Age | Commit message (Collapse) | Author |
|
'PciEmulation' is not a proper naming for a driver, which
uses NonDiscoverableDeviceRegistrationLib for handling
the devices attached via internal bus of the SoC.
Rename it to MvNonDiscoverableDxe and move under
Silicon/Marvell/Drivers directory.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
|
|
MvEeprom driver nesting into additional Devices directory
under Silicon/Marvell/Drivers/I2c is redundant. Align
its level with MvI2cDxe.
Change MvEeprom to MvEepromDxe in order to be
more consistent in Marvell drivers naming.
On the occasion add sorting in the .inf file.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
|
|
MvPhyDxe driver nesting into additional Phy directory
under Silicon/Marvell/Drivers/Net is redundant. Align
its level with MvMdioDxe and Pp2Dxe.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
|
|
Hitherto MvSpiDxe driver name was very generic. In order to
be ready for adding new SPI master drivers, use the controller's
traditional name (it's called SPI Orion in Linux and
U-Boot) for files and the entry point.
Additionally, move the files to new 'MvSpiOrionDxe' directory
and rename the 'Devices' and 'Variables' to reflect actual drivers'
names. Change MvSpiFlash to MvSpiFlashDxe in order to be
more consistent. On the occasion add sorting in the .inf files.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
|
|
Hitherto fd file name was pretty generic, so use one,
which reflects actual board model.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
|
|
As a part of files reorganization, switch to using '7k8k'
in all SoC-specific driver/library code instead of
'70x0'/'7040', so that to ensure consistent naming for
entire SoC family.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
|
|
In edk2-platforms it is expected to provide a separation between
SoC and boards files in 'Silicon' and 'Platform' directories
accordingly.
This patch aligns Marvell code to this requirement with no functional
changes in the actual source files, unless required due to modified
paths. Change the supported board's files names to proper Armada70x0Db.
Also rename 'Armada' directory to 'Armada7k8k' in order to properly
refer to the SoC family and prevent confusion in future, when
adding new Armada machines. On the occasion add ARM copyright,
which was wrongly missing in the dsc.inc file.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
|
|
This patch removes unused macros defined in MvPhyDxe.h, as well
as improves the style and comments. Pick single definition
of the autonegotiation timeout - two different macros were used
for the same purpose.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
fupdate command's usage information referred to a deprecated
'-f' flag in 'examples' section. It was a residue from the
initial version of the application, removed during review
before merging to upstream branch. Correct the help information
and also on the occasion improve the message and its style.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Proper watchdog control base is 0xf0610000, so fix the
incorrect value.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
After removal of the tftp library in EDK2, the tftp was
disabled on Armada platform. Re-enable this functionality
as a dynamic command on Armada 70x0 DB board. For this
purpose add it as an option, depending on a new
INCLUDE_TFTP_COMMAND parameter, which can be passed
in the command line during build time.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
The function ArmPlatformInitializeSystemMemory () has been removed from
ArmPlatformLib, so remove all the [empty] implementations provided by
the various platforms.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Wire up the non-volatile EFI variable store support, by switching from
the emulation driver to the real one. Define default values for
memory mapped SPI access, which must be configured by the early
firmware. In order to ensure proper execution, configure initialization
order with Depex entries.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
This patch applies necessary modifications, which allow to use
MvSpiDxe driver in variable support as a runtime service.
The driver's type is modified to DXE_RUNTIME_DRIVER, as well as
a new callback is introduced as a part of the SpiMasterProtocol.
It configures the memory space for mmio access to the host
controller registers.
Apply locking in the driver only during boot services. Once at
runtime, resource protection is handled by the operating system.
Moreover ensure proper execution order before MvSpiFlashDxe
(and hence MvFvbDxe) by setting according Depex dependency.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
This patch applies necessary modifications, which allow to use
MvSpiFlash driver in variable support as a runtime service.
Its type is modified to DXE_RUNTIME_DRIVER, as well as
an event is created, which converts the pointers to the
SpiMasterProtocol and its routines. In order to ensure proper
execution of the MvFvbDxe driver, configure initialization
order with Depex entry.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
MvFvbDxe driver introduces non-volatile EFI variable support
for Armada platforms. It relies on memory-mapped SPI read access.
