Since cbmem is not initialized in bootblock, CAR_GLOBAL variables
can only be accessed directly similar to verstage.
Change-Id: Ifc705016290807c49dc8c49b581864cac2ad3f80
Signed-off-by: Andrey Petrov <andrey.petrov@intel.com>
Reviewed-on: https://review.coreboot.org/13641
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins)
Some platforms may want to use C code in bootblock so they need
writable memory and CAR can be used for it. This change reserves
memory in CAR that can be used by bootblock and other CAR stages.
Change-Id: I8dec768cf8763dbe235f0ba1339079ebc49cbd9a
Signed-off-by: Andrey Petrov <andrey.petrov@intel.com>
Reviewed-on: https://review.coreboot.org/13640
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins)
The PSTATE mask bits for Debug exceptions, external Aborts, Interrupts
and Fast interrupts are usually best left unset: under normal
circumstances none of those exceptions should occur in firmware, and if
they do it's better to get a crash close to the code that caused it
(rather than much later when the kernel first unmasks them). For this
reason arm64_cpu_init unmasks them right after boot. However, the EL2
payload was still running with all mask bits set, which this patch
fixes.
BL31, on the other hand, explicitly wants to be entered with all masks
set (see calling convention in docs/firmware-design.md), which we had
previously not been doing. It doesn't seem to make a difference at the
moment, but since it's explicitly specified we should probably comply.
BRANCH=None
BUG=None
TEST=Booted Oak, confirmed with raw_read_daif() in payload that mask
bits are now cleared.
Change-Id: I04406da4c435ae7d44e2592c41f9807934bbc802
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: 6ba55bc23fbde962d91c87dc0f982437572a69a8
Original-Change-Id: Ic5fbdd4e1cd7933c8b0c7c5fe72eac2022c9553c
Original-Signed-off-by: Julius Werner <jwerner@chromium.org>
Original-Reviewed-on: https://chromium-review.googlesource.com/325056
Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/13596
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
On ARM64, the memory type for accessing page table descriptors during
address translation is governed by the Translation Control Register
(TCR). When the MMU code accesses the same descriptors to change page
mappings, it uses the standard memory type rules (defined by the page
table descriptor for the page that contains that table, or 'device' if
the MMU is off).
Accessing the same memory with different memory types can lead to all
kinds of fun and hard to debug effects. In particular, if the TCR says
"cacheable" and the page tables say "uncacheable", page table walks will
pull stale entries into the cache and later mmu_config_range() calls
will write directly to memory, bypassing those cache lines. This means
the translations will not get updated even after a TLB flush, and later
cache flushes/evictions may write the stale entries back to memory.
Since page table configuration is currently always done from SoC code,
we can't generally ensure that the TTB is always mapped as cacheable.
We can however save developers of future SoCs a lot of headaches and
time by spot checking the attributes when the MMU gets enabled, as this
patch does.
BRANCH=None
BUG=None
TEST=Booted Oak. Manually tested get_pte() with a few addresses.
Change-Id: I3afd29dece848c4b5f759ce2f00ca2b7433374da
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: f3947f4bb0abf4466006d5e3a962bbcb8919b12d
Original-Change-Id: I1008883e5ed4cc37d30cae5777a60287d3d01af0
Original-Signed-off-by: Julius Werner <jwerner@chromium.org>
Original-Reviewed-on: https://chromium-review.googlesource.com/323862
Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/13595
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Instead of instructing users to edit xcompile when they want to build
a quark platform, give the build a way to set -march=586 so that
the quark code will build correctly. The Quark processor does not
support the instructions introduced with the Pentium 6 architecture.
Change-Id: I0ed69aadc515f86f76800180e0e33bcd75feac5a
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/13552
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Tested-by: build bot (Jenkins)
Reviewed-by: Leroy P Leahy <leroy.p.leahy@intel.com>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: FEI WANG <wangfei.jimei@gmail.com>
Some newer x86 systems can boot from non-memory-mapped boot media
(e.g. EMMC). The bootblock may be backed by small amounts of SRAM, or
other memory, similar to how most ARM chipsets work. In such cases, we
may not have enough code space for romstage very early on. This means
that CAR setup and early boot media (e.g. SPI, EMMC) drivers need to
be implemented within the limited amount memory of storage available.
Since the reset vector has to be contained in this early code memory,
the bootblock is the best place to implement loading of other stages.
