Alex Kiernan [Thu, 5 Jul 2018 12:38:15 +0000 (12:38 +0000)]
env: Include bootdelay in environment if negative
The test for (CONFIG_BOOTDELAY >= 0) has been in U-Boot since the
beginning, but the meaning of it has changed over time. Allow the
default to be set for any value, including -ve ones. This allows
(for example) CONFIG_ENV_IS_NOWHERE to have values for bootdelay in
its compiled in environment.
The only thing this changes is where the default for bootdelay can be
fetched from; before this change you get a compiled in default, after
you'll pull it from the default value in the environment, but both values
will be the same. Also if there's a value set in the environment then
that will take precedence (as before).
Signed-off-by: Alex Kiernan <alex.kiernan@gmail.com>
Marek BehĂșn [Wed, 4 Jul 2018 18:23:01 +0000 (20:23 +0200)]
fs: btrfs: Fix wrong comparison in logical to physical mapping
The comparison
logical > item->logical + item->length
in btrfs_map_logical_to_physical is wrong and should be instead
logical >= item->logical + item->length
For example, if
item->logical = 4096
item->length = 4096
and we are looking for logical = 8192, it is not part of item (item is
[4096, 8191]). But the comparison is false and we think we have found
the correct item, although we should be searing in the right subtree.
board/BuR/brppt1: convert brppt1 boards to driver model
- add a devicetree for each variant (mmc, spi, nand)
- drop unneeded code from board and bur/common
- drop unneeded stuff from config header files
- minor adaptions to be compliant with driver model (requesting gpio,..)
- harmonize the commandset over all brppt1 targets
board/BuR/brppt1: implement more flexible boot process
With this commit we do:
- set the bootdelay in all brppt1 defconfigs to 0, this makes
development easier, since we can break into serial console.
- move CONFIG_BOOTCOMMAND from header file to defconfig
- introduce b_mode variable for selecting the final boot-target.
This b_mode represents the boot-switch, which can found on most b&r
targets. On the brppt1 this boot-switch is derived from some gpio and
the bootcounter within the RTC block, making it so possible to force
a boot-target (as example for repair-case).
- refactor the environment for booting new flexible way
primary we want to get some bootscr.img within the mass-storage,
this script then loads everything needed for the boot.
For legacy reason we implement the t30lgcy#x boot targets, booting the
already delivered linux-images.
- make space for the cfgscr within mtdparts on brppt1_nand
On other OS, not one provided by B&R, it is not guaranteed that there
are factory-settings within a devicetree. So we must not treat the
absence of them as error.
Further we've the fact that on different version of the device-tree
files there are different namings of the factory-settings, we consider
this with searching for an alternative name.
changing things as following:
- don't treat as error if the bootloader version cannot written into
devicetree.
- since the naming of the factory-settings are different in different
versions of the provided device-tree we search for the alternate name
"/fset"
If a board-code calls the pmicsetup(u32 mpupll) with a mpupll value
!= 0 it wants to force some frequency with the value provided by mpupll.
Setting up 1 GHz is wrong here.
Nobody did take notice about that yet, since every board calls this
function with zero.
board/BuR/common: remove interface Label from summary screen
This interface names may vary over different products, to consider this
fact we replace the interface label "IF1" and "IF2" on the summary
screen with some more generic wording "MAC1" and "MAC2".
On this linux target long time ago the OS is using DRM driver for
handling video output, the pre initialization of u-boot and the display
summary screen is obsolete. With this patch we drop the LCD-support from
thisd board.
Since we're going to drop LCD-support on brppt1 boards, we have to make
this stuff here optional and remove the #error path.
We also move out the ft_board_setup(...) from this #ifdef because
there's no relationship with the LCD-code and on the other hand this is
still needed in future even with LCD-support off.
