platform/kernel/u-boot.git
3 years agoMakefile: Pass the U-Boot phase to dtoc
Simon Glass [Wed, 3 Feb 2021 13:01:03 +0000 (06:01 -0700)]
Makefile: Pass the U-Boot phase to dtoc

Pass the U-Boot phase as a parameter so dtoc can use it. At present it is
ether "spl" or "tpl".

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Support tracking the phase of U-Boot
Simon Glass [Wed, 3 Feb 2021 13:01:02 +0000 (06:01 -0700)]
dtoc: Support tracking the phase of U-Boot

U-Boot operates in several phases, typically TPL, SPL and U-Boot proper.
The latter does not use dtoc.

In some rare cases different drivers are used for two phases. For example,
in TPL it may not be necessary to use the full PCI subsystem, so a simple
driver can be used instead.

This works in the build system simply by compiling in one driver or the
other (e.g. PCI driver + uclass for SPL; simple_bus for TPL). But dtoc has
no way of knowing which code is compiled in for which phase, since it does
not inspect Makefiles or dependency graphs.

So to make this work for dtoc, we need to be able to explicitly mark
drivers with their phase. This is done by adding an empty macro to the
driver. Add support for this in dtoc.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Track nodes which are actually used
Simon Glass [Wed, 3 Feb 2021 13:01:01 +0000 (06:01 -0700)]
dtoc: Track nodes which are actually used

Mark all nodes that are actually used, so we can perform extra checks on
them.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Process nodes to set up required properties
Simon Glass [Wed, 3 Feb 2021 13:01:00 +0000 (06:01 -0700)]
dtoc: Process nodes to set up required properties

Add logic to assign property values to nodes as required by dtoc. The
references allow nodes to refer to each other in C code. The macros used
by dtoc are not yet defined in driver model. They will be added along
with the actual driver model implementation.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Make use of node properties
Simon Glass [Wed, 3 Feb 2021 13:00:59 +0000 (06:00 -0700)]
dtoc: Make use of node properties

Now that we have these available, use them instead of recalculating
things each time.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Add some extra properties to nodes
Simon Glass [Wed, 3 Feb 2021 13:00:58 +0000 (06:00 -0700)]
dtoc: Add some extra properties to nodes

It is convenient to attach drivers, etc. to nodes so that we can use the
Node object as the main data structure in this module.

Add a function which adds the new properties, along with documentation.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Rename sandbox_i2c_test and sandbox_pmic_test
Simon Glass [Wed, 3 Feb 2021 13:00:57 +0000 (06:00 -0700)]
dtoc: Rename sandbox_i2c_test and sandbox_pmic_test

These have '_test' suffixes which are not present on the drivers in the
source code. Drop the suffixes to avoid a mismatch when scanning.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Move test files into a test/ directory
Simon Glass [Wed, 3 Feb 2021 13:00:56 +0000 (06:00 -0700)]
dtoc: Move test files into a test/ directory

It is confusing to have the test files in the same places as the
implementation. Move them into a separate directory.

Add a helper function for test_dtoc, to avoid repeating the same
path.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Support scanning of structs in header files
Simon Glass [Wed, 3 Feb 2021 13:00:55 +0000 (06:00 -0700)]
dtoc: Support scanning of structs in header files

Drivers can have private / platform data contained in structs and these
struct definitions are generally kept in header files. In order to
generate build-time devices, dtoc needs to generate code that declares
the data contained in those structs. This generated code must include the
relevant header file, to avoid a build error.

We need a way for dtoc to scan header files for struct definitions. Then,
when it wants to generate code that uses a struct, it can make sure it
includes the correct header file, first.

Add a parser for struct information, similar to drivers. Keep a dict of
the structs that were found.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Support scanning of uclasses
Simon Glass [Wed, 3 Feb 2021 13:00:54 +0000 (06:00 -0700)]
dtoc: Support scanning of uclasses

Uclasses can have per-device private / platform data so dtoc needs to
scan these drivers. This allows it to find out the size of this data so
it can be allocated a build time.

Add a parser for uclass information, similar to drivers. Keep a dict of
the uclasses that were found.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Collect priv/plat struct info from drivers
Simon Glass [Wed, 3 Feb 2021 13:00:53 +0000 (06:00 -0700)]
dtoc: Collect priv/plat struct info from drivers

In order to output variables to hold the priv/plat information used by
each device, dtoc needs to know the struct for each. With this, it can
declare this at build time:

   u8 xxx_priv [sizeof(struct <name>)];

Collect the various struct names from the drivers.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Ignore unwanted files when scanning for drivers
Simon Glass [Wed, 3 Feb 2021 13:00:52 +0000 (06:00 -0700)]
dtoc: Ignore unwanted files when scanning for drivers

We should ignore anything in the .git directory or any of the
build-sandbox, etc. directories created by 'make check'. These can confuse
dtoc. Update the code to ignore these.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Save scan information across test runs
Simon Glass [Wed, 3 Feb 2021 13:00:51 +0000 (06:00 -0700)]
dtoc: Save scan information across test runs

At present most of the tests scan the U-Boot source tree as part of their
run. This information does not change across tests, so we can save time
by remembering it.

Add a way to set up this information and use it for each test, taking a
copy first, so as not to mess up the original.

