From what I can see this bug dates back to the commit from 2011 where
support for this was added: 2cf4b60895
The first problem is that PopulateLevel2PageTable overflows the
translation table buffer because it doesn't verify that the size
actually fits within one level 2 page table.
The second problem is that the loop in FillTranslationTable doesn't
care about the PhysicalBase or the RemainLength and always substracts
one section size from RemainLength.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Michael Zimmermann <sigmaepsilon92@gmail.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Add a missing newline to the version string output that is displayed
on the serial console, or the next line will be appended to it.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
If gEfiMdeModulePkgTokenSpaceGuid.PcdFirmwareVersionString is set to
a non-empty string, print it to the console at boot. Note that this
is independent of DEBUG/RELEASE or graphical vs serial console,
although we do attempt to stay clear of the logo and progress bar
in graphical mode, by printing it top center.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
This patch add implementation of following new API introduced into
CpuExceptionHandlerLib. Since this lib hasn't support Stack Guard
and stack switch, the new method just calls original
InitializeCpuExceptionHandlers.
EFI_STATUS
EFIAPI
InitializeCpuExceptionHandlersEx (
IN EFI_VECTOR_HANDOFF_INFO *VectorInfo OPTIONAL,
IN CPU_EXCEPTION_INIT_DATA *InitDataEx OPTIONAL
);
Cc: Leif Lindholm <leif.lindholm@linaro.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Jiewen Yao <jiewen.yao@intel.com>
Suggested-by: Ayellet Wolman <ayellet.wolman@intel.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Jian J Wang <jian.j.wang@intel.com>
Reviewed-by: Jeff Fan <vanjeff_919@hotmail.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
With the last user FdtPlatformDxe removed, we can finally get rid of the
last bit of ARM BDS related cruft.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Currently, each ARM platform built with RVCT that uses ArmHvcLib
or ArmSmcLib needs to specify a CPU target that implements both the
security and virtualization extensions, so that the assembler does
not choke on the 'hvc' and 'smc' instructions in ArmHvcLib/ArmSvcLib.
Let's move these overrides into the module .INFs so we can lift this
requirement at the platform side.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Flash region needs to be set as cacheable (write back) to increase
performance, if PEI is still XIP on flash or DXE FV is decompressed
from flash FV. However some ARM platforms do not support to set flash
as inner shareable since flash is not normal DDR memory and it will
not respond to cache snoop request, which will causes system hang
after MMU is enabled.
So we need a new ARM memory region attribute WRITE_BACK_NONSHAREABLE
for flash region on these platforms specifically. This attribute will
set the region as write back but not inner shared.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Peicong Li <lipeicong@huawei.com>
Signed-off-by: Heyi Guo <heyi.guo@linaro.org>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
This patch adds a library that enables invocation of SVCs from Exception
Level EL0. It will be used by the Standalone MM environment to request
services from a software running in a privileged EL e.g. ARM Trusted
Firmware. The library is derived directly from Arm SMC Library.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Achin Gupta <achin.gupta@arm.com>
Signed-off-by: Supreeth Venkatesh <supreeth.venkatesh@arm.com>
[ardb: assign frame pointer (AArch64)
keep stack alignment (ARM)]
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
The ESRT hook call that I just added invokes the protocol before
retrieving a pointer to it, which interestingly enough did not
result in any crashes, nor did it get picked up by GCC. Clang did
notice, though, so let's fix it right away.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
The ESRT management protocol needs to be invoked at the appropriate times
to get the ESRT config table to be published when the ReadyToBoot event
is signalled. So add this handling to the default ArmPkg implementation
of PlatformBootManagerLib.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Process any capsule HOBs that were left for us by CapsulePei. This
involves calling ProcessCapsules() twice, as explained in the comment
in DxeCapsuleLibFmp [sic].
1) The first call must be before EndOfDxe. The system capsules is processed.
If device capsule FMP protocols are exposted at this time and device FMP
capsule has zero EmbeddedDriverCount, the device capsules are processed.
Each individual capsule result is recorded in capsule record variable.
System may reset in this function, if reset is required by capsule and
all capsules are processed.
If not all capsules are processed, reset will be defered to second call.
2) The second call must be after EndOfDxe and after ConnectAll, so that all
device capsule FMP protocols are exposed.
The system capsules are skipped. If the device capsules are NOT processed
in first call, they are processed here.
Each individual capsule result is recorded in capsule record variable.
System may reset in this function, if reset is required by capsule
processed in first call and second call.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
The ARMv8.2-FP16 extension introduces support for half precision
floating point and the processor ID registers have been updated to
enable detection of the implementation.