The linux systems running on the brppt1 targets are using modern DRM
drivers since long time ago. Further we are going to drop the LCD
support completely on this board, so the simple-framebuffer setup
becomes obsolete.
board/BuR: drop devicetree loading and lcd setup for linux-targets
This patch drops the lcd-screen setup, the summary screen and getting
mac-addresses based on a previous loaded device-tree for linux targets.
Selecting those linux target is simple, since we have only the brppt1.
In detail we do:
- drop the common lcd-setup code which relys on a fdt_blob
- drop the common dtb loading mechanism
- drop the now obsolete CONFIG_USE_FDT from board header and whitelist.
Ramon Fried [Sun, 1 Jul 2018 23:57:56 +0000 (02:57 +0300)]
soc: qualcomm: Add Shared Memory Manager driver
The Shared Memory Manager driver implements an interface for allocating
and accessing items in the memory area shared among all of the
processors in a Qualcomm platform.
Adapted from the Linux driver (4.17)
Changes from the original Linux driver:
* Removed HW spinlock mechanism, which is irrelevant
in U-boot particualar use case, which is just reading from the smem.
* Adapted from Linux driver model to U-Boot's.
Cc: Bjorn Andersson <bjorn.andersson@linaro.org> Signed-off-by: Ramon Fried <ramon.fried@gmail.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Ramon Fried [Sun, 1 Jul 2018 23:57:55 +0000 (02:57 +0300)]
dm: SMEM (Shared memory) uclass
This is a uclass for Shared memory manager drivers.
A Shared Memory Manager driver implements an interface for allocating
and accessing items in the memory area shared among all of the
processors.
Signed-off-by: Ramon Fried <ramon.fried@gmail.com> Reviewed-by: Simon Glass <sjg@chromium.org> Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
QEMU provides an emulated ARM AMBA PrimeCell PL031 RTC.
The patch sets the base address in the board include file according to the
definition in hw/arm/virt.c of the QEMU source. It defines the Kconfig
option for the existing driver, and enables the RTC driver in
qemu_arm64_defconfig and qemu_arm_defconfig as well as the date command.
We need an RTC to provide the GetTime() runtime service in the UEFI
subsystem.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Reviewed-by: Tuomas Tynkkynen <tuomas.tynkkynen@iki.fi> Tested-by: Tuomas Tynkkynen <tuomas.tynkkynen@iki.fi> Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
arm/arm64: bitops: fix find_next_zero_bit to be compat with arm64
Current implementation of find_next_zero_bit() is incompatible with arm64.
Hence fix it by using BITS_PER_LONG define instead of constants and
use generic ffz() implementation.
This patch adds a delay when regulators are disabled.
This delay is set to 5 ms to cover all use cases.
The worst use case actually seen is during a SD card power cycle.
generic_phy_get_by_index() must be updated to first look for
PHY phandle as previously and in case of error looks for PHY
provider by finding the parent's current node which is the PHY
provider.
args (ofnode_phandle_args struct) must also be updated by inserting
the phy index into the PHY provider as args[0].
Patrick Delaunay [Wed, 27 Jun 2018 08:15:33 +0000 (10:15 +0200)]
mmc: stm32_sdmmc2: update pwron management
Correctly manage the SDMMC reset and card cycle power
to fully handle the power cycle added in the MMC uclass
and avoid issue with level-shifter with some uSDCARD.
3 states managed in driver:
1/ reset: SDMMC disable, signal HiZ
2/ power-cycle: SDMMC disable, signals drive to 0
3/ power-on: SDMMC enabled
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com> Signed-off-by: Patrice Chotard <patrice.chotard@st.com>
Yaniv Levinsky [Sun, 24 Jun 2018 16:16:57 +0000 (19:16 +0300)]
env: common: accept flags on reset to default env
The function set_default_env() sets the hashtable flags for import_r().
Formally set_default_env() doesn't accept flags from its callers. In
practice the caller can (un)set the H_INTERACTIVE flag, but it has to be
done using the first character of the function's string argument. Other
flags like H_FORCE can't be set by the caller.