This reduces the run time from about 1.6 seconds to 1.5 seconds on my
machine. For code coverage (which cannot run in parallel), it reduces from
33 seconds to 5.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agodtoc: Scan drivers for available information
Simon Glass [Wed, 3 Feb 2021 13:00:50 +0000 (06:00 -0700)]
dtoc: Scan drivers for available information

At present we simply record the name of a driver parsed from its
implementation file. We also need to get the uclass and a few other
things so we can instantiate devices at build time. Add support for
collecting this information. This requires parsing each driver file.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agosandbox: add FAT to the list of usable env drivers
Heinrich Schuchardt [Thu, 4 Mar 2021 18:28:37 +0000 (18:28 +0000)]
sandbox: add FAT to the list of usable env drivers

Add the FAT environment driver to the priority list.

When testing the UEFI sub-system the EFI system partition is formatted with
FAT so it is reasonable to store the environment there.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agodm: error handling dev_get_dma_range()
Heinrich Schuchardt [Sat, 20 Feb 2021 09:41:22 +0000 (10:41 +0100)]
dm: error handling dev_get_dma_range()

goto after return has not effect. Calling of_node_put() in case of some
errors and not for others is inconsistent.

Fixes: 51bdb50904b ("dm: Introduce xxx_get_dma_range()")
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agosandbox: enable cros-ec-keyb in test.dtb
Heinrich Schuchardt [Thu, 18 Feb 2021 12:01:35 +0000 (13:01 +0100)]
sandbox: enable cros-ec-keyb in test.dtb

Currently keyboard input fails in the GUI window opened by

    ./u-boot -T -l

Add the missing include to test.dts.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agopatman: Use less for help file, if available
Nicolas Boichat [Wed, 17 Feb 2021 10:41:43 +0000 (18:41 +0800)]
patman: Use less for help file, if available

It's convenient to be able to scroll up in `patman -H`.

Signed-off-by: Nicolas Boichat <drinkcat@chromium.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agotest: print_ut: Fix potential build error
Bin Meng [Wed, 17 Feb 2021 09:31:47 +0000 (17:31 +0800)]
test: print_ut: Fix potential build error

This files uses the macro U_BOOT_CMD which is defined in command.h,
but command.h is conditionally included. Fix it.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agotest: cmd: setexpr: Fix a typo
Bin Meng [Wed, 17 Feb 2021 09:04:06 +0000 (17:04 +0800)]
test: cmd: setexpr: Fix a typo

SETEXPR_TEST is for a new setexpr test, not mem.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agommc: pci: Fix Kconfig dependency
Bin Meng [Mon, 15 Feb 2021 12:01:44 +0000 (20:01 +0800)]
mmc: pci: Fix Kconfig dependency

The PCI MMC driver depends on the generic MMC SDHCI driver,
otherwise it does not compile.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
3 years agotools: fdtgrep: Use unsigned chars for arrays
Samuel Dionne-Riel [Thu, 11 Feb 2021 00:43:09 +0000 (19:43 -0500)]
tools: fdtgrep: Use unsigned chars for arrays

Otherwise, values over 127 end up prefixed with ffffff.

Signed-off-by: Samuel Dionne-Riel <samuel@dionne-riel.com>
Cc: Simon Glass <sjg@chromium.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agosandbox: allow cross-compiling sandbox
Heinrich Schuchardt [Wed, 10 Feb 2021 17:54:25 +0000 (18:54 +0100)]
sandbox: allow cross-compiling sandbox

UEFI test files like helloworld.efi require an architecture specific
PE-COFF header.

Currently this does not work for cross compiling. If $CROSS_COMPILE is set,
use the first part of the architecture triplet from the variable to
choose the PE-COFF header.

Now we can cross-compile the sandbox, e.g.

    make sandbox_defconfig NO_SDL=1
    CROSS_COMPILE=/opt/bin/aarch64-linux-gnu- NO_SDL=1 MK_ARCH=aarch64 make

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agopatman: Quieten down the alias checking
Simon Glass [Sat, 23 Jan 2021 15:56:15 +0000 (08:56 -0700)]
patman: Quieten down the alias checking

When a tag is used in a patch subject (e.g. "tag: rest of message") and
it cannot be found as an alias, patman currently reports a fatal error,
unless -t is provided, in which case it reports a warning.

Experience suggest that the fatal error is not very useful. Instead,
default to reporting a warning, with -t tell patman to ignore it
altogether.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agopatman: Update documentation to match new usage
Simon Glass [Sat, 23 Jan 2021 15:56:14 +0000 (08:56 -0700)]
patman: Update documentation to match new usage

With the subcommands some of the documentation examples are no-longer
correct. Fix all of them, so it is consistent.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agoazure: Use --board flag with sandbox_spl
Simon Glass [Sun, 21 Mar 2021 19:22:53 +0000 (08:22 +1300)]
azure: Use --board flag with sandbox_spl

At present there is only one board which uses sandbox SPL. But with
sandbox_noinst being added, this is no longer true. Add a --board flag
so that we just build one board on azure, as is done in gitlab.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
3 years agoMerge branch '2021-03-20-mediatek-updates' into next
Tom Rini [Sat, 20 Mar 2021 23:27:22 +0000 (19:27 -0400)]
Merge branch '2021-03-20-mediatek-updates' into next

- A number of MediaTek platform updates

3 years agotools: mtk_image: add an option to set device header offset
Weijie Gao [Tue, 9 Mar 2021 07:52:31 +0000 (15:52 +0800)]
tools: mtk_image: add an option to set device header offset

This patch adds an option which allows setting the device header offset.
This is useful if this tool is used to generate ATF BL2 image of mt7622 for
SD cards.

Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
3 years agoconfigs: mt7622: enable debug uart for mt7622_rfb_defconfig
Weijie Gao [Fri, 5 Mar 2021 02:41:33 +0000 (10:41 +0800)]
configs: mt7622: enable debug uart for mt7622_rfb_defconfig

Enable debug uart for mt7622_rfb_defconfig

Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
3 years agoboard: mt7629: enable compression of u-boot to reduce the size of final image
Weijie Gao [Fri, 5 Mar 2021 02:39:55 +0000 (10:39 +0800)]
board: mt7629: enable compression of u-boot to reduce the size of final image

This patch makes use of the decompression mechanism implemented for mt7628
previously to reduce the total image size. Binman will be also removed.

Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
3 years agoserial: mtk: rewrite the setbrg function
Weijie Gao [Fri, 5 Mar 2021 02:35:39 +0000 (10:35 +0800)]
serial: mtk: rewrite the setbrg function

Currently the setbrg logic of serial-mtk is messy, and should be rewritten.
Also an option is added to make it possible to use highspeed-3 mode for all
bauds.

The new logic is:
1. If baud clock > 12MHz
   a) If baud <= 115200, highspeed-0 mode will be used (ns16550 compatible)
   b) If baud <= 576000, highspeed-2 mode will be used
   c) any bauds > 576000, highspeed-3 mode will be used
2. If baud clock <= 12MHz
   Forced highspeed-3 mode
   a) If baud <= 115200, calculates the divisor using DIV_ROUND_CLOSEST
   b) any bauds > 115200, the same as 1. c)

Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
3 years agodts: mt7629: enable JTAG pins by default
Weijie Gao [Fri, 5 Mar 2021 02:27:46 +0000 (10:27 +0800)]
dts: mt7629: enable JTAG pins by default

The EPHY LEDs belongs to the built-in FE switch of MT7629, which is barely
used. These LED pins on reference boards are used as JTAG socket. So it's
a good idea to change the default state to JTAG, and this will make it
convenience for debugging.

Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
3 years agoconfigs: mt7622: use ARMv8 Generic Timer instead of mtk_timer
Weijie Gao [Fri, 5 Mar 2021 02:24:58 +0000 (10:24 +0800)]
configs: mt7622: use ARMv8 Generic Timer instead of mtk_timer

It's better to use the generic timer which is correctly initialized by
the ATF. The generic timer has higher resolution than the mtk_timer.

Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
3 years agopinctrl: mt7629: add jtag function and pin group
Weijie Gao [Fri, 5 Mar 2021 02:22:31 +0000 (10:22 +0800)]
pinctrl: mt7629: add jtag function and pin group

The EPHY LEDs of mt7629 can be used as JTAG. This patch adds the jtag pin
group to the pinctrl driver.

Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
3 years agopinctrl: mediatek: do not probe gpio driver if not enabled
Weijie Gao [Fri, 5 Mar 2021 02:22:26 +0000 (10:22 +0800)]
pinctrl: mediatek: do not probe gpio driver if not enabled

The mtk pinctrl driver is a combination driver with support for both
pinctrl and gpio. When this driver is used in SPL, gpio support may not be
enabled, and this will result in a compilation error.

To fix this, macros are added to make sure gpio related code will only be
compiled when gpio support is enabled.

Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
3 years agopinctrl: mediatek: add get_pin_muxing ops for mediatek pinctrl
Sam Shih [Fri, 5 Mar 2021 02:22:19 +0000 (10:22 +0800)]
pinctrl: mediatek: add get_pin_muxing ops for mediatek pinctrl

This patch add get_pin_muxing support for mediatek pinctrl drivers

Signed-off-by: Sam Shih <sam.shih@mediatek.com>
3 years agopinctrl: mediatek: fix wrong assignment in mtk_get_pin_name
Sam Shih [Fri, 5 Mar 2021 02:22:11 +0000 (10:22 +0800)]
pinctrl: mediatek: fix wrong assignment in mtk_get_pin_name

This is a bug fix for mtk pinctrl common part. Appearently pins should be
used instead of grps in mtk_get_pin_name().

Signed-off-by: Sam Shih <sam.shih@mediatek.com>
3 years agoboard: Add MT8183 pumpkin board support
Fabien Parent [Mon, 15 Feb 2021 18:21:12 +0000 (19:21 +0100)]
board: Add MT8183 pumpkin board support

Add the MT8183 pumpkin board support.

Signed-off-by: Fabien Parent <fparent@baylibre.com>
3 years agoARM: mediatek: Add MT8183 support
Fabien Parent [Mon, 15 Feb 2021 18:21:11 +0000 (19:21 +0100)]
ARM: mediatek: Add MT8183 support

Add the MT8183 SoC support.

Signed-off-by: Fabien Parent <fparent@baylibre.com>
3 years agoconfigs: mt8516: use bootcmd from config_distro_bootcmd.h
Fabien Parent [Mon, 15 Feb 2021 18:07:46 +0000 (19:07 +0100)]
configs: mt8516: use bootcmd from config_distro_bootcmd.h

Instead of redefining our own way to boot, let's just use
config_distro_bootcmd.h.

Signed-off-by: Fabien Parent <fparent@baylibre.com>
3 years agoboard: mediatek: mt8516: init USB Ether for pumpkin board
Fabien Parent [Mon, 15 Feb 2021 18:07:45 +0000 (19:07 +0100)]
board: mediatek: mt8516: init USB Ether for pumpkin board

Init USB Ether if CONFIG_USB_ETHER is enabled.

