlike DXE USB enumeration that enumerates all changed ports on timer
interrupt, PEI USB Enumeration runs once at the driver entry point.
USB3.x devices initially appear in USB2.0 ports. When the USB2.0 port is
reset, the USB3.x device disappears from the USB2.0 port and appears on the
USB3.0 port. The USB3.x device won't be enumerated if the USB2.0 port
number is greater than the USB3.0 port number. Re-enumerate USB to make
sure USB3.x devices in this case.
Signed-off-by: Phil Noh <Phil.Noh@amd.com>
This reverts commit 58766a472932c485d41163b1746fb1d9e7984f07.
In edk2 commit 58766a4, validation of the two reserved FAT entries
was added.
However, it also checked the return of FatGetFatEntry to MAX_UINT32,
which is what FatGetFatEntry returns when it encounters an error,
e.g. not being able to read the disk. However, MAX_UINT32 is also a
valid value for the reserved FAT entries and under some conditions
these will be returned in the success case.
A FAT volume formatted with these valid values of the reserved FAT
entries will fail to boot an OS because the opening of the volume
will fail.
However, the reason FatGetFatEntry returns MAX_UINT32 is that most
other uses of the function are comparing it against the END_OF_CHAIN
mark, which MAX_UINT32 will trip and those functions will fail out.
Because this is a critical bug that can prevent OS booting and the
bug the original commit was solving was accounting for a bad FAT
filesystem formatting tool, this commit is reverted for now.
Future work will clean up FatGetFatEntry so that it returns an
EFI_STATUS, but that involves more work and this bug needs to be
resolved in the meantime.
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
ArmFfaLibCommonInit will return EFI_UNSUPPORTED when there is no FFA
supported on the platform. This is expected behavior. However, the return
of error code will incur program asserts.
This change fixed the non-FFA path for the Standalone MM instance.
Signed-off-by: Kun Qin <kun.qin@microsoft.com>
SEC, PEI_CORE and PEIM type modules generally execute in place from
memory that is not writable. This means that it is not generally
possible to use an entrypoint implementation that stores a dynamically
generated stack cookie into a global variable. For PEIMs in particular,
there may be other options, such as a DEPEX on the permanent memory PPI,
but the current dynamic implementations of the stack checking library
entrypoints for PEI_CORE and PEIM modules must not be used as-is.
So remove them, and update the readme accordingly.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Remove some source files that were part of an earlier incarnation of
DynamicStackCookieEntryPointLib but are no longer actually in use.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Commit efbf5ed moves StackCheckLibStaticInit to StackCheckLib,
and each SEC module has a dependency on StackCheckLib now.
Add StackCheckLib in FitUniversalPayloadEntry.inf to fix build
issue.
Signed-off-by: Ajan Zhong <ajan.zhong@newfw.com>
StandaloneMmCoreMemoryAllocationLib.c and StandaloneMmServicesTableLib.c
are both defining gMmst, StandaloneMmCoreMemoryAllocationLib will be
linked to StandaloneMmCore directly, StandaloneMmServicesTableLib may be
linked to StandaloneMmCore indirectly, when they are both linked to
StandaloneMmCore, there will be "lld-link: error: duplicate symbol: gMmst"
build error with Clang compiler.
gMmst is declared in MmServicesTableLib.h and its definition should be
owned by MmServicesTableLib.
This patch renames gMmst in StandaloneMmCoreMemoryAllocationLib.c to
mMemoryAllocationMmst to avoid this build error.
Signed-off-by: Star Zeng <star.zeng@intel.com>
Adds X64 support to the SRAT table generator library.
Adds EX64ObjLocalApicX2ApicAffinity CM object to get
platform configuration information.
Update the parser with newly added CM object.
Cc: Sami Mujawar <Sami.Mujawar@arm.com>
Cc: Pierre Gondois <pierre.gondois@arm.com>
Signed-off-by: Abdul Lateef Attar <AbdulLateef.Attar@amd.com>
In order to provide an example and test out dynamic stack
cookies in edk2, dynamic stack cookies are added to ArmVirtPkg
platforms using ArmVirt.dsc.inc, per maintainer request.