Implement a bootblock which does the minimal initialization, up to,
and including switch to protected mode. This then transfers control
to platform-specific code. No stack is needed, and control is
transferred via a "jmp" such that no stack operations are involved.
Change-Id: I009b42b9a707cf11a74493bd4d8c189dc09b8ace
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: https://review.coreboot.org/13485
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
bootblock.S was used strictly for setting up the system so that the
assembly generated by ROMCC could be executed. Since the
infrastructure now exists to run a bootblock wihtout ROMCC, rename
this file accordingly. this is done to prevent any future confusion.
Change-Id: Icbf5804b66b9517f9ceb352bed86978dcf92228f
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: https://review.coreboot.org/11784
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
If coreboot's build process is reproducible (eg. using the latest git
timestamp as source), bl31 is, too.
This requires an arm-trusted-firmware side merge first (in progress) and
an update of our reference commit for the submodule, but it also doesn't
hurt anything because it merely sets a variable that currently goes
unused.
Change-Id: If139538a2fab5b3a70c67f4625aa2596532308f7
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Reviewed-on: https://review.coreboot.org/13497
Tested-by: build bot (Jenkins)
Reviewed-by: Alexander Couzens <lynxis@fe80.eu>
Instead of tagging object files with .<class>, move them to a <class>
directory below $(obj)/. This way we can keep a 1:1 mapping between
source- and object-file names.
The 1:1 mapping is a prerequisite for Ada, where the compiler refuses
any other object-file name.
Tested by verifying that the resulting coreboot.rom files didn't change
for all of Jenkins' abuild configurations.
Change-Id: Idb7a8abec4ea0a37021d9fc24cc8583c4d3bf67c
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/13181
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins)
There were several spots in the tree where the path to a per class
object file was hardcoded. To make use of the src-to-obj macro for
this, it had to be moved before the inclusion of subdirs. Which is
fine, as it doesn't have dependencies beside $(obj).
Tested by verifying that the resulting coreboot.rom files didn't change
for all of Jenkins' abuild configurations.
Change-Id: I2eb1beeb8ae55872edfd95f750d7d5a1cee474c4
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/13180
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
When C_ENVIRONMENT_BOOTBLOCK is selected link bootblock using the
memlayout.ld scripts and infrastructure. This allows bootblock on
x86 to utilize all the other coreboot infrastructure without
relying romcc.
Change-Id: Ie3e077d553360853bf33f30cf8a347ba1df1e389
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/13069
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com>
Replace with the more familiar AT&T syntax.
Tested by sha1sum(1)ing the object files, and checking the objdump that
the code in question was actually compiled.
Change-Id: Ie85b8ee5dad1794864c18683427e32f055745221
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Reviewed-on: https://review.coreboot.org/13132
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
This provides symbols needed by CBFS and FMAP APIs, and allows running
run_romstage() in an x86 bootblock. Note that console-related files
are not added in this patch, as they are not essential for the
functinality on an x86 environment bootbock.
Change-Id: I36558b672a926ab22bc9018cd51aee32213792c2
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: https://review.coreboot.org/12880
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Instead of depending BOOTBLOCK_CONSOLE on a set of architectures,
allow the arch or platform to specify whether it can provide a C
environment. This simplifies the selection logic.
Change-Id: Ia3e41796d9aea197cee0a073acce63761823c3aa
Signed-off-by: Alexandru Gagniuc <alexandrux.gagniuc@intel.com>
Reviewed-on: https://review.coreboot.org/12871
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
This header is only used for the bootblock compiled with ROMCC. As the
follow-on patches introduce a bootblock which does not make use of
ROMCC, rename this header to prevent confusion.
Change-Id: Id29c5bc6928c11cc7cb922fcfac71e5a3dcd113c
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: https://review.coreboot.org/12867
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Trivial fix for syntax highlighting in editors. Some get confused by
the double quote that doesn't have a close quote and stop highlighting
at that point. This comment closes the quote and the paren pair so
that they can recover.
Change-Id: I2bdb7c953a86905fc302d77eb9ad1200958800b7
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/13017
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
This just updates existing guard name comments on the header files
to match the actual #define name.
As a side effect, if there was no newline at the end of these files,
one was added.