Signed-off-by: Fabien Parent <fparent@baylibre.com>
3 years agoboard: mediatek: rename pumpkin board into mt8516
Fabien Parent [Mon, 15 Feb 2021 18:07:44 +0000 (19:07 +0100)]
board: mediatek: rename pumpkin board into mt8516

More than one pumpkin board has been made with different MediaTek SoCs.
Rename the pumpkin board to follow the naming convention of all
other MediaTek boards and also to not be confusing when other pumpkin
boards will be added in follow-up commits.

Signed-off-by: Fabien Parent <fparent@baylibre.com>
3 years agoMerge branch 'next' of git://source.denx.de/u-boot-sh into next
Tom Rini [Fri, 19 Mar 2021 12:20:34 +0000 (08:20 -0400)]
Merge branch 'next' of git://source.denx.de/u-boot-sh into next

- New platforms and related support

3 years agoboard: silinux: Enable recovery SPL for EK874 board
Lad Prabhakar [Mon, 15 Mar 2021 22:24:06 +0000 (22:24 +0000)]
board: silinux: Enable recovery SPL for EK874 board

Enable building SPL for EK874 board which is based on R8A774C0 SoC.

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Biju Das <biju.das.jz@bp.renesas.com>
3 years agoarm: rmobile: Add Silicon Linux EK874 board support
Lad Prabhakar [Mon, 15 Mar 2021 22:24:05 +0000 (22:24 +0000)]
arm: rmobile: Add Silicon Linux EK874 board support

The EK874 development kit from Silicon Linux is made of CAT874 (the main
board) and CAT875 (the sub board that goes on top of CAT874).

This patch adds the required board support to boot Si-Linux EK874 board
based on R8A774C0 SoC.

DTS files apart from r8a774c0-ek874-u-boot.dts and r8a774c0-u-boot.dtsi
have been imported from Linux kernel 5.11 commit f40ddce88593
("Linux 5.11").

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Biju Das <biju.das.jz@bp.renesas.com>
3 years agopinctrl: renesas: Add support for R8A774C0
Lad Prabhakar [Mon, 15 Mar 2021 22:24:04 +0000 (22:24 +0000)]
pinctrl: renesas: Add support for R8A774C0

Renesas RZ/G2E (a.k.a. r8a774c0) is pin compatible with R-Car
E3 (a.k.a. r8a77990), however it doesn't have several automotive
specific peripherals.

This patch hooks R8A774C0 SoC with the pfc driver.

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Biju Das <biju.das.jz@bp.renesas.com>
3 years agopinctrl: renesas: pfc-r8a77990: Sync PFC tables with Linux 5.11
Lad Prabhakar [Mon, 15 Mar 2021 22:24:03 +0000 (22:24 +0000)]
pinctrl: renesas: pfc-r8a77990: Sync PFC tables with Linux 5.11

Sync the R8A77990 SoC PFC tables with Linux 5.11 , commit f40ddce88593.

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Biju Das <biju.das.jz@bp.renesas.com>
3 years agoarm: dts: r8a774c0: Resync R8A774C0 SoC DTSI with Linux 5.11
Lad Prabhakar [Mon, 15 Mar 2021 22:24:02 +0000 (22:24 +0000)]
arm: dts: r8a774c0: Resync R8A774C0 SoC DTSI with Linux 5.11

Resync the R8A774C0 SoC DTSI with Linux kernel 5.11 commit f40ddce88593
("Linux 5.11").

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
3 years agoarm: rmobile: Add HopeRun HiHope RZ/G2H board support
Biju Das [Mon, 1 Mar 2021 17:08:49 +0000 (17:08 +0000)]
arm: rmobile: Add HopeRun HiHope RZ/G2H board support

The HiHope RZ/G2H board from HopeRun consists of main board
(HopeRun HiHope RZ/G2H main board) and sub board(HopeRun
HiHope RZ/G2H sub board). The HiHope RZ/G2H sub board sits
below the HiHope RZ/G2H main board.

This patch adds the required board support to boot HopeRun HiHope
RZ/G2H board.

Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
3 years agoarm: rmobile: Add HopeRun HiHope RZ/G2N board support
Biju Das [Mon, 1 Mar 2021 17:08:48 +0000 (17:08 +0000)]
arm: rmobile: Add HopeRun HiHope RZ/G2N board support

The HiHope RZ/G2N board from HopeRun consists of main board
(HopeRun HiHope RZ/G2N main board) and sub board(HopeRun
HiHope RZ/G2N sub board). The HiHope RZ/G2N sub board sits
below the HiHope RZ/G2N main board.

This patch adds the required board support to boot HopeRun HiHope
RZ/G2N board.

Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
3 years agoarm: rmobile: Add HopeRun HiHope RZ/G2M board support
Biju Das [Mon, 1 Mar 2021 17:08:47 +0000 (17:08 +0000)]
arm: rmobile: Add HopeRun HiHope RZ/G2M board support

The HiHope RZ/G2M board from HopeRun consists of main board
(HopeRun HiHope RZ/G2M main board) and sub board(HopeRun
HiHope RZ/G2M sub board). The HiHope RZ/G2M sub board sits
below the HiHope RZ/G2M main board.

This patch adds the required board support to boot HopeRun HiHope
RZ/G2M board.

Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
3 years agoarm: dts: rmobile: r8a774e1: Synchronize DTs with Linux 5.11
Biju Das [Mon, 1 Mar 2021 17:08:46 +0000 (17:08 +0000)]
arm: dts: rmobile: r8a774e1: Synchronize DTs with Linux 5.11

Synchronize r8a774e1 device trees with Linux 5.11,
commit f40ddce88593482919 ("Linux 5.11").

Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
3 years agoarm: dts: rmobile: r8a774b1: Synchronize DTs with Linux 5.11
Biju Das [Mon, 1 Mar 2021 17:08:45 +0000 (17:08 +0000)]
arm: dts: rmobile: r8a774b1: Synchronize DTs with Linux 5.11

Synchronize r8a774b1 device trees with Linux 5.11,
commit f40ddce88593482919 ("Linux 5.11")

Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
3 years agoMerge tag 'v2021.04-rc4' into next
Tom Rini [Mon, 15 Mar 2021 16:15:38 +0000 (12:15 -0400)]
Merge tag 'v2021.04-rc4' into next

Prepare v2021.04-rc4

3 years agoPrepare v2021.04-rc4
Tom Rini [Mon, 15 Mar 2021 16:06:41 +0000 (12:06 -0400)]
Prepare v2021.04-rc4

Signed-off-by: Tom Rini <trini@konsulko.com>
3 years agoconfigs: Resync with savedefconfig
Tom Rini [Mon, 15 Mar 2021 14:50:47 +0000 (10:50 -0400)]
configs: Resync with savedefconfig

Rsync all defconfig files using moveconfig.py

Signed-off-by: Tom Rini <trini@konsulko.com>
3 years agoMerge tag 'ti-v2021.04-rc4' of https://source.denx.de/u-boot/custodians/u-boot-ti
Tom Rini [Mon, 15 Mar 2021 12:43:29 +0000 (08:43 -0400)]
Merge tag 'ti-v2021.04-rc4' of https://source.denx.de/u-boot/custodians/u-boot-ti

- Fix boot for da850-evm and omap3_logic
- Optimize SPL size for am65x boards

3 years agoMerge tag 'u-boot-stm32-20210312' of https://source.denx.de/u-boot/custodians/u-boot-stm
Tom Rini [Mon, 15 Mar 2021 12:43:19 +0000 (08:43 -0400)]
Merge tag 'u-boot-stm32-20210312' of https://source.denx.de/u-boot/custodians/u-boot-stm

- Add WATCHDOG_RESET() in MTD framework and STM32 QSPI driver
- stm32mp1_trusted_defconfig rely on SCMI support
- Remove the nand MTD configuration for NOR boot in stm32mp1 board
- STM32programmer update
- Bsec: manage clock when present in device tree
- stm32mp15: move bootdelay configuration in defconfig
- Update for stm32 dsi and dw_mipi_dsi
- STM32 MCU's cleanup
- Fix compilation issue depending on SYS_DCACHE_OFF and SYS_ICACHE_OFF flags
- Update stm32mp1 doc

3 years agoMerge branch '2021-03-12-assorted-improvements' into next
Tom Rini [Mon, 15 Mar 2021 12:41:14 +0000 (08:41 -0400)]
Merge branch '2021-03-12-assorted-improvements' into next

- More log enhancements
- A few warning fixes in some cases
- Secure Channel Protocol 03 (SCP03) support for TEEs

3 years agoconfigs: am65x_evm_r5: Enable checks for spl and stack sizes
Lokesh Vutla [Tue, 9 Mar 2021 18:02:45 +0000 (23:32 +0530)]
configs: am65x_evm_r5: Enable checks for spl and stack sizes

Enable relevant configs that checks for the size of image and stack:

 BSS: 3KB
 Initial MALLOC: ~22KB
 Initial Stack: 8K
 SPL Image size can be: ~215KB

Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com>
3 years agoinclude: configs: am65x_evm: Optimize size of SPL BSS
Lokesh Vutla [Tue, 9 Mar 2021 18:02:44 +0000 (23:32 +0530)]
include: configs: am65x_evm: Optimize size of SPL BSS

Current BSS allocation of SPL is as below:
$ size spl/u-boot-spl
   text    data     bss     dec     hex filename
 132369    7852    1496  141717   22995 spl/u-boot-spl

But 20KB is allocated currently for BSS. Reduce it to 3KB and save some
space for stack.

Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com>
3 years agoARM: da850-evm: Fix boot issues from missing SPL_PAD_TO
Adam Ford [Sat, 6 Mar 2021 02:48:50 +0000 (20:48 -0600)]
ARM: da850-evm: Fix boot issues from missing SPL_PAD_TO

In a previous attempt to unify config options and remove items
from the whitelist file, SPL items were moved into a section
enabled with CONFIG_SPL_BUILD.  Unfortunately, SPL_PAD_TO
is referenced at the head Makefile and uses this define
to create padding of the output file.  When it was moved
to CONFIG_SPL_BUILD, it caused boot errors with devices
that are not booting from NOR.  Fix the boot issues by moving
SPL_PAD_TO out so it's always.

Fixes: 7bb33e4684aa ("ARM: da850-evm: Unify config options with Kconfig")
Signed-off-by: Adam Ford <aford173@gmail.com>
3 years agoconfigs: omap3_logic: Enable CONFIG_SPL_ALLOC_BD
Adam Ford [Thu, 4 Mar 2021 16:31:58 +0000 (10:31 -0600)]
configs: omap3_logic: Enable CONFIG_SPL_ALLOC_BD

With bd_info dropped from the data section, the Logic PD OMAP3 boards
and AM3517 fail to boot. Enabling CONFIG_SPL_ALLOC_BD restores
them.