This does require moving the ArmVirt.dsc.inc to be included
before MdeLibs.dsc.inc as the BaseTools does not expand all
the Defines sections before processing.
Signed-off-by: Oliver Smith-Denny <osde@linux.microsoft.com>
In order to use dynamic stack cookies in ArmVirtQemu, we need
RNDR support. This is added by using the neoverse-n2 cpu.
Signed-off-by: Oliver Smith-Denny <osde@linux.microsoft.com>
To provide an example and test the code within edk2, this
adds stack cookie checking to OvmfIA32X64, doing no checking
for SEC and PEI_CORE modules, static cookies for PEIMs, and
dynamic cookies for all other module types.
Signed-off-by: Oliver Smith-Denny <osde@linux.microsoft.com>
In order to use dynamic stack cookies, we need RDRAND support
from QEMU, so this updates the QEMU launching code for OvmfPkg
to include RDRAND support.
Signed-off-by: Oliver Smith-Denny <osde@linux.microsoft.com>
Adds dynamic stack cookies in the form of copies of the entry
point libraries that use shared logic to update stack cookies
at runtime.
This relies on RDRAND on IA32/X64 and RNDR on AARCH64 to get a
random number to apply to the stack cookie on module entry point.
This simplifies the logic a platform must do to include stack
check functionality.
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
This commit oves StackCheckLib from a NULL lib to an instance of
StackCheckLib. This requires every entry point to add a library
dependency on StackCheckLib. It also requires every SEC module
to have a dependency on StackCheckLib because there is no
standard SEC entry point.
It allows for greater flexibility for a platform to apply stack
cookies and simplifies DSC logic.
Continuous-integration-options: PatchCheck.ignore-multi-package
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
Now that the X64 StandaloneMmCoreEntryPoint has been moved to
MdePkg, it can be removed from StandaloneMmPkg and consumed
from MdePpkg.
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
Moves StandaloneMmCoreEntryPoint for X64 to MdePkg to live with
the other entry point libs. It does not move the ARM64 version,
as this was just moved to ArmPkg due to its heavy coupling with
ArmPkg code. This will need to be revisited when dynamic stack
cookie support is added to ARM64 StMM.
This commit just adds the library in MdePkg, a separate commit
will remove it from StandaloneMmPkg to make it easier to consume.
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
StackCheckLib defines the interface between a compiler
and the stack checking code. It is being converted from
a NULL library class to an actual library class to make
it easier to use for a platform and be easier to define
the expected interface with a compiler, so if there is
a compiler change it can be tracked and caught.
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
RNDR is a standard register defined in the ARM ARM for
AARCH64. Move the definition from BaseRngLib to AArch64.h.
Furthermore, move the inclusion of this register definition
to the ARM specific header file.
Signed-off-by: Oliver Smith-Denny <osde@microsoft.com>
REF:https://github.com/tianocore/edk2/issues/10529
The `RngPei` PEIM can be used if RNG should be provided over a
dynamic binary interface to other PEIMs on a platform.
Tested with:
- SecurityPkg CI
- Verify RNG linked with RngLib is executed as expected
- Verify random numbers are generated successfully with a valid
RngLib
Integration instructions:
Use the `RngPei` module if a platform needs to produce
`gEfiRngPpiGuid`.
The platform should usually link a different `RngLib` instance to
`RngPei` than other PEIMs that may use the RNG PPI produced since
`RngPei` is responsible for producing the PPI.
For example, a `RngLib` instance that uses the rdrand instruction
may be linked against `RngPei` and a `RngLib` instance that uses
the RNG PPI may be linked against other PEIMs.
Signed-off-by: Michael Kubacki <michael.kubacki@microsoft.com>
REF:https://github.com/tianocore/edk2/issues/10529
Adds a new PEI library instance for RngLib that uses the RNG services
provided by the RNG PPI.
This library instance will add a DEPEX on gEfiRngPpiGuid on modules
it links against. It can be used to allow PEIMs to get RNG support
over a dynamic interface.