Implementation of EFI_FIRMWARE_VOLUME_BLOCK2_PROTOCOL
is done with using existing Marvell SPI infrastructure
(SpiMasterProtocol and SpiFlashProtocol), thanks to which
this driver will be able to support various combinations of
flash devices and host controllers.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Cello, Overdrive, Overdrive1000 and Armada still include the
now-defunct library implementation of the Shell tftp command.
Since the command is a nonstandard one anyway, just remove
it from the build for these platforms.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Leif Lindholm <leif.lindholm@linaro.org>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
|
|
Remove all gEmbeddedTokenSpaceGuid.PcdEmbeddedPrompt assignments, which
are no longer meaningful with EBL removed.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
None of these platforms still include EBL, but some references
remained to its support libraries. Get rid of that.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
In the MvSpiDxe driver obtaining host register base address,
controller clock and device maximum frequency directly from PCDs
was done all over the code. This patch cleans up the parameters'
handling and enables accessing them from SPI_DEVICE structure fields.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Spansion SPI flash devices use different command for bank
selection. Update it, basing on the first byte of flash ID.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Current usage of sf command requires running 'sf probe' prior to
executing any other option. Because it is done in two separate steps,
it turned out that SpiMasterProtocol->SetupDevice could easily
overwrite valid Slave pointer when performing second operation.
Fix the issue by allocating Slave device only once and keep it
as global variable in the SpiTool application. This patch
also updates FirmwareUpdate command to follow the modified
SetupDevice operation.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Flash commands macros are already defined in MvSpiFlash.h, so
remove them from the protocol header.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Hitherto mechanism of fixing SPI flash model in the PCDs,
occured to be very inefficient and problematic. Enable
dynamic detection by reworking MvSpiFlashReadId() command,
which now uses newly added NorFlashInfoLib, that helps to
obtain description of the JEDEC compliant devices.
This patch updates the MvSpiFlashProtocol ReadId() protocol
callback on both producer's (MvFlashDxe) and consumers' sides
(FirmwareUpdate and SpiTool applications). Because all
information about detected SPI NOR flash is now stored in
the obtained NorFlashInfo structure fields, use them instead
of the PCDs.
Enable compilation of the NorFlashInfoLib and update
PortingGuide documentation accordingly.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Fix the ReadId routine by using master's ReadWrite callback
instead of the raw Transfer - no longer swapping and byte
shifting is needed. Simplify code by using local array
instead of dynamic allocation. Moreover store the FlashId
in an UINT8 array PCD instead of the concatenated UINT32
format - this way less overhead in the driver is needed
for comparing the buffers.
The new handling allowed for cleaning Fupdate and Sf
shell commands FlashProbe routines.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Changing controller's FIFO default values is not necessary and
possibly can cause instabilities, when using some devices.
Disable the modification and rely on initial settings.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Some SdMmc host controllers are run by clocks with different
frequency than it is reflected in Capabilities Register 1.
Because the bitfield is only 8 bits wide, a maximum value
that could be obtained from hardware is 255(MHz).
In case the actual frequency exceeds 255MHz, the 8-bit BaseClkFreq
member of SD_MMC_HC_SLOT_CAP structure occurs to be not sufficient
to be used for setting the clock speed in SdMmcHcClockSupply
function.
This patch adds new UINT32 array ('BaseClkFreq[]') to
SD_MMC_HC_PRIVATE_DATA structure for specifying
the input clock speed for each slot of the host controller.
All routines that are used for clock configuration are
updated accordingly.
Thanks to above the Xenon host controller driver
could be modified to configure clock speed relatively
to actual 400MHz input.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
This patch fixes incorrect settings for UHS mode in
SD_MMC_HC_HOST_CTRL2 register for SDR50 and SDR25, of which
the latter was missing. This field should be set to:
0x4 for DDR52
0x2 for SDR50
0x1 for SDR25
0x0 for others.
This way EmmcSwitchToHighSpeed function is on par with Linux
set_uhs_signaling routine in the Xenon driver.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Currently initial forcing link status happened for all ports, not only
marked as 'always-up'. Although this didn't actually matter for the MAC
settings, because MAC is automatically updated with PHY HW polling
feature of the controller, perform mv_gop110_fl_cfg only when
the appropriate flag is true. Also in such case, force the link as up,
using a new library routine.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
The MppSel definition PCDs contain 0xFF placeholders for values that
should be left untouched. MppLib needs to be taught how to take those
into account.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
MppLib may be used very early (in SEC), at which point stack protection
measures are more likely to cause harm than help, given that not even
the UART has been configured to the point where we can complain usefully.
