'name' and 'env' are supposed to be file system paths,
but could overflow the buffer if configured incorrectly.
Let's avoid that entirely.
Found-by: Coverity Scan #1362515
Signed-off-by: Jacob Garber <jgarber1@ualberta.ca>
Change-Id: I1aef36819d49ebcbde1c51995dc0961c85e74150
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32686
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Systems with C_EVIRONMENT_BOOTBLOCK and VBOOT enabled requires functions
tsc_freq_mhz(), vbnv_cmos_failed() and vboot_platform_is_resuming()
in verstage.
Add tsc_freq.c and pmutil.c to verstage.
BUG=NA
TEST=Booting Embedded Linux on Facebook FBG-1701 + Building Google Banon
Change-Id: Ia509eda6bf415aaa63be71013249493aa472289d
Signed-off-by: Frans Hendriks <fhendriks@eltan.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32733
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
It is possible that 'malloc_fo_sub' and 'remalloc_fo' can
fail, so add appropriate error checks for those cases.
This incidentally fixes a possible memory leak when
'malloc_fo_sub' succeeds but 'remalloc_fo' does not.
Found-by: Coverity Scan #1396050
Signed-off-by: Jacob Garber <jgarber1@ualberta.ca>
Change-Id: I944b67f5cdcfd7a687e81d8bb01a209c9dc9b0b8
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32696
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
As the previous comment indicated, this null check is
currently superfluous, but adding it in makes Coverity
happy, and future-proofs the code in case someone changes
the internals of 'find_fcba' later and forgets/doesn't know
to update this error check.
Found-by: Coverity Scan #1395066
Signed-off-by: Jacob Garber <jgarber1@ualberta.ca>
Change-Id: I594cd0098f5b36cef5b3efc4c904710d3ba9b815
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32691
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
This patch ensures start_cpu() function to store default_apic_id using
common cpu_add_map_entry() function to make cpu_index() implementation
generic.
BRANCH=none
BUG=b:79562868
Change-Id: Iac4d6e9e6e6f9ba644335b4b70da8689c405f638
Signed-off-by: Subrata Banik <subrata.banik@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32722
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
This cpu_index() implementation assumes that cpu_index() function
might always getting called from coreboot context (ESP stack
pointer will always refer to coreboot).
This might not be true in case of proposed PI spec MP_SERVICES_PPI
implementation, where FSP context (stack pointer refers to fsp)
will request to get cpu_index(), natural alignment logic will
use ESP and retrieve struct cpu_info *ci from (stack_top - 8 byte).
This is not the place where cpu_index is actually stored by
ramstage c_start.S
Hence this patch tries to remove those dependencies while retrieving
cpu_index(), rather it uses cpuid to fetch lapic id and matches with
cpus_default_apic_id[] variable to return correct cpu_index().
BRANCH=none
BUG=b:79562868
TEST=Ensures functions can be run on APs without any failure and
cpu_index() also provides correct index number.
Change-Id: I55023a3e0cf42f0496d45bc6af8ead447f402350
Signed-off-by: Subrata Banik <subrata.banik@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/26346
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
We keep the support, though. Just now that `libgfxinit` is fixed, we
don't need the distinction anymore. Causally, we also don't need
CPU_INTEL_MODEL_306AX any more.
TEST=Played tint on kontron/ktqm77. Score 606
Change-Id: Id1e33c77f44a66baacba375cbb2aeb71effb7b76
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32737
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Beside one tiny fix for framebuffer scaling, this contains a major
refactoring of libgfxinit's configuration infrastructure. With this,
we are finally able to detect CPUs at runtime and only have to confi-
gure a CPU/GPU generation.
Change-Id: Iccf4557453878536f527e4a1902439a1961ab701
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32736
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Beside some refactorings that don't affect coreboot, this contains
bd0ed91 (Makefile: Revise support for generated sources) that fixes
an issue with upcoming libgfxinit configuration changes.
Change-Id: Ib47aeff8f6426ae27ddbc235a954e3bd60029072
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32735
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The Kconfig MAINBOARD_FAMILY sets the family field of SMBIOS entry 1.