Change the function to accept flags argument. The benefits are:
1. The caller will have to explicitly set the H_INTERACTIVE flag,
instead of un-setting it using a special char in a string.
2. Add the ability to propagate flags from the caller to himport(),
especially the H_FORCE flag from do_env_default() in nvedit.c that
currently gets ignored for "env default -a -f" commands.
3. Flags and messages will not be coupled together. A caller will be
able to set flags without passing a string and vice versa.
Please note:
The propagation of H_FORCE from do_env_default() does not introduce any
functional changes, because currently himport_r() is set to destroy the
old environment regardless if H_FORCE flag is set or not. More changes
are needed to utilize the propagation of H_FORCE.
Signed-off-by: Yaniv Levinsky <yaniv.levinsky@compulab.co.il> Acked-by: Igor Grinberg <grinberg@compulab.co.il>
Yaniv Levinsky [Sun, 24 Jun 2018 16:16:56 +0000 (19:16 +0300)]
cmd: nvedit: set H_INTERACTIVE in do_env_default
The function set_default_vars() in common.c adds H_INTERACTIVE to the
h_import() flag, but the function has no way of telling if the command
actually was user directed like this flag suggest. The flag should be
set by the calling function do_env_default() in nvedit.c instead, where
the command is certainty user directed.
Move the H_INTERACTIVE flag from set_default_vars() to do_env_default().
Signed-off-by: Yaniv Levinsky <yaniv.levinsky@compulab.co.il> Acked-by: Igor Grinberg <grinberg@compulab.co.il>
Yaniv Levinsky [Sun, 24 Jun 2018 16:16:55 +0000 (19:16 +0300)]
cmd: nvedit: propagate envflag to set_default_vars
The env_flag in do_env_default() doesn't get propagated and therefore
gets ignored by himport_r(). This breaks to ability to "forcibly" reset
variables to their default values using the environment command.
Scenario example of the problem:
# setenv kernel uImage
# setenv .flags kernel:so
# env default -f kernel
## Error: Can't overwrite "kernel"
himport_r: can't insert "kernel=zImage" into hash table
Change the call path so it will pass the flag correctly.
Signed-off-by: Yaniv Levinsky <yaniv.levinsky@compulab.co.il> Acked-by: Igor Grinberg <grinberg@compulab.co.il>
Yaniv Levinsky [Sun, 24 Jun 2018 16:16:54 +0000 (19:16 +0300)]
cmd: nvedit: rename flags in do_env_default
The naming convention for flags in nvedit.c is:
* The hashtable flag (defined in search.h) is named "env_flag"
* The command flag argument (defined in command.h) is named "flag"
This convention is kept in functions like do_env_print(), do_env_set()
and do_env_delete(), but not in do_env_default().
Rename the hashtable flag in do_env_default() from "flag" to "env_flag".
Rename the command flag in do_env_default() from "__flag" to "flag".
No functional change.
Signed-off-by: Yaniv Levinsky <yaniv.levinsky@compulab.co.il> Reviewed-by: Igor Grinberg <grinberg@compulab.co.il>
Alex Kiernan [Fri, 22 Jun 2018 14:58:02 +0000 (14:58 +0000)]
cmd: fit_image: Add default property for FIT format scripts
When sourcing a FIT format script, if we've not been told the unit name
to use, look for a default property at the root of /images to work out
which unit we should use.