The possible values for the FP bits in ID_AA64PFR0_EL1[19:16] are:
- 0000 : Floating-point is implemented.
- 0001 : Floating-point including Half-precision support is
implemented.
- 1111 : Floating-point is not implemented.
- All other values are reserved.
Previously ArmEnableVFP() compared the FP bits with 0000b to see if
the FP was implemented, before enabling FP. Modified this check to
enable the FP if the FP bits 19:16 are not 1111b.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Sami Mujawar <sami.mujawar@arm.com>
Signed-off-by: Evan Lloyd <evan.lloyd@arm.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Now that we have a generic DmaLib implementation for non-coherent DMA,
let's get rid of the ARM specific one.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Now that ArmDmaLib can take care of its own memory allocation needs,
let's get rid of UncachedMemoryAllocationLib entirely. This forces
platforms to declare the required semantics (non-cache coherent DMA,
whichever way it is implemented), rather than using uncached memory
allocations directly, which may not always be the right choice, and
prevents sharing of drivers between platforms if one is cache coherent
and the other is not.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Now that ArmDmaLib no longer uses uncached mappings for short-lived
bounce buffers used for streaming DMA, the only place we allocate
uncached memory is in DmaAllocateBuffer (), which is used for static
mappings shared between the host and the device, e.g., for packet
descriptor rings etc.
There is no performance concern around such long lived mappings, and
so we can really do without the overhead of UncachedMemoryAllocationLib,
which is a sizable chunk of poorly maintained code that never actually
releases any memory, and despite the fact that it implements pool based
routines, it always performs page based allocations anyway.
So let's invoke the DXE services directly to manage memory attributes
on allocations, and keep track of the allocations in a linked list so
we can restore the attributes and free the memory properly after use.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
The ArmPkg implementation of DmaLib uses double buffering to ensure
that any attempt to perform non-coherent DMA on unaligned buffers cannot
corrupt adjacent unrelated data which happens to share cachelines with
the data we are exchanging with the device.
Such corruption can only occur on bus master write, in which case we have
to invalidate the caches to ensure the CPU will see the data written to
memory by the device. In the bus master read case, we can simply clean
and invalidate at the same time, which may purge unrelated adjacent data
from the caches, but will not corrupt its contents.
Also, this double buffer does not necessarily have to be allocated from
uncached memory: by the same reasoning, we can perform cache invalidation
on an ordinary pool allocation as long as we take the same alignment
constraints into account.
So update our code accordingly: remove double buffering from the bus
master read path, and switch to a pool allocation for the double buffer.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Move IS_DEVICE_PATH_NODE into header to share it.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jun Nie <jun.nie@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
This adds an implementation of the ResetSystemLib library class as
defined in MdeModulePkg. It is used as the platform glue by the generic
ResetSystemRuntimeDxe which lives in the same package.
This implementation is intended to replace the EfiResetSystemLib based
implementation that is deprecated now that we have decided that there is
no longer a reason to keep a different ResetSystem() implementation
under EmbeddedPkg.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
ARM ArmHvcLib looks like it was created from copy of ArmSmcLib which
looks like it was created from a copy of the AArch64 version.
Both of these files include AsmMacroIoLibV8.h instead of
AsmMacroIoLib.h, although since they only use macros that are identical
between the two, there was no functional issue caused by this.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Leif Lindholm <leif.lindholm@linaro.org>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
In order to be able to produce meaningful diagnostic output when taking
synchronous exceptions that have been caused by corruption of the stack
pointer, prepare the EL0 stack pointer and switch to it when handling the
'Sync exception using SPx' exception class.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Currently, we only attempt to walk the call stack and print a backtrace
if the program counter refers to a location covered by a PE/COFF image.
However, regardless of the value of PC, the frame pointer may still have
a meaningful value, and so we can still produce the remainder of the
backtrace.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Add the gEfiDebugImageInfoTableGuid, which is referenced in the code,
to both .INF files describing this module.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Replace the duplicated and outdated code in QuietBoot.c with a reference
to BootLogoLib, which provides the same functionality. This also allows
us to drop all references to IntelFrameworkModulePkg in this module.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Instead of indirecting the reference to the Shell binary via a PCD
that is defined in IntelFrameworkModulePkg, and which invariably
gets set to the same value by all users of this library, refer to
the UEFI Shell application by its declared symbolic GUID.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Commit e7b24ec978 ("ArmPkg/UncachedMemoryAllocationLib: map uncached
allocations non-executable") adds code that manipulates the GCD memory
space attributes of a newly allocated uncached region without checking
whether this region expose these attributes in its capabilities mask.