Match what vendor firmware does and use the same value as in the
version field.
Required for fwupd which uses the family field to generate a GUID.
Change-Id: I0033c42c5eac6b9d47d0acd16c67467b6d419534
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32727
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
This patch ensures to have correct timestamp value in postcar.
Change-Id: I3ba3a54c20dfcdaf5b87818cc5da9a812f5f2edf
Signed-off-by: Subrata Banik <subrata.banik@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32726
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
This patch enables coreboot flow to skip ramstage as
individual stage to load payload. Instead it is expected
to load payload from postcar stage.
Change-Id: I839f2d34a93b69ca6bf3de6594e2ad9f66ee7135
Signed-off-by: Subrata Banik <subrata.banik@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32618
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: ron minnich <rminnich@gmail.com>
This patch renames mp_get_apic_id() to cpu_get_apic_id() and
add_cpu_map_entry() to cpu_add_map_entry() in order access it
outside CONFIG_PARALLEL_MP kconfig scope.
Also make below changes
- Make cpu_add_map_entry() function available externally to call
it from mp_init.c and lapic_cpu_init.c.
BRANCH=none
BUG=b:79562868
Change-Id: I6a6c85df055bc0b5fc8c850cfa04d50859067088
Signed-off-by: Subrata Banik <subrata.banik@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32701
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Enable VT-d through fsp upd VtdDisable. Update remapping structure
types in numerical order as all remapping structures of type 0 (DRHD)
enumerated before remapping structures of type 1 (RMRR), and so forth.
BUG=b:130351429
TEST=Booted to kernel and verified the DMAR table contents.
Change-Id: I1d20932e417b9d324edd98c8f2195dc228d2e092
Signed-off-by: John Zhao <john.zhao@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32432
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Pratikkumar V Prajapati <pratikkumar.v.prajapati@intel.com>
We've been assuming that ENV_RAMSTAGE is always the payload loader.
In order to test out different models, we need a way to mark the
"stage we are in" as the payload loader.
Define a new rule, ENV_PAYLOAD_LOADER. For now, it is set
to ENV_RAMSTAGE. It is not used yet pending approval of this
approach.
Change-Id: I7d4aa71bad92987374d57ff350b9b0178ee7c12b
Signed-off-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32709
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This patch moves vboot_logic_executed() (and its dependencies) into a
header and turns it into a static inline function. The function is used
to guard larger amounts of code in several places, so this should allow
us to save some more space through compile-time elimination (and also
makes it easier to avoid undefined reference issues in some cases).
Change-Id: I193f608882cbfe07dc91ee90d02fafbd67a3c324
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32716
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Frans Hendriks <fhendriks@eltan.com>
Some edk2 vendorcode headers define an ASSERT macro. They're guarded
with an #ifndef ASSERT, but if coreboot's assert.h gets included after
that header, we still have a problem. Add code to assert.h to undefine
any rogue definitions that may have already been set by vendorcode
headers.
This is ugly and should only be a stopgap... it would be nice if someone
maintaining those vendorcode parts could eventually replace it with a
better solution. One option would be to use a "guard header" for every
vendorcode header we want to pull into normal coreboot code which would
chain-include the vendorcode header and then undefine anything that
clashes with coreboot again.
Change-Id: Ibf8dc8b2365821e401ce69705df20aa7540aefb2
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32715
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
This patch adds another check to vboot_logic_executed() to make sure we
only do a runtime check for verstage_should_load() if
CONFIG_VBOOT_RETURN_FROM_VERSTAGE is enabled. That's the only case where
the stage that's loading the verstage can execute after verification has
run (because the verstage will return to it when it's done). In the
other case, the stage that loads verstage really just loads it and will
never do anything again after hand-off, so it's guaranteed to always
execute before verification.
This change may allow extra dead-code elimination in some cases.
Change-Id: I7019b6f7b0acfbf0a8173914b53364751b08f2cf
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32714
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Replace locations in the source that explicitely use the CD6/CD7
index/data pair with utility function calls.
