If the TPM completely fails to respond then the vendor structure may not
have assigned handlers yet, so catch that case and return error so the
boot can continue to recovery mode instead of asserting over and over.
Change-Id: If3a11567df89bc73b4d4878bf89d877974044f34
Signed-off-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: https://review.coreboot.org/16416
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Enable the I2C based TPM on the reef board at
bus 2 and address 0x50.
This makes vboot functional without needing MOCK_TPM and
results in the following in the SSDT:
Device (TPMI)
{
Name (_HID, "GOOG0005") // _HID: Hardware ID
Name (_UID, Zero) // _UID: Unique ID
Name (_DDN, "I2C TPM") // _DDN: DOS Device Name
Method (_STA, 0, NotSerialized) // _STA: Status
{
Return (0x0F)
}
Name (_CRS, ResourceTemplate () // _CRS: Current Resource Settings
{
I2cSerialBus (0x0050, ControllerInitiated, 0x00061A80,
AddressingMode7Bit, "\\_SB.PCI0.I2C2",
0x00, ResourceConsumer)
Interrupt (ResourceConsumer, Edge, ActiveLow, Exclusive)
{
0x00000039
}
})
}
Change-Id: Ia9775caabeac3e6a3bd72de38f9611b4cea7cea4
Signed-off-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: https://review.coreboot.org/16398
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Add code to generate an ACPI descriptor for an I2C TPM based
on the device as described in devicetree.cb.
This currently requires the devicetree to provide the HID,
since we don't currently talk to the TPM in ramstage and I
didn't want to add yet another init path for it here.
This was tested on a reef board to ensure that the device
is described properly in the SSDT.
Change-Id: I43d7f6192f48e99a4074baa4e52f0a9ee554a250
Signed-off-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: https://review.coreboot.org/16397
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Add support for the cr50 TPM used in apollolake chromebooks.
This requires custom handling due to chip limitations, which
may be revisited but are needed to get things working today.
- timeouts need to be longer
- must use the older style write+wait+read read protocol
- all 4 bytes of status register must be read at once
- same limitation applies when reading burst count from status reg
- burst count max is 63 bytes, and burst count behaves
slightly differently than other I2C TPMs
- TPM expects the host to drain the full burst count (63 bytes)
from the FIFO on a read
Luckily the existing driver provides most abstraction needed to
make this work seamlessly. To maximize code re-use the support
for cr50 is added directly instead of as a separate driver and the
style is kept similar to the rest of the driver code.
This was tested with the cr50 TPM on a reef board with vboot
use of TPM for secdata storage and factory initialization.
Change-Id: I9b0bc282e41e779da8bf9184be0a11649735a101
Signed-off-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: https://review.coreboot.org/16396
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Philipp Deppenwiese <zaolin.daisuki@googlemail.com>
Allow the sleep durations used by the driver to be set by the
specific chip so they can be tuned appropriately.
Since we need to read the chip id to know the values use very
conservative defaults for the first command and then set it
to the current values by default.
Change-Id: Ic64159328b18a1471eb06fa8b52b589eec1e1ca2
Signed-off-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: https://review.coreboot.org/16395
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins)
Reviewed-by: Furquan Shaikh <furquan@google.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Use CAR accessors where needed for accessing static data.
In some cases this required some minor restructuring to pass
in a variable instead of use a global one.
For the tpm_vendor_init the structure no longer has useful
defaults, which nobody was depending on anyway. This now
requires the caller to provide a non-zero address.
Tested by enabling I2C TPM on reef and compiling successfully.
Change-Id: I8e02fbcebf5fe10c4122632eda1c48b247478289
Signed-off-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: https://review.coreboot.org/16394
Tested-by: build bot (Jenkins)
Reviewed-by: Furquan Shaikh <furquan@google.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Fix a few more instances of global variable usage in the tlcl
and marshaling code for tpm2.
For the tlcl case this buffer doesn't need to be static as it
isn't used after this function exits.
Change-Id: Ia739c81d79c6cee9046ae96061045fe4f7fb7c23
Signed-off-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: https://review.coreboot.org/16393
Tested-by: build bot (Jenkins)
Reviewed-by: Philipp Deppenwiese <zaolin.daisuki@googlemail.com>
Place config DEVICETREE after the sourced mainboard Kconfig. This
gives the mainboard the opportunity to set a unique default value.
Change-Id: Id877e1e8f555334a99b6c0ee1782d06a4a2b7a04
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/16493
Tested-by: build bot (Jenkins)
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Use the config switch CONFIG_DEBUG_SPI_FLASH on compiler level rather
then on preprocessor level to ensure that the code is compiled even if
the switch is not selected. In addition the following two changes are
introduced:
1. Prepend the debug messages with 'SPI:' to make the output more
meaningful.