So just disable it.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
After enabling dynamic PCDs, it is possible to reconfigure
MPP during platform initialization. It occurred that due to
a faulty way of passing temporary values, information obtained
from PCDs was overwritten. This patch fixes the issue, which
on the occasion simplifies PcdToMppRegs function.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Joe Zhou <shjzhou@marvell.com>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
During each transaction start, clearing the I2C_CONTROL_FLAG
was surrounded by 3 uncoditional stalls. This was not necessary,
so replace them with one busy-wait loop, whose polling
count could be also safely reduced.
Above improvements result in faster transfer initialization
and allow to reduce the I2C bus occupation.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: David Greeson <dgreeson@cisco.com>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
In MvI2cStartRequest the status was assigned to the variable
without dereferencing a pointer. Fix it.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Although the I2C transaction routines were prepared to
return their status, they were never used. This could
cause bus lock-up e.g. in case of failing to send a
slave address, the data transfer was attempted to be
continued anyway.
This patch fixes faulty behavior by checking transaction
status and stopping it immediately, once the fail
is detected. On the occasion fix style around modified
functions calls.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: David Greeson <dgreeson@cisco.com>
[Style adjustment and cleanup]
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Update the included components and library classes to make this platform
build for 32-bit ARM.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Add an ARM implementation of ArmPlatformHelper.S.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Instead of using hardcoded value in PcdSystemMemorySize PCD,
obtain DRAM size directly from SoC registers, which are filled
by firmware during early initialization stage.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
The default MemoryInitPeiLib implementation insists on reserving the
region occupied by our own FV, while this is not necessary at all (the
compressed payload is uncompressed elsewhere, so the moment we enter
DXE core, we don't care about the FV contents in memory)
So clone MemoryInitPeiLib and modify it to suit our needs.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
The Armada 70x0/80x0 DRAM controller allows a single window of DRAM
to be remapped to another location in the physical address space.
Since DRAM starts at address 0, and normally exceeds 4GB on these
systems, this allows us to free up some memory in the 32-bit
addressable region for peripheral MMIO and PCI MMIO32 and CONFIG
spaces.
This patch adjusts memory blocks to the configuration done in ARM-TF.
The remap parameters are otained directly from the registers. Moreover,
the configuration space base address is now configurable via PCD, so
that to satisfy a case, when remap is not enabled in the early firmware
and ensure, that PcdSystemMemorySize is not overlapping it.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
When switching to generic PSCI reset library, obsolete parts
of previous custom reset library (PCDs, documentation) remained.
Remove them.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
In order to prevent fragmentation of the UEFI memory map, increase the
sizes of the preallocated regions. Note that this does not increase the
memory footprint of UEFI, it just modifies it allocation policy to keep
similar region types together.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Add an implementation of EFI_RNG_PROTOCOL so that the OS loader has
access to entropy for KASLR and other purposes (i.e., seeding the OS's
entropy pool very early on).
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
This driver allows automatic booting via the network.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
The FDFs no longer require explicit alignment for sections containing
aligned objects, so change it to 'Auto' and FIXED (which allows some
padding to be removed), and remove some other cruft while at it.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
There is no point in using the MPCore PrePi, given that only the primary
core will enter UEFI at EL2, and the secondaries will be held in EL3
until summoned by the OS. So use the unicore flavour instead.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Remove the gEfiMdePkgTokenSpaceGuid.PcdPerformanceLibraryPropertyMask
setting so it reverts to its default of 0, and disables performance
profiling.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
The GIC architecture mandates that the CPU interface, which consists
of 2 consecutive 4 KB frames, can be mapped using separate mappings.
Since this is problematic on 64 KB pages, the MMU-400 aliases each
frame 16 times, and the two consecutive frames can be found at offset
0xf000.
Therefore use the last alias from the first series of aliases as the
base address, so that the first frame from the second series becomes
directly adjacent, whilst remaining covered by a separate 64KB page.
This patch is intended to expose correct GICC alias via
MADT, once ACPI support is added.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
For some reason, one of the early ARM platforms disabled all the
diagnostics related to the UEFI driver model, resulting in the
output of UEFI shell utilities such as 'devices' or 'drivers' to
become completely useless. Armada's shared .DSC include file
inherited this for no good reason, so let's revert it.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|
|
Switch from the Intel BDS to the generic BDS, which is preferred for
ARM platforms given that it is completely legacy free.
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
|