Change-Id: I6e7ba472ef2551e363987d18a79408fcd2074de4
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32648
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Leakage power is observed from TOUCH_SCREEN_PD# (GPP_E7 which is connected
to RESET pin of Wacom controller) during S5.
To avoid leakage power, GPP_E7 needs to be turned off before S5 entry.
BUG=b:129899315
TEST=Measure leakage power in S5 from both Arcada and Sarien
Signed-off-by: Lijian Zhao <lijian.zhao@intel.com>
Change-Id: Ie4229477b7149c0a75f4a8c6c7c453a37cc1c78c
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32367
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Roy Mingi Park <roy.mingi.park@intel.com>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
PCI_MIN_GNT is defined at offset 0x3e in <pci_def.h> which does not
apply to this PCI bridge because it is only defined for
"Header type 0 (normal devices)" (line 82).
Some lines obove that code line, the "write" on BCTRL is already done.
Change-Id: I8f1b98ba627947ab6652a4ba31d2acb159dd3e32
Signed-off-by: Elyes HAOUAS <ehaouas@noos.fr>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32700
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
PCI_MIN_GNT is defined at offset 0x3e in <pci_def.h> which does not
apply to this PCI bridge because it is only defined for
"Header type 0 (normal devices)" (line 82).
BCTRL registry for D30:F0 is defined at offset 0x3e for i82801gx
(see ICH7 Family Datasheet page 355).
The write on that register is already done some lines above.
So remove wrong register name and the wrong code line.
Change-Id: Ib8a0514200f424049503bb8e4bc076ee6ae86ce3
Signed-off-by: Elyes HAOUAS <ehaouas@noos.fr>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32699
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Add vbt files for nami variants and select Kconfig option
to utilize them. The default vbt is automatically added
by the Kconfig selection and so does not need to be
specified in the makefile with the others.
Test: boot vayne and akali nami variants, verify
display functional and correct vbt loaded.
Change-Id: Iaf49bdee7ae82a0a61192327351267f098eb5ab1
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/30427
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Set tcc offset value to 1 degree celsius for Arcada system.
BRANCH=None
BUG=b:122636962
TEST=Built and tested on Arcada system
Signed-off-by: Bonnie Lin <bonnie_ty_lin@wistron.corp-partner.google.com>
Change-Id: I3ca4be2f7b92e29fb133ecc32023526b177d2ac2
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32680
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Casper Chang <casper_chang@wistron.corp-partner.google.com>
Reviewed-by: Lijian Zhao <lijian.zhao@intel.com>
Reviewed-by: Sumeet R Pawnikar <sumeet.r.pawnikar@intel.com>
Correct Micron MT40A512M16TB-062E:J SPD CRC to 0x5330 to fix post hang
in AGESA TestPoint:05 TpProcMemSPDChecking.
BUG=b:127394249
BRANCH=master
TEST=emerge-grunt coreboot chromeos-bootimage
Signed-off-by: Kevin Chiu <Kevin.Chiu@quantatw.com>
Change-Id: I8fa49e6e938b3195945b3199438cc53f3e9c92e7
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32678
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Enable early SMBus support compatible with SPD library using Intel SB
common SMBus API.
TEST=boot Protectli FW2B with new FSP, MemoryInit should pass without
errors
Signed-off-by: Michał Żygowski <michal.zygowski@3mdeb.com>
Change-Id: I92a2c5a6d0b38e5658cfdc017041f12717dabdd5
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32062
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Frans Hendriks <fhendriks@eltan.com>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Cannonlake PCH LP have total 6 pcie clocks and Cannonlake PCH H have
total 16 pcie clocks. It is different with pcie root port numbers.
BUG=CID 1381814
TEST=Build and boot up fine on sarien platform.
Signed-off-by: Lijian Zhao <lijian.zhao@intel.com>
Change-Id: I909b5b584c596e6fe878ffe24d9cabc53c4576ed
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32672
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: John Zhao <john.zhao@intel.com>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Furquan Shaikh <furquan@google.com>
Reviewed-by: Subrata Banik <subrata.banik@intel.com>
The ASL 2.0 syntax for "!X" resolves to "LNot(X)" which will evaluate
the object as an integer and turn into a boolean. This may not do the
right thing if the object is actually a string and it can lead to
unexpected behavior.