Given that the intent is to remove executable permissions from the region,
this is a fairly pointless exercise to begin with, regardless of whether
it is correct or not. The reason is that RO/XP memory attributes in the
GCD memory space map or the UEFI memory map are completely disconnected
from the actual mapping permissions used in the page tables.
So instead, invoke the CPU arch protocol directly, and add the non-exec
attributes in the page tables directly.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Tested-by: Ryan Harkin <ryan.harkin@linaro.org>
modsi3.S references the symbol '__divsi3' by '___divsi3' which assumes
the prefix is always required and supported. Use ASM_PFX() instead
to support all compilers.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Marvin Haeuser <Marvin.Haeuser@outlook.com>
The primary use case for UncachedMemoryAllocationLib is non-coherent DMA,
which implies that such regions are not used to fetch instructions from.
So let's map them as non-executable, to avoid creating a security hole
when the rest of the platform may be enforcing strict memory permissions
on ordinary allocations.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Uncached pool allocations are aligned to the data cache line length under
the assumption that this is sufficient to prevent cache maintenance from
corrupting adjacent allocations. However, the value to use in such cases
is architecturally called the Cache Writeback Granule (CWG), which is
essentially the maximum Dcache line length rather than the minimum.
Note that this is mostly a cosmetical fix, given that the pool allocation
is turned into a page allocation later, and rounded up accordingly.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
In order to play nice with platforms that use strict memory permission
policies, restore the original mapping attributes when freeing uncached
allocations.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Now that we have the prerequisite functionality available in ArmMmuLib,
wire it up into ArmSetMemoryRegionNoExec, ArmClearMemoryRegionNoExec,
ArmSetMemoryRegionReadOnly and ArmClearMemoryRegionReadOnly. This is
used by the non-executable stack feature that is configured by DxeIpl.
NOTE: The current implementation will not combine RO and XP attributes,
i.e., setting/clearing a region no-exec will unconditionally
clear the read-only attribute, and vice versa. Currently, we
only use ArmSetMemoryRegionNoExec(), so for now, we should be
able to live with this.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
We no longer make use of the ArmMmuLib 'feature' to create aliased
memory ranges with mismatched attributes, and in fact, it was only
wired up in the ARM version to begin with.
So remove the VirtualMask argument from ArmSetMemoryAttributes()'s
prototype, and remove the dead code that referred to it.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
... where it belongs, since AARCH64 already keeps it there, and
non DXE users of ArmMmuLib (such as DxeIpl, for the non-executable
stack) may need its functionality as well.
While at it, rename SetMemoryAttributes to ArmSetMemoryAttributes,
and make any functions that are not exported STATIC. Also, replace
an explicit gBS->AllocatePages() call [which is DXE specific] with
MemoryAllocationLib::AllocatePages().
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
The routines ArmConfigureMmu(), SetMemoryAttributes() [*] and the
various set/clear read-only/no-exec routines are declared as returning
EFI_STATUS in the respective header files, so align the definitions with
that.
* SetMemoryAttributes() is declared in the wrong header (and defined in
ArmMmuLib for AARCH64 and in CpuDxe for ARM)
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
The debug implementation of the UncachedMemoryAllocationLib library
class relies on the creation of an uncached alias of a memory range,
while keeping the original cached mapping, but with read-only attributes
to trap inadvertent write accesses.
This is not a terribly good idea, given that the ARM architecture does
not allow mismatched attributes, and so creating them deliberately is
not something we should encourage by doing it in reference code.
So remove the library, and replace all references to it with a reference
to the non-debug version (unless the platform does not require a resolution
for it in the first place, in which case all UncachedMemoryAllocationLib
references can be removed altogether).
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Enable the hardware stack alignment check, as mandated by the UEFI spec.
This ensures that the stack pointer is 16 byte aligned at each instance
where it is used as the base address in a load/store operation.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
In preparation of enabling stack alignment checking, which is mandated
by the UEFI spec for AARCH64, add the code to manage this bit to ArmLib.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Stack and unstack the frame pointer according to the AAPCS in
AArch64AllDataCachesOperation ().
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Since the new DXE page protection for PE/COFF images may invoke
EFI_CPU_ARCH_PROTOCOL.SetMemoryAttributes() with only permission
attributes set, add support for this in the AARCH64 MMU code.
Move the EFI_MEMORY_CACHETYPE_MASK macro to a shared location between
CpuDxe and ArmMmuLib so we don't have to introduce yet another
definition.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Current Arm CpuDxe driver uses EFI_MEMORY_WP for write protection,
according to UEFI spec, we should use EFI_MEMORY_RO for write protection.