2. Change the address mask from 0xffff to 0x3ff and remove the subtraction
of the constant value 0xf020 in order to print only the register
offset within the SPI controller and avoid the visibility of any
fragments from SPI base address.
3. Switch to uint8_t and friends instead of u8 to sync up with other
code in the same file.
Change-Id: Iaf46f29a775039007a402fe862839df06a4cbfaa
Signed-off-by: Werner Zeh <werner.zeh@siemens.com>
Reviewed-on: https://review.coreboot.org/16499
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
The pull up for CLKRUN is required to resolve keyboard slowness
and malfunctioning observed on some reef systems. The CLKRUN
signal was probed and found to be floating when the pull up
was not enabled. Also Added pull ups for the LPC Multiplexed
command, address and data lines LAD0:3 because the LPC
Interface specification requires them.
BUG=chrome-os-partner:55586
BRANCH=none
TEST=When a key is pressed, the character is immediately visible
on the screen. Also the interrupt count for i8042 increments
immediately in /proc/interrupts.
Change-Id: I16df1a0301a3994c926a609f61291761219f9e01
Signed-off-by: Shamile Khan <shamile.khan@intel.com>
Reviewed-on: https://review.coreboot.org/16426
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins)
The last vestige of the proto boards is the memory sku id
gpios. The internal pullups are still required because there's
only pulldown stuffing options available on the reef boards.
BUG=chrome-os-partner:56791
Change-Id: I04d541a897ec9aacbf2011293d18242fa32896d2
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16432
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Furquan Shaikh <furquan@google.com>
Move the current NHLT configuration implementation to the baseboard
area such that other variants can leverage it or provide their
own configuration.
BUG=chrome-os-partner:56677
Change-Id: If0d48cacdc793492e1618d0eda02a149e33f0650
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16431
Tested-by: build bot (Jenkins)
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Move the current memory configuration implementation to the baseboard
area such that other variants can leverage it. The swizzle config
is exported as a global to allow duplicate swizzles to use the same
structure while still allowing different memory SKUs.
BUG=chrome-os-partner:56677
Change-Id: I57201118053051c01f0e3f164ab4bbaf650b892b
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16430
Tested-by: build bot (Jenkins)
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Add support for Chrome OS gpio ACPI table information by
providing weak implementation from the baseboard.
BUG=chrome-os-partner:56677
Change-Id: I517764b78f47fb7b3637482ff9efc053cdd1ac69
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16422
Tested-by: build bot (Jenkins)
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Since multiple boards will be living within one directory move all
the macros for defining anyting related to GPIOs to the gpio.h
header file. That way, when other boards land they can override
or use them as is.
BUG=chrome-os-partner:56677
Change-Id: I36967e57fc61ef354e0b51d1ff1396ce562fa805
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16421
Tested-by: build bot (Jenkins)
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
There's no common EC header file in the code base, and I didn't
want to use a header file for single declaration. Therefore,
just move the declaration to each file that uses that symbol.
BUG=chrome-os-partner:56677
Change-Id: Ibaebb0ea6a07029aec02d5185cf05ffb8593b117
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16420
Tested-by: build bot (Jenkins)
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Provide APIs for the board_id() and gpio table functionality.
Default and weak implementations are provided from the baseboard.
BUG=chrome-os-partner:56677
Change-Id: I02d8deb7f60f8c4842916a9d35f51d8af74b1da4
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16419
Tested-by: build bot (Jenkins)
Reviewed-by: Furquan Shaikh <furquan@google.com>
Commit e96543e1 (vendorcode/intel: Add UDK 2015 Bindings)
had an extra underscore at the end of one of the make lines that
we missed in the review. Remove it.
Fixes this build warning:
.../Makefile.inc:34: Extraneous text after `ifeq' directive
Change-Id: I0bc76d827207b4f641ac5ff08f540a114347533b
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/16411
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Alexander Couzens <lynxis@fe80.eu>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-by: Werner Zeh <werner.zeh@siemens.com>
A recently announced Turbot system populates two Ethernet
controllers. Enable the remaining disabled PCIe port.
Also add a clarifying comment regarding the port associated
with Function 0. Coreboot must not be allowed to disable the
function which breaks PCI compatibility.
Change-Id: I2815ba7e6d68b9898091fbc21c96eeeb49c8e05a
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/16429
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
MinnowBoard Turbot systems have a GPIO-controlled LED that is
generally used to indicate the CPU is running. Commit 2ae9cce8
changed the parameter for GPIO_NC, exposing an issue with the
assumed behavior of the signal. Use a pull-down to turn on the
LED.