Fixes: 38d6b7ebdaee ("spl: Drop bd_info in the data section")
Signed-off-by: Adam Ford <aford173@gmail.com>
3 years agotest: py: add initial coverage for scp03 cmd
Igor Opaniuk [Sun, 14 Feb 2021 15:27:28 +0000 (16:27 +0100)]
test: py: add initial coverage for scp03 cmd

Add initial test coverage for SCP03 command.

Signed-off-by: Igor Opaniuk <igor.opaniuk@foundries.io>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agosandbox: imply SCP03 and CMD_SCP03
Igor Opaniuk [Sun, 14 Feb 2021 15:27:27 +0000 (16:27 +0100)]
sandbox: imply SCP03 and CMD_SCP03

Enable by default SCP_03/CMD_SCP03 for sandbox target.

Signed-off-by: Igor Opaniuk <igor.opaniuk@foundries.io>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agodoc: describe the scp03 command
Jorge Ramirez-Ortiz [Sun, 14 Feb 2021 15:27:26 +0000 (16:27 +0100)]
doc: describe the scp03 command

The Secure Channel Protocol 03 command sends control requests
(enable/provision) to the TEE implementing the protocol between the
processor and the secure element.

Signed-off-by: Jorge Ramirez-Ortiz <jorge@foundries.io>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agodrivers: tee: sandbox: SCP03 control emulator
Jorge Ramirez-Ortiz [Sun, 14 Feb 2021 15:27:25 +0000 (16:27 +0100)]
drivers: tee: sandbox: SCP03 control emulator

Adds support for a working SCP03 emulation. Input parameters are
validated however the commands (enable, provision) executed by the TEE
are assumed to always succeed.

Signed-off-by: Jorge Ramirez-Ortiz <jorge@foundries.io>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agocmd: SCP03: enable and provision command
Jorge Ramirez-Ortiz [Sun, 14 Feb 2021 15:27:24 +0000 (16:27 +0100)]
cmd: SCP03: enable and provision command

Enable and provision the SCP03 keys on a TEE controlled secured elemt
from the U-Boot shell.

Executing this command will generate and program new SCP03 encryption
keys on the secure element NVM.

Depending on the TEE implementation, the keys would then be stored in
some persistent storage or better derived from some platform secret
(so they can't be lost).

Signed-off-by: Jorge Ramirez-Ortiz <jorge@foundries.io>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Igor Opaniuk <igor.opaniuk@foundries.io>
3 years agocommon: SCP03 control (enable and provision of keys)
Jorge Ramirez-Ortiz [Sun, 14 Feb 2021 15:27:23 +0000 (16:27 +0100)]
common: SCP03 control (enable and provision of keys)

This Trusted Application allows enabling SCP03 as well as provisioning
the keys on TEE controlled secure element (ie, NXP SE050).

All the information flowing on buses (ie I2C) between the processor
and the secure element must be encrypted. Secure elements are
pre-provisioned with a set of keys known to the user so that the
secure channel protocol (encryption) can be enforced on the first
boot. This situation is however unsafe since the keys are publically
available.

For example, in the case of the NXP SE050, these keys would be
available in the OP-TEE source tree [2] and of course in the
documentation corresponding to the part.

To address that, users are required to rotate/provision those keys
(ie, generate new keys and write them in the secure element's
persistent memory).

For information on SCP03, check the Global Platform HomePage and
google for that term [1]
[1] globalplatform.org
[2] https://github.com/OP-TEE/optee_os/
    check:
    core/drivers/crypto/se050/adaptors/utils/scp_config.c

Signed-off-by: Jorge Ramirez-Ortiz <jorge@foundries.io>
Reviewed-by: Simon Glass <sjg@chromium.org>
3 years agodoc: sandbox: Update instructions on quitting
Simon Glass [Sun, 7 Feb 2021 21:27:07 +0000 (14:27 -0700)]
doc: sandbox: Update instructions on quitting

The 'reset' command now resets sandbox but does not quit it. Fix the
instructions.

Fixes: 329dccc0675 ("sandbox: implement reset")
Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: acpi: Fix warnings on 32-bit build
Simon Glass [Sun, 7 Feb 2021 21:27:05 +0000 (14:27 -0700)]
test: acpi: Fix warnings on 32-bit build

Some format strings use the wrong type. Fix them.

Example warnings:

In file included from test/dm/acpi.c:22:
test/dm/acpi.c: In function ‘dm_test_acpi_cmd_list’:
test/dm/acpi.c:362:21: warning: format ‘%lx’ expects argument of type
  ‘long unsigned int’, but argument 4 has type ‘unsigned int’ [-Wformat=]
  ut_assert_nextline("RSDP %08lx %06lx (v02 U-BOOT)", addr,
                     ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
       sizeof(struct acpi_rsdp));
       ~~~~~~~~~~~~~~~~~~~~~~~~
include/test/ut.h:282:33: note: in definition of macro ‘ut_assert_nextline’
  if (ut_check_console_line(uts, fmt, ##args)) {   \
                                 ^~~

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotpm: Correct warning on 32-bit build
Simon Glass [Sun, 7 Feb 2021 21:27:04 +0000 (14:27 -0700)]
tpm: Correct warning on 32-bit build