The EFI_MEMORY_WP is the cache attribute instead of memory attribute.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jiewen Yao <jiewen.yao@intel.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
This reverts commit d32702d2c2.
Using a pool allocation for the root translation table seemed like
a good idea at the time, but as it turns out, such allocations are
handled in a way that makes them unsuitable for this purpose: they
are backed by HOBs that don't remain in the same place during the
various PI phase changes, which means the address programmed into
the TTBR register is no longer valid, and may refer to memory that
is reported as available to the OS.
So switch back to using a page based allocation.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
The generic timer support libraries call the actual system register
accessor function via a single pair of functions ArmArchTimerReadReg()
and ArmArchTimerWriteReg(), which take an enum argument to identify
the register, and return output values by pointer reference.
Since these functions are never called with a non-immediate argument,
we can simply replace each invocation with the underlying system register
accessor instead. This is mostly functionally equivalent, with the
exception of the bounds check for the enum (which is pointless given the
fact that we never pass a variable), the check for the presence of the
architected timer (which only makes sense for ARMv7, but is highly unlikely
to vary between platforms that are similar enough to run the same firmware
image), and a check for enum values that refer to the HYP view of the timer,
which we never referred to anywhere in the code in the first place.
So get rid of the middle man, and update the ArmGenericTimerPhyCounterLib
and ArmGenericTimerVirtCounterLib implementations to call the system
register accessors directly.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Tested-by: Ryan Harkin <ryan.harkin@linaro.org>
Commit 0a99a65d2c ("fix incorrect device address of double buffer")
retained an explicit cast on the variable "Buffer" which became
incorrect with the other changes, leading to compilation failures
with some toolchains. Drop the cast.
Contributed-under: TianoCore Contribution Agreement 1.0
Reported-by: Sudeep Holla <sudeep.holla@arm.com>
Signed-off-by: Leif Lindholm <leif.lindholm@linaro.org>
Tested-by: Sudeep Holla <sudeep.holla@arm.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Some devices, such as the Raspberry Pi3, have a fixed offset between memory
addresses as seen by the host and as seen by the other bus masters. So add
a new PCD that allows this fixed offset to be recorded, and to be used when
returning device addresses from the DmaLib mapping routines.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Tested-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
In preparation of adding support to ArmDmalib for DMA bus masters whose
view of memory is offset by a constant compared to the CPU's view, clean
up some abuse of the device address.
The device address is not defined in terms of the CPU's address space,
and so it should not be used in CopyMem () or cache maintenance operations
that require a valid mapping. This not only applies to the above use case,
but also to the DebugUncachedMemoryAllocationLib that unmaps the
primary, cached mapping of an allocation, and returns a host address
which is an uncached alias offset by a constant.
Since we should never access the device address from the CPU, there is
no need to record it in the MAPINFO struct. Instead, record the buffer
address in case of double buffering, since we do need to copy the contents
(in case of a bus master write) and free the buffer (in all cases) when
DmaUnmap() is called.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Tested-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
If double buffering is not required in DmaMap(), the returned device
address is passed through ConvertToPhysicalAddress () to convert the
host address (which in case of DebugUncachedMemoryAllocationLib is not
1:1 mapped) to a physical address, which is what a device would expect
to be able to perform DMA.
By the same reasoning, a double buffer allocated using DmaAllocateBuffer ()
should be converted in the same way, considering that the buffer is allocated
using UncachedAllocatePages (), to which the above equally applies.
So add the missing ConvertToPhysicalAddress () invocation.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Tested-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Instead of depending on ArmLib to retrieve the CWG directly, use
the DMA buffer alignment exposed by the CPU arch protocol. This
removes our dependency on ArmLib, which makes the library a bit
more architecture independent.
While we're in there, rename gCpu to mCpu to better reflect its
local scope, and reflow some lines that we're modifying anyway.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Tested-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Translation table walks are always cache coherent on ARMv8-A, so cache
maintenance on page tables is never needed. Since there is a risk of
loss of coherency when using mismatched attributes, and given that memory
is mapped cacheable except for extraordinary cases (such as non-coherent
DMA), restrict the page table walker to performing cacheable accesses to
the translation tables.
For DEBUG builds, retain some of the logic so that we can double check
that the memory holding the root translation table is indeed located in
memory that is mapped cacheable.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
Michael Zimmermann
Ard Biesheuvel
Jian J Wang
Peicong Li
Supreeth Venkatesh
Sami Mujawar
Jun Nie
Leif Lindholm
Marvin Häuser
Jiewen Yao
Daniil Egranov