Change-Id: I153870904c007d89016c0d47bb3db9b824ebbcff
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/16428
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
Taken from upstream Linux kernel.
Change-Id: Id8c9d48b8ca3c7592dc4d78cfd092714bbbb2abf
Signed-off-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-on: https://review.coreboot.org/15980
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
Replace the use of the old device_t definition inside
northbridge/intel/i945.
The patch has been tested both with the arch/io.h definition of
device_t enabled and disabled in order to ensure compatibility
while the transaction takes place.
Change-Id: I041c150a7b50261e26955ad9287ef05b9a06e412
Signed-off-by: Antonello Dettori <dev@dettori.io>
Reviewed-on: https://review.coreboot.org/16371
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Change the argument to #ifdef from __PRE_RAM__ to __SIMPLE_DEVICE__
in order to account for the coreboot stages that do not define device_t
and are not __PRE_RAM__ (i.e. smm) device_t
Change-Id: Ic6e9b504803622b60b5217c9432ce57caefc5065
Signed-off-by: Antonello Dettori <dev@dettori.io>
Reviewed-on: https://review.coreboot.org/16369
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Replace the use of the old device_t definition inside
northbridge/intel/sandybridge.
The patch has been tested both with the arch/io.h definition of device_t
enabled and disabled in order to ensure compatibility while the
transaction takes place.
Change-Id: I35cc76ec7b6baa216666d06f6f325f43ac69067e
Signed-off-by: Antonello Dettori <dev@dettori.io>
Reviewed-on: https://review.coreboot.org/16409
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
Avoid the inclusion of a function declaration if the argument type
device_t is not defined.
This was not a problem until now because the
old declaration of device_t and the new one overlapped.
Change-Id: I05a6ef1bf65bf47f3c6933073ae2d26992348813
Signed-off-by: Antonello Dettori <dev@dettori.io>
Reviewed-on: https://review.coreboot.org/16404
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
Replace the use of the old device_t definition inside
southbridge/intel/bd82x6x.
The patch has been tested both with the arch/io.h definition of device_t
enabled and disabled in order to ensure compatibility while the
transaction takes place.
Change-Id: I7166bfab7904f80b745855d3bbcfb910cbc89f56
Signed-off-by: Antonello Dettori <dev@dettori.io>
Reviewed-on: https://review.coreboot.org/16407
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
To further the ability of multiple variant boards to share code
provide a place to land the split up changes. This patch provides
the tooling using a new Kconfig value, VARIANT_DIR, as well as
the Make plumbing. The directory layout with a single variant,
reef (which is also the baseboard), looks like this:
variants/baseboard - code
variants/baseboard/include/baseboard - headers
variants/reef - code
variants/reef/include/variant - headers
New boards would then add themselves under their board name
within the 'variants' directory.
No split has been done with providing different logic yet.
This is purely a organizational change.
BUG=chrome-os-partner:56677
Change-Id: Ib73a3c8a3729546257623171ef6d8fa7a9f16514
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16418
Tested-by: build bot (Jenkins)
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Instead of completely duplicating the a reference board's directory
when doing a variant or follower device start providing a means to
share code within a single directory. This change just starts the
process from the Kconfig side, but subsequent patches will follow
which disentangles the board specific pieces from and common
logic.
BUG=chrome-os-partner:56677
Change-Id: I96628920d78012e488ec008e35daac9c1be0cf79
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16417
Tested-by: build bot (Jenkins)
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
The superio.asl wasn't being included within the right scope.
Fix that as well as clean up the per-mainboard header includes
to be in one place.
BUG=chrome-os-partner:56677
Change-Id: I5e6a82f9f2e3c7455132263d19b32b2f06220376
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16413
Tested-by: build bot (Jenkins)
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-by: Shaunak Saha <shaunak.saha@intel.com>
Some of the macros in gpio.h are no longer used because
devicetree.cb is being used to autogeneric the ACPI AML.
Therefore remove the unused macros.
BUG=chrome-os-partner:56677
Change-Id: I433a929229a0318f6c1df652655d046a5152cc63
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16412
Tested-by: build bot (Jenkins)
Reviewed-by: Sathyanarayana Nujella <sathyanarayana.nujella@intel.com>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Commit 93ef3ff makes the following only print the part number when
the ROM is built. In Makefile.inc, $(MAINBOARDDIR) is the variable
that has the quotes stripped off from $(CONFIG_MAINBOARD_DIR), so
use it instead of $(MAINBOARD_DIR).
build_complete:: coreboot
printf "\nBuilt %s (%s)\n" $(MAINBOARD_DIR) \
$(CONFIG_MAINBOARD_PART_NUMBER)
Change-Id: I729a583182937db7a926eb75aa28dfb53360046c
Signed-off-by: Iru Cai <mytbk920423@gmail.com>
Reviewed-on: https://review.coreboot.org/16410
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
Increase the default timeout in the LPSS I2C driver to 4ms
from 2ms. During testing with some slower devices I found
that the existing timeout could be too short leading to
transaction failures.