Fix the warning:

drivers/tpm/tpm2_tis_sandbox.c: In function ‘sandbox_tpm2_xfer’:
drivers/tpm/tpm2_tis_sandbox.c:288:48: warning: format ‘%ld’ expects
argument of type ‘long int’, but argument 2 has type ‘size_t’
{aka ‘unsigned int’} [-Wformat=]
   printf("TPM2: Unmatching length, received: %ld, expected: %d\n",
                                              ~~^
                                              %d
          send_size, length);
          ~~~~~~~~~

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
3 years agotpm: Don't select LOG
Simon Glass [Thu, 21 Jan 2021 03:10:56 +0000 (20:10 -0700)]
tpm: Don't select LOG

We don't need to enable logging to run this command since the output will
still appear. Drop the 'select'.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agolog: Add return-checking macros for 0 being success
Simon Glass [Thu, 21 Jan 2021 03:10:54 +0000 (20:10 -0700)]
log: Add return-checking macros for 0 being success

The existing log_ret() and log_msg_ret() macros consider an error to be
less than zero. But some function may return a positive number to indicate
a different kind of failure. Add macros to check for that also.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agolog: Handle line continuation
Simon Glass [Thu, 21 Jan 2021 03:10:53 +0000 (20:10 -0700)]
log: Handle line continuation

When multiple log() calls are used which don't end in newline, the
log prefix is prepended multiple times in the same line. This makes the
output look strange.

Fix this by detecting when the previous log record did not end in newline.
In that case, setting a flag.

Drop the unused BUFFSIZE in the test while we are here.

As an example implementation, update log_console to check the flag and
produce the expected output.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agolog: Set up a flag byte for log records
Simon Glass [Thu, 21 Jan 2021 03:10:52 +0000 (20:10 -0700)]
log: Set up a flag byte for log records

At present only a single flag (force_debug) is used in log records. Before
adding more, convert this into a bitfield, so more can be added without
using more space.

To avoid expanding the log_record struct itself (which some drivers may
wish to store in memory) reduce the line-number field to 16 bits. This
provides for up to 64K lines which should be enough for anyone.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agoMerge branch '2021-03-12-test-improvements' into next
Tom Rini [Fri, 12 Mar 2021 20:57:20 +0000 (15:57 -0500)]
Merge branch '2021-03-12-test-improvements' into next

- Assorted improvements to the pytest framework

3 years agodoc: Explain briefly how to write new tests
Simon Glass [Mon, 8 Mar 2021 00:35:17 +0000 (17:35 -0700)]
doc: Explain briefly how to write new tests

Add a second on writing tests, covering when to use Python and C, where
to put the tests, etc. Add a link to the existing Python test
documentation.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: sandbox: Move sandbox test docs into doc/develop
Simon Glass [Mon, 8 Mar 2021 00:35:16 +0000 (17:35 -0700)]
test: sandbox: Move sandbox test docs into doc/develop

At present some of the documentation about running sandbox tests is in the
sandbox docs. It makes more sense to put it in with the other testing
docs, with a link there from sandbox. Update the documentation
accordingly.

Also add a paragraph explaining why sandbox exists and the test philosophy
that it uses.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agospl: test: Add a test for spl_load_simple_fit()
Simon Glass [Mon, 8 Mar 2021 00:35:15 +0000 (17:35 -0700)]
spl: test: Add a test for spl_load_simple_fit()

As an example of an SPL test, add a new test for loading a FIT within
SPL. This runs on sandbox_spl. For this to work, the text base is adjusted
so that there is plenty of space available.

While we are here, document struct spl_load_info properly, since this is
currently ambiguous.

This test only verifies the logic path. It does not actually check that
the image is loaded correctly. It is not possible for sandbox's SPL to
actually run u-boot.img since it currently includes u-boot.bin rather than
u-boot. Further work could expand the test in that direction.

The need for this was noted at:

   http://patchwork.ozlabs.org/project/uboot/patch/20201216000944.2832585-3-mr.nuke.me@gmail.com/

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agospl: Convert spl_fit to work with sandbox
Simon Glass [Mon, 8 Mar 2021 00:35:14 +0000 (17:35 -0700)]
spl: Convert spl_fit to work with sandbox

At present this casts addresses to pointers so cannot work with sandbox.
Update the code to use map_sysmem() instead.

As part of this change, the existing load_ptr is renamed to src_ptr since
it is not a pointer to load_addr. It is confusing to use a similar name
for something that is not actually related. For the alignment code,
ALIGN() is used instead of open-coded alignment. Add a comment to the line
that casts away a const.

Use a (new) load_ptr variable to access memory at address load_addr.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agosandbox: Update os_find_u_boot() to find the .img file
Simon Glass [Mon, 8 Mar 2021 00:35:13 +0000 (17:35 -0700)]
sandbox: Update os_find_u_boot() to find the .img file

At present this function can only locate the u-boot ELF file. For SPL it
is handy to be able to locate u-boot.img since this is what would normally
be loaded by SPL.

Add another argument to allow this to be selected.

While we are here, update the function to load SPL when running in TPL,
since that is the next stage.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Allow SPL to run any available test
Simon Glass [Mon, 8 Mar 2021 00:35:12 +0000 (17:35 -0700)]
test: Allow SPL to run any available test

At present SPL only runs driver model tests. Update it to run all
available tests, i.e. in any test suite.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Rename all linker lists to have a ut_ prefix
Simon Glass [Mon, 8 Mar 2021 00:35:11 +0000 (17:35 -0700)]
test: Rename all linker lists to have a ut_ prefix

At present each test suite has its own portion of the linker_list section
of the image, but other lists are interspersed. This makes it hard to
enumerate all the available tests without knowing the suites that each one
is in.