Signed-off-by: Alex Kiernan <alex.kiernan@gmail.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Tom Rini [Thu, 19 Jul 2018 15:48:33 +0000 (11:48 -0400)]
Merge tag 'xilinx-for-v2018.09' of git://git.denx.de/u-boot-microblaze
Xilinx changes for v2018.09
clk:
- Fix zynqmp clock driver
common:
- Handle CMD_RET_USAGE in cmd_process_error
- Use return macros in cmd_process_error
- Fix duplication of CONFIG_SYS_PROMPT_HUSH_PS2
- Support watchdog in usb_kbd.c
- Fix name usage in usb_kbd.c
- Support systems with non zero memory start initialized from DT only
gpio:
- Add support for manual relocation in uclass
- zynq - use live tree
- zynq - fix match data reading
- zynq - setup bank name
- xilinx - convert driver to DM
microblaze:
- Use generic iounmap/ioremap implementations
- Redesign reset logic with sysreset features
- Use watchdog and gpio over DM
- Remove unused macros and fix some checkpatch issues
- Fix timer initialization not to be called twice
serial:
- zynq - Use platdata intead of priv data
sysreset:
- Add support for manual relocation in uclass
- Add gpio-restart driver
- Add microblaze soft reset driver
watchdog:
- Add support for aliases in uclass
- Add support for manual relocation in uclass
- Convert xilinx driver to DM
- cadence - update info in the driver and not stop wdt in probe
xilinx:
- Enable LED gpio for some targets with gpio-leds DT node
- Setup variables via Kconfig
zynq:
- Add support for watchdog aliases
- Add support for mini nand/nor configurations
- Wire FPGA initalization in SPL
zynqmp:
- Enable mass storage for zcu100
- Handle external pmufw files
- Add support for secure images
- Some Kconfig movements and alignments
- Add support for watchdog aliases
- Use subcommands style for platform command
- Add mmio_read/write platform commands
- DT updates
- Add support for mini qspi configuration
Luis Araneda [Thu, 19 Jul 2018 07:10:18 +0000 (03:10 -0400)]
arm: zynq: spl: fix FPGA initialization
commit 4aba5fb857c1 ("arm: zynq: Rework FPGA initialization")
moved FPGA initialization from board_init() to arch_early_init_r(),
which is not called as part of the SPL
Fix this by calling arch_early_init_r() in the spl_board_init()
function, so the FPGA is correctly initialized
Signed-off-by: Luis Araneda <luaraneda@gmail.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Luis Araneda [Thu, 19 Jul 2018 07:10:17 +0000 (03:10 -0400)]
drivers: fpga: zynqpl: fix compilation with SPL
Disable the use of function zynq_loadfs when compiling
the driver for the SPL, as the following filesystem
functions are not found by the linker:
- fs_set_blk_dev
- fs_read
- fs_set_blk_dev
- fs_read
- fs_read
Signed-off-by: Luis Araneda <luaraneda@gmail.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
arm64: zynqmp: Add QSPI flash mini u-boot configuration
Add configuration files/dtses for mini u-boot configuration
which runs on smaller footprint of internal memory. This
configuration has only required qspi flash support and it
uses DCC as serial.
Signed-off-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
arm: zynq: Add parallel NOR flash mini u-boot configuration for zynq
Add configuration files/dtses for mini u-boot configuration
which runs on smaller footprint OCM memory. This configuration
only has required parallel nor flash support.
Signed-off-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
arm: zynq: Add Nand flash mini u-boot configuration for zynq
Add configuration files/dtses for mini u-boot configuration
which runs on smaller footprint of memory. This configuration
has only required nand flash support.
Signed-off-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
lib: fdtdec: Update ram_base to store ram start adddress
This patch updates the ram_base to store the start address of
the first bank DRAM and the use this ram_base to calculate ram_top
properly. This patch fixes the erroneous calculation of ram_top
incase of non zero ram start address.
Signed-off-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com> Reviewed-by: Tom Rini <trini@konsulko.com>
Michal Simek [Tue, 17 Jul 2018 11:17:39 +0000 (13:17 +0200)]
watchdog: cadence: Do not stop wdt in probe
Watchdog can be started before probe and u-boot should just take control
over it. That's why do not stop watchdog in probe to cover cases where
watchdog can expire before probe and start.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
arm64: zynqmp: Added support of mmio read and write commands
This patch added support of mmio read and write commands. These commands
can be used to read and write registers from the u-boot command line.