Change-Id: Ied86c7a0aa26d55b31f447c5938803c194d0045e
Signed-off-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: https://review.coreboot.org/16392
Reviewed-by: Furquan Shaikh <furquan@google.com>
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
- Don't require a description of every config statement in the
Kconfig changes
- Don't complain about externs in .c files
- Don't complain about the use of the volatile keyword. The kernel
may not want it, but we definitely need it.
- Disable checks that seem to be broken.
Change-Id: Ic419b81cd36852a91e887e610d4a04984ab5fbd7
Signed-off-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/16010
Tested-by: build bot (Jenkins)
Reviewed-by: Alexander Couzens <lynxis@fe80.eu>
FSP header files should be located in vendorcode, not soc directory.
This patch includes changes any references to the old location to
the new location.
Change-Id: I44270392617418ec1b9dec15ee187863f2503341
Signed-off-by: Brandon Breitenstein <brandon.breitenstein@intel.com>
Reviewed-on: https://review.coreboot.org/16310
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
The following patch is based off of the UEFI 2.6 patch. The FSP header files
are temporarily staying in soc/intel/apollolake and FspUpd.h has been relocated
since the other headers expect it to be in the root of an includable directory.
Any struct defines were removed since they are defined in the headers and no
longer need to be explicity declared as struct with the UEFI 2.6 includes.
BUG=chrome-os-partner:54100
BRANCH=none
TEST=confirmed coreboot builds successfully
Change-Id: I10739dca1b6da3f15bd850adf06238f7c51508f7
Signed-off-by: Brandon Breitenstein <brandon.breitenstein@intel.com>#
Signed-off-by: Andrey Petrov <andrey.petrov@intel.com>
Reviewed-on: https://review.coreboot.org/16308
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
UEFI 2.6 spec casts the return of FFS_FILE2_SIZE to a UINT32
which cannot be read using read_le32(&returnval). Add in a
cast in order to safeguard for any non x86 architecture that may
use this relocate. The proper change will be to get the UEFI
header files changed to not cast this return value.
Change-Id: Ie1b50d99576ac42a0413204bbd599bab9f01828e
Signed-off-by: Brandon Breitenstein <brandon.breitenstein@intel.com>
Reviewed-on: https://review.coreboot.org/16309
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Replace the use of the old device_t definition inside
mainboard/lenovo/x60.
The patch has been tested both with the arch/io.h definition of device_t
enabled and disabled in order to ensure compatibility while the
transaction takes place.
Change-Id: Icaceeae2fc7276efa82d37582ecac93aaf37c41c
Signed-off-by: Antonello Dettori <dev@dettori.io>
Reviewed-on: https://review.coreboot.org/16372
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Uses gpio.h instead of default_southbridge_gpio_setup to configure
southbridge GPIO's. This is more consistent with how GPIO's are
configured on newer targets.
Change-Id: I6ccd0564b929e958864739b7cde04f5592c58479
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/16379
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
The checksum command might appear to be unpredictable only by
checking the OS. Just list the candidates, sorted by possibility.
Change-Id: Ia3f4f5f0f98ff47d322a4f70689cca0bd4fa79fa
Signed-off-by: Zheng Bao <zheng.bao@amd.com>
Signed-off-by: Zheng Bao <fishbaozi@gmail.com>
Reviewed-on: https://review.coreboot.org/11483
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
All but ga-g41m-es2l/cmos.default had multiple final newlines.
ga-g41m-es2l/cmos.default had no final newline.
Change-Id: Id350b513d5833bb14a2564eb789ab23b6278dcb5
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/16361
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins)
Reviewed-by: Antonello Dettori <dev@dettori.io>
Instead of using a mix of _N and _NORTH, _NW and _NORTHWEST for GPIO
community names, follow one single convention. This allows for re-using
macros easily.
Change-Id: Icd9cf9ef70d03576d864688cf5d6946124c259c3
Signed-off-by: Furquan Shaikh <furquan@google.com>
Reviewed-on: https://review.coreboot.org/16353
Reviewed-by: Lijian Zhao <lijian.zhao@intel.com>
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Duncan Laurie