Instead be specific about the object type and check for zero or an
empty string depending on what is being returned.
This fixes an issue where some VPD keys were causing the search to
stop and miss subsequent entries.
Change-Id: I1688842964f9c2f81ca31073da9c2d71a8c81767
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32694
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
The definition of bootblock_systemagent_early_init was guarded by
!ENV_RAMSTAGE. But it's only called in the bootblock. So guard it
with ENV_BOOTBLOCK instead.
Change-Id: I143cf72e4a63b176e4772575e7a60a2a611e4ad9
Signed-off-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32697
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Lijian Zhao <lijian.zhao@intel.com>
Reviewed-by: Subrata Banik <subrata.banik@intel.com>
Should include vb2_sha.h header when SHA library functions or
constants are required. This replaces NEED_VB2_SHA_LIBRARY.
BUG=b:124141368, chromium:956474
TEST=make clean && make test-abuild
BRANCH=none
Change-Id: I9f32174dbf3de05fbe5279cb8017888757abf368
Signed-off-by: Joel Kitching <kitching@google.com>
Cq-Depend: chromium:1583820
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32454
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Input: tell vb2api_fw_phase1 if display unconditionally available
Output: vb2api_fw_phase1 may request coreboot to initialize
display, if needed based on some internal request
Move setting the VBOOT_FLAG_DISPLAY_REQUESTED flag into
verstage_main.
BUG=b:124141368, b:124192753, chromium:948529
TEST=make clean && make test-abuild
BRANCH=none
Change-Id: I81c82c46303564b63b8a32e7f80beb9d891a4628
Cq-Depend: chromium:1564232
Signed-off-by: Joel Kitching <kitching@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32324
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
The VbInitParams struct will be deprecated. Remove its
use in preparation.
Additionally, remove use of the flag
VB_INIT_OUT_ENABLE_USB_STORAGE, which is no longer used
downstream since vboot_reference CL:347257.
BUG=b:124141368, chromium:960226
TEST=make clean && make test-abuild
BRANCH=none
Change-Id: Ibe02cb6ba639de0d7cbdf79fc4dbf49044c92278
Signed-off-by: Joel Kitching <kitching@google.com>
Cq-Depend: chromium:1583943
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32664
Reviewed-by: Furquan Shaikh <furquan@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Use CondRefOf to replace config optios for PTS/WAK acpi method dynamic
loading. Then we can move EC PTS and WAK method to be under mainboard.
BUG=N/A
TEST=Build sarien source code, check build/dsdt.dsl have EC.PTS method
included, build whlrvp soure, check build/dsdt.dsl don't have EC.PTS
method. Both able to build pass.
Signed-off-by: Lijian Zhao <lijian.zhao@intel.com>
Change-Id: I9f4bd7240832caf070e65039e4ba2d8656371da8
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32371
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add functions to read and write the region in the AcpiMmio block.
Convert gpio.c to use them instead of creating pointers.
Change-Id: I2a0f44b6ec7261648cf0357b44a6c18dd40d1504
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32647
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Add 16 and 32-bit versions of read / write_misc functions. Find one
access of the MISC block still using read8() and write8(), and convert
it.
Change-Id: I296c521ea7f43210db406013bbe79362545ce6f3
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32646
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Add 8-bit functions to access the AOAC registers and use them in
southbridge.c. At this time, there is no reason to pursue WORD or
DWORD access and it's not known if those transaction sizes are
supported.
Change-Id: I3a8f493625f941fb855c0b8a0eff511a9a5ddfe8
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32645
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-by: Martin Roth <martinroth@google.com>
Convert smbus_read8() and smbus_write8() functions to use the same
arguments as the other AcpiMmio blocks, and add 16 and 32 bit versions.
Add matching functions for the ASF controller.
Change-Id: I3b0ecf21f20472245da98ab5e711a54e99dca93a
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32644
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-by: Martin Roth <martinroth@google.com>
Nico Huber