Place all tests together in a single contiguous list by giving them
common prefix not used elsewhere in U-Boot. This makes it possible to find
the start and end of all tests.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Add a macros for finding tests in linker_lists
Simon Glass [Mon, 8 Mar 2021 00:35:10 +0000 (17:35 -0700)]
test: Add a macros for finding tests in linker_lists

At present we use the linker list directly. This is not very friendly, so
add a helpful macro instead. This will also allow us to change the naming
later without updating this code.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: log: Rename log main test file to log_ut.c
Simon Glass [Mon, 8 Mar 2021 00:35:09 +0000 (17:35 -0700)]
test: log: Rename log main test file to log_ut.c

The current name is the same as the main test runner file. Rename it to
avoid confusion.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
3 years agotest: Move restoring of driver model state to ut_run_list()
Simon Glass [Mon, 8 Mar 2021 00:35:08 +0000 (17:35 -0700)]
test: Move restoring of driver model state to ut_run_list()

Add this functionality to ut_run_list() so it can be removed from
dm_test_run().

At this point all tests are run through ut_run_list().

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Move the devicetree check into ut_run_list()
Simon Glass [Mon, 8 Mar 2021 00:35:07 +0000 (17:35 -0700)]
test: Move the devicetree check into ut_run_list()

Add a check to ut_run_list() as to whether a list has driver model tests.
Move the logic for the test devicetree into that function, in an effort
to eventually remove all logic from dm_test_run().

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Use return values in dm_test_run()
Simon Glass [Mon, 8 Mar 2021 00:35:06 +0000 (17:35 -0700)]
test: Use return values in dm_test_run()

Update this function to use the return value of ut_run_list() to check for
success/failure, so that they are in sync. Also return a command success
code so that the caller gets what it expects.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Run driver-model tests using ut_run_list()
Simon Glass [Mon, 8 Mar 2021 00:35:05 +0000 (17:35 -0700)]
test: Run driver-model tests using ut_run_list()

Use this function instead of implementing it separately for driver model.

Make ut_run_tests() private since it is only used in test-main.c

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Use a local variable for test state
Simon Glass [Mon, 8 Mar 2021 00:35:04 +0000 (17:35 -0700)]
test: Use a local variable for test state

At present we use a global test state for all driver-model tests. Make use
of a local struct like we do with the other tests.

To make this work, add functions to get and set this state. When a test
starts, the state is set (so it can be used in the test). When a test
finishes, the state is unset, so it cannot be used by mistake.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Add ut_run_test_live_flat() to run tests twice
Simon Glass [Mon, 8 Mar 2021 00:35:03 +0000 (17:35 -0700)]
test: Add ut_run_test_live_flat() to run tests twice

Driver model tests are generally run twice, once with livetree enable and
again with it disabled. Add a function to handle this and call it from the
driver model test runner.

Make ut_run_test() private since it is not used outside test-main.c now.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Drop dm_do_test()
Simon Glass [Mon, 8 Mar 2021 00:35:02 +0000 (17:35 -0700)]
test: Drop dm_do_test()

In an effort to make use of a common test runner, use ut_run_test()
directly to run driver model tests.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Use ut_run_test() to run driver model tests
Simon Glass [Mon, 8 Mar 2021 00:35:01 +0000 (17:35 -0700)]
test: Use ut_run_test() to run driver model tests

Instead of having a separate function for running driver model tests, use
the common one. Make the pre/post-run functions private since we don't
need these outside of test-main.c

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Move test running into a separate function
Simon Glass [Mon, 8 Mar 2021 00:35:00 +0000 (17:35 -0700)]
test: Move test running into a separate function

Add a function to handle the preparation for running a test and the
post-test clean-up.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Move dm_test_destroy() into test-main.c
Simon Glass [Mon, 8 Mar 2021 00:34:59 +0000 (17:34 -0700)]
test: Move dm_test_destroy() into test-main.c

Move this function into the common test runner and rename it to
dm_test_post_run() so that its purpose is clear.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Move dm_test_init() into test-main.c
Simon Glass [Mon, 8 Mar 2021 00:34:58 +0000 (17:34 -0700)]
test: Move dm_test_init() into test-main.c

Move this function into test-main so that all the init is in one place.
Rename it so that its purpose is clearer.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Drop struct dm_test_state
Simon Glass [Mon, 8 Mar 2021 00:34:57 +0000 (17:34 -0700)]
test: Drop struct dm_test_state

Driver model is a core part of U-Boot. We don't really need to have a
separate test structure for the driver model tests and it makes it harder
to write a test if you have to think about which type of test it is.

Subsume the fields from struct dm_test_state into struct unit_test_state
and delete the former.

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Handle driver model reinit in test_pre_run()
Simon Glass [Mon, 8 Mar 2021 00:34:56 +0000 (17:34 -0700)]
test: Handle driver model reinit in test_pre_run()

For driver model tests we want to reinit the data structures so that
everything is in a known state before the test runs. This avoids one test
changing something that breaks a subsequent tests.

Move the call for this into test_pre_run().

Signed-off-by: Simon Glass <sjg@chromium.org>
3 years agotest: Move delay skipping to test_pre_run()
Simon Glass [Mon, 8 Mar 2021 00:34:55 +0000 (17:34 -0700)]
test: Move delay skipping to test_pre_run()

This allows delays to be skipped in sandbox tests. Move it to the
common pre-init function.

Signed-off-by: Simon Glass <sjg@chromium.org>