Signed-off-by: Michael Kubacki <michael.kubacki@microsoft.com>
REF:https://github.com/tianocore/edk2/issues/10529
Adds a new PPI that serves the same purpose as EFI_RNG_PROTOCOL in
DXE. This PPI can be produced by a PEIM to provide a dynamic interface
to RNG services in PEI.
Signed-off-by: Michael Kubacki <michael.kubacki@microsoft.com>
Move the structure definition to a common location that can be shared
with the RNG PPI. Move the algorithms to a common header that can be
referenced independent of the protocol.
Include the algorithm header in the interface header since the
algorithms are directly used in the interface and for compatibility
with existing code.
Signed-off-by: Michael Kubacki <michael.kubacki@microsoft.com>
We've migrated bugzilla bugs to github issues, so update the references
to specific bugs called out in the top-level readme.
Also delete the references to BZ2661 and BZ2668 which are both *closed*:
https://github.com/tianocore/edk2/issues/8453https://github.com/tianocore/edk2/issues/8458
And drop bugzilla.tianocore.org from "references".
The server is no longer in use, and the certificate has expired,
so stop telling people to go there.
Signed-off-by: Leif Lindholm <leif.lindholm@oss.qualcomm.com>
Update BZ links to corresponding GitHub issues links now that the
BZ -> GitHub issue migration is complete.
Signed-off-by: Michael Kubacki <michael.kubacki@microsoft.com>
By default, Doxygen tries to nicely format all text within doc blocks.
Unfortunately this mangles ASCII art: add @verbatim statements
to EdkiiSystemFmpCapsule.h to cause Doxygen to display the text as
provided.
Signed-off-by: Rebecca Cran <rebecca@bsdio.com>
With the latest GccBase.lds combining with CLANG build, there is no
need to enforce 4KB alignment of in UPL CLANGDWARF build.
This reverts commit b3bfb8f22d43bd9577df85bec20e75eb2f659990.
Signed-off-by: Chasel Chiu <chasel.chiu@intel.com>
This adds #defines and struct typedefs for the various structure
types in the CXL Early Discovery Table (CEDT).
Signed-off-by: Yuquan Wang <wangyuquan1236@phytium.com.cn>
If command line options are moved into a response file
of a GCC family build, then the file path separators are
converted from '\' to '/'. However, this can corrupt
command line options that are quoted strings.
Update GenMake to no convert '\' to '/' in quoted strings.
Signed-off-by: Michael D Kinney <michael.d.kinney@intel.com>
The linker option 'no-warn-rwx-segments' breaks both the LLVM linker and
versions of the binutils ld.bfd linker prior to 2.39.
Now that the ELF image is made up of separate R-X and RW- segments, this
warning is no longer emitted and so there is no longer a need to
suppress it either.
While at it, move GCC_DLINK_FLAGS_COMMON (which is not common but only
used by Ia32 and X64) into its only user so it can be dropped.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
To prevent the ELF linkers from complaining about emitted ELF segments
that require both writable and executable permissions, define two
separate R-X and RW- ELF segments, and emit the output sections
explicitly into those segments as appropriate.
Note that this has no bearing on the PE image, and using a single RW-
segment would probably be fine too.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
The original reason for creating a separate version of the ELF linker
script for Clang was the difference between COMMONPAGESIZE and
MAXPAGESIZE, which can we provided on the command line to the respective
linkers (ld.bfd versus lld). That difference no longer exists, and both
use COMMONPAGE_SIZE. So there is no longer a need to maintain a fork,
which has already been going out of sync with the original for no good
reason.
So merge the two and call it GccBase.lds
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
The GccBase.lds and ClangBase.lds ELF linker scripts have been laid out
very carefully to ensure that the memory mappings of .text and .data are
such that they can be preserved in the PE/COFF memory image. This
removes the need to update any place-relative ELF relocations when
generating the PE/COFF image, making its job much easier, and
potentially allowing it to disregard static ELF relocations altogether,
and rely solely on dynamic ELF relocations.
Adding an arbitrary .entry section before .text breaks those
assumptions, so instead of emitting it as a separate section, move its
payload to the start of .text.
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