Change-Id: Ia2cd8057f2b1ceb0fa1b946e85e0c16a327a04d7
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/12900
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Most of these files are original to coreboot and get the standard
coreboot GPL header.
encoding.h and atomic.h are from the riscv codebase and have their
license.
Change-Id: I32506b0ecf88be2f5794dc1e312a6cd9b2a271ad
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/12906
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
This was breaking the build on OS X, but also wasn't working correctly
under linux anymore either. It wouldn't print the illegal symbols
when it failed.
- Split the generation of the offenders file from the actual check for
offending symbols and just send all output to /dev/null.
- Rewrite the check for offending symbols in a way that works with OS X.
Tested by adding a global variable to romstage and verifying the
failure is shown correctly. Verified that it works correctly with no
illegal variables.
Change-Id: I5b3ac32448851884d78c3b3449508ffe014119ab
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/13018
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
These were mostly written as part of the coreboot project, so get
the standard coreboot license header.
memmove.c came from the linux kernel, so also gets the standard
coreboot v2 license header, but gets the added attribution that it
was derived from the linux kernel. Unlike many coreboot files,
this file may not be re-licensed as GPL V3.
Change-Id: I1fdc26b543e059f7a42d4b886f7222f4c74b959d
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/12916
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
These were all written as part of the coreboot project, so get
the standard coreboot license header.
Change-Id: I51e1e504b3bc7be2a00c9356d8775b87f2a1db5a
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/12912
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
These were all written as part of the coreboot project, so get
the standard coreboot license header.
Change-Id: I4fccc8055755816be64e9e1a185f1e6fcb2b89ae
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/12911
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
These were all written as part of the coreboot project, so get
the standard coreboot license header.
Change-Id: I74438e8032c84f4190ef49f306969f7157234001
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/12910
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
When microcode updates are enabled, this fixes an issue identical
to that described in GIT hash 7b22d84d:
* drivers/pc80: Add optional spinlock for nvram CBFS access
Change-Id: Ib7e8cb171f44833167053ca98a85cca23021dfba
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: https://review.coreboot.org/12063
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
When used with a U-boot payload it will need this region
identity mapped also, so we're defining it in preparation
for that functionality.
Change-Id: I27cee5b58cb899433b52bd06df07b5f2105212af
Signed-off-by: Ionela Voinescu <ionela.voinescu@imgtec.com>
Reviewed-on: https://review.coreboot.org/12768
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
In order for a U-boot payload to work properly the soc_registers
region (device registers) needs to be mapped as uncached.
Therefore, add a coherency argument to the identity mapping funcion
which will establish the type of mapping.
Change-Id: I26fc546378acda4f4f8f4757fbc0adb03ac7db9f
Signed-off-by: Ionela Voinescu <ionela.voinescu@imgtec.com>
Reviewed-on: https://review.coreboot.org/12769
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
When enabling the IOMMU on certain systems dmesg is spammed with I/O page faults like the following:
AMD-Vi: Event logged [IO_PAGE_FAULT device=00:14.0 domain=0x000a address=0x000000fdf9103300 flags=0x0030]
Decoding the faulting address:
0x000000fdf9103300
fdf91x Hypertransport system management region
33 SysMgtCmd (System Management Command) = 0x33
3 Base Command Type = 0x3: STPCLK (Stop Clock request)
3 SMAF (System Management Action Field) = [3:1] = 0x1
1 Signal State Bit Map = [0] = 0x1
Therefore, the error appears to be triggered by an upstream C1E request.
This was eventually traced to concurrent access to the SP5100's SPI Flash controller by
multiple APs during startup. Calls to the nvram read functions get_option and read_option
call CBFS functions, which in turn make near-simultaneous requests to the SPI Flash
controller, thus placing the SP5100 in an invalid state. This limitation is not documented
in any public AMD errata, and was only discovered through considerable debugging effort.
Change-Id: I4e61b1ab767b1b7958ac7c1cf20eee41d2261bef
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: https://review.coreboot.org/12061
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
Section 6.1.3 (Text Strings) of the SMBIOS specification states:
If a string field references no string, a null (0) is placed in that
string field.
Change smbios_add_string() to do that.
Change-Id: I9c28cb89dcfe2c8ef2366c23ee6203e15b7c2513
Signed-off-by: Ben Gardner <gardner.ben@gmail.com>
Reviewed-on: https://review.coreboot.org/12697
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
$(objgenerated)/empty would touch files before the directory
is created on parallel builds.