It can be useful in debugging.
Signed-off-by: Vipul Kumar <vipul.kumar@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Michal Simek [Mon, 16 Jul 2018 11:34:55 +0000 (13:34 +0200)]
watchdog: cdns: Add comment for expire_now function
IP itself has no reg/no bit which can be used for this functionality.
Add this note to the driver to make sure that none will be asking for
that. Current method is to setup 1s timeout and hang() which is done via
wdt_expire_now().
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Michal Simek [Fri, 13 Jul 2018 06:26:28 +0000 (08:26 +0200)]
microblaze: Convert generic platform to DM gpio
Converting GPIO to DM requires to do changes in reset subsystem
that's why support for Microblaze soft reset via sysreset and GPIO
sysreset support was added.
These two patches enables enabling GPIO DM.
Microblaze soft reset is bind at last reset method.
GPIO reset is handled via sysreset with adding this fragment to DT.
Michal Simek [Thu, 12 Jul 2018 14:05:46 +0000 (16:05 +0200)]
gpio: xilinx: Convert driver to DM
This patch is enabling GPIO_DM support to have an option to use this
driver together with zynq gpio driver.
!DM part is kept there till Microblaze is cleanup which will be done
hopefully soon.
Just a note:
There is no reason to initialize uc-priv->name because it is completely
unused.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Michal Simek [Fri, 13 Jul 2018 15:00:13 +0000 (17:00 +0200)]
sysreset: Add support for Microblaze soft reset jump
Microblaze is storing reset vector at address 0x0.
It means soft reset can be done by just jumping to this address.
This code was in platform code but sysreset interface is providing
enough capabilities to have more options how to reset the system. It can
go from gpio reset through watchdog reset till soft reset.
The driver has not compatible string because this is cpu specific and DM
core is not able to detect compatible string in DT root that's why this
driver will be instantiated from platform code by calling
device_bind_driver(gd->dm_root, "mb_soft_reset", "reset_soft",
NULL);
It should be bind as the last reset method to ensure that hw reset is
called before this.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Michal Simek [Fri, 13 Jul 2018 09:04:56 +0000 (11:04 +0200)]
sysreset: Add support for gpio-restart
The Linux kernel has binding for gpio-restart node.
This patch is adding basic support without supporting any optional
properties.
This driver was tested on Microblaze system where gpio is connected to
SoC reset logic.
Output value is handled via gpios cells values.
In gpio_reboot_request() set_value is writing 1 because
dm_gpio_set_value() is capable to changing it when it is ACTIVE_LOW.
...
if (desc->flags & GPIOD_ACTIVE_LOW)
value = !value;
...
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Reviewed-by: Simon Glass <sjg@chromium.org>
dm tree fragment for above configuration with patch applied:
ZynqMP> dm tree
Class index Probed Driver Name
-----------------------------------------
...
watchdog 0 [ ] cdns_wdt | |-- watchdog@ff150000
watchdog 1 [ + ] cdns_wdt | `-- watchdog@fd4d0000
...
Michal Simek [Wed, 11 Jul 2018 12:08:26 +0000 (14:08 +0200)]
microblaze: Do not call timer init that early
Timer needs to be converted to DM but as of now it can't be called so
early because intc controller is not ready. Call it later in board_r.c.
Before this patch timer_init is called twice which is wrong.
The patch is blocking initialization before relocation.
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Michal Simek [Thu, 12 Jul 2018 05:50:59 +0000 (07:50 +0200)]
arm64: zynqmp: Sync defconfigs in connection to DEFINE_TCM_OCM_MMAP
CONFIG_MP was added to Kconfig with enabling CONFIG_DEFINE_TCM_OCM_MMAP=y
for zynqmp boards. This option is enabled by default that's why it
shouldn't be in defconfig.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
This patch basically adds two new commands for loadig secure
images.
1. zynq rsa adds support to load secure image which can be both
authenticated or encrypted or both authenticated and encrypted
image in xilinx bootimage(BOOT.bin) format.
2. zynq aes command adds support to decrypt and load encrypted
image back to DDR as per destination address. The image has
to be encrypted using xilinx bootgen tool and to get only the
encrypted image from tool use -split option while invoking
bootgen.
Signed-off-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Michal Simek [Thu, 14 Jun 2018 08:32:27 +0000 (10:32 +0200)]
serial: zynq: Use platdata for storing static data instead of priv
Explanation from Simon Glass
"Private data is created when the device is probed and freed when the
device is removed.
Platform data is created when the device is bound, and survives
probe/remove cycles.
Strictly speaking, platform data should be used to hold the decoded
device tree properties. Private data should be used for run-time
things the device needs to keep track of."
Based on description the driver needs to be switch to use platdata
instead of priv.
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Luca Ceresoli [Fri, 22 Jun 2018 10:40:16 +0000 (12:40 +0200)]
arm/arm64: zynq/zynqmp: pass the PS init file as a kconfig variable
U-Boot needs to link ps7_init_gpl.c on Zynq or psu_init_gpl.c on
ZynqMP (PS init for short). The current logic to locate this file for
both platforms is:
1. if a board-specific file exists in
board/xilinx/zynq[mp]/$(CONFIG_DEFAULT_DEVICE_TREE)/ps?_init_gpl.c
then use it
2. otherwise use board/xilinx/zynq/ps?_init_gpl.c
In the latter case the file does not exist in the U-Boot sources and
must be copied in the source tree from the outside before starting the
build. This is typical when it is generated from Xilinx tools while
developing a custom hardware. However making sure that a
board-specific file is _not_ found (and used) requires some trickery
such as removing or overwriting all PS init files (e.g.: the current
meta-xilinx yocto layer).
This generates a few problems:
* if the source tree is shared among different out-of-tree builds,
they will pollute (and potentially corrupt) each other
* the source tree cannot be read-only
* any buildsystem must add a command to copy the PS init file binary
* overwriting or deleting files in the source tree is ugly as hell
Simplify usage by allowing to pass the path to the desired PS init
file in kconfig variable XILINX_PS_INIT_FILE. It can be an absolute
path or relative to $(srctree). If the variable is set, the
user-specified file will always be used without being copied
around. If the the variable is left empty, for backward compatibility
fall back to the old behaviour.
Since the issue is the same for Zynq and ZynqMP, add one kconfig
variable in a common place and use it for both.
Also use the new kconfig help text to document all the ways to give
U-Boot the PS init file.
Build-tested with all combinations of:
- platform: zynq or zynqmp
- PS init file: from XILINX_PS_INIT_FILE (absolute, relative path,
non-existing), in-tree board-specific, in board/xilinx/zynq[mp]/
- building in-tree, in subdir, in other directory
Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net> Cc: Albert Aribaud <albert.u.boot@aribaud.net> Cc: Michal Simek <michal.simek@xilinx.com> Cc: Nathan Rossi <nathan@nathanrossi.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Michal Simek [Thu, 21 Jun 2018 12:49:26 +0000 (14:49 +0200)]
common: command: Handle USAGE failure separately
command_ret_t enum contains 3 return values but only two are handled
now. Extend cmd_process_error() and handle CMD_RET_USAGE separately.
These commands are affected by this change.
cmd/demo.c
cmd/efi.c
cmd/gpio.c
cmd/qfw.c
cmd/x86/fsp.c
test/dm/cmd_dm.c
And scripts shouldn't be affected because return value is not 0. But
every command implementation can choose what it is correct to pass.
I would expect that RET_USAGE is called when parameters are not
correctly passed (have incorrect value, missing parameters)
and RET_FAILURE when correct parameters are passed but command fails.
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Reviewed-by: Simon Glass <sjg@chromum.org>