Thanks to reproducible-builds.org for hitting this bug.
Change-Id: I7565e9fe130b4e9deaf1c7b9d568ff90b00dda52
Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
Reviewed-on: https://review.coreboot.org/12717
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
The ALIGN_CURRENT macro relied on a local variable name
as well as being defined in numerous compilation units.
Replace those instances with an acpi_align_current()
inline function.
Change-Id: Iab453f2eda1addefad8a1c37d265f917bd803202
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/12707
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Now that only CBFS access is supported for finding resources
within the boot media the assets infrastructure can be removed.
Remove it.
BUG=chromium:445938
BRANCH=None
TEST=Built and ran on glados.
Change-Id: I383fd6579280cf9cfe5a18c2851baf74cad004e9
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/12690
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The Chrome OS verified boot path supported multiple CBFS
instances in the boot media as well as stand-alone assets
sitting in each vboot RW slot. Remove the support for the
stand-alone assets and always use CBFS accesses as the
way to retrieve data.
This is implemented by adding a cbfs_locator object which
is queried for locating the current CBFS. Additionally, it
is also signalled prior to when a program is about to be
loaded by coreboot for the subsequent stage/payload. This
provides the same opportunity as previous for vboot to
hook in and perform its logic.
BUG=chromium:445938
BRANCH=None
TEST=Built and ran on glados.
CQ-DEPEND=CL:307121,CL:31691,CL:31690
Change-Id: I6a3a15feb6edd355d6ec252c36b6f7885b383099
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/12689
Tested-by: build bot (Jenkins)
Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
- Move initialization of entry to later in main.
- Make boot_mode an unsigned char - no need to use int.
- Remove unnecessary variable filenames.
- Only get and try to boot fallback once.
Change-Id: I823092c60dd8c2de0a36ec7fdbba3e68f6b7567a
Test: compiled.
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/12574
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
We need mmu interfaces in these two stages for,
1. bootblock: to support mmu initialization in bootblock
2. romstage: to be able to add dram range to mmu table
BRANCH=none
BUG=none
TEST=build pass
Change-Id: I56dea5f958a48b875579f546ba17a5dd6eaf159c
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: cf72736bda2233f8e0bdd7a8ca3245f1d941ee86
Original-Change-Id: I1e27c0a0a878f7bc0ff8712bee640ec3fd8dbb8b
Original-Signed-off-by: Jimmy Huang <jimmy.huang@mediatek.com>
Original-Reviewed-on: https://chromium-review.googlesource.com/292665
Original-Commit-Ready: Yidi Lin <yidi.lin@mediatek.com>
Original-Tested-by: Yidi Lin <yidi.lin@mediatek.com>
Original-Reviewed-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/12585
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Tested-by: build bot (Jenkins)
We currently race in SMM init on Atom 230 (and potentially
other CPUs). At least on the 230, this leads to a hang on
RSM, likely because both hyperthreads mess around with
SMBASE and other SMM state variables in parallel without
coordination. The same behaviour occurs with Atom D5xx.
Change it so first APs are spun up and sent to sleep, then
BSP initializes SMM, then every CPU, one after another.
Only do this when SERIALIZE_SMM_INITIALIZATION is set.
Set the flag for Atom CPUs.
Change-Id: I1ae864e37546298ea222e81349c27cf774ed251f
Signed-off-by: Patrick Georgi <patrick@georgi-clan.de>
Signed-off-by: Damien Zammit <damien@zamaudio.com>
Reviewed-on: https://review.coreboot.org/6311
Tested-by: build bot (Jenkins)
Tested-by: BSI firmware lab <coreboot-labor@bsi.bund.de>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Instead of having to remember to strip the quotes everywhere so that
string comparisons (of which there are a few) match up, do it right at
the beginning.
Fixes building the image with a .config where CONFIG_CBFS_PREFIX
contains quotes.
Change-Id: I4d63341cd9f0bc5e313883ef7b5ca6486190c124
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Reviewed-on: https://review.coreboot.org/12578
Tested-by: build bot (Jenkins)
Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com>
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Instead of having to have an ifeq() all across the code base,
use $(target-objcopy). And correct target-objcopy to a value
that objcopy actually understands.
Change-Id: Id5dea6420bee02a044dc488b5086d109e806d605
Signed-off-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-on: http://review.coreboot.org/11090
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
