oweals/u-boot.git
10 years agoarm: kirkwood: Change naming of dram functions from km_foo() to mvebu_foo()
Stefan Roese [Wed, 22 Oct 2014 10:13:13 +0000 (12:13 +0200)]
arm: kirkwood: Change naming of dram functions from km_foo() to mvebu_foo()

Additionally the SDRAM address decoding register address is not hard coded
in the C code any more. A define is introduced for this base address.

This makes is possible to use those gpio functions from other MVEBU SoC's
as well.

Signed-off-by: Stefan Roese <sr@denx.de>
Tested-by: Luka Perkov <luka@openwrt.org>
Acked-by: Prafulla Wadaskar <prafulla@marvell.com>
10 years agospi: kirkwood_spi.c: Change KW_SPI_BASE to MVEBU_SPI_BASE
Stefan Roese [Wed, 22 Oct 2014 10:13:12 +0000 (12:13 +0200)]
spi: kirkwood_spi.c: Change KW_SPI_BASE to MVEBU_SPI_BASE

This makes is possible to use this SPI driver from other MVEBU SoC's as well.
As the upcoming Armada XP support will do.

Signed-off-by: Stefan Roese <sr@denx.de>
Reviewed-by: Jagannadha Sutradharudu Teki <jaganna@xilinx.com>
Tested-by: Luka Perkov <luka@openwrt.org>
Acked-by: Prafulla Wadaskar <prafulla@marvell.com>
10 years agoarm: marvell: Extract kirkwood gpio functions into new common file gpio.c
Stefan Roese [Wed, 22 Oct 2014 10:13:11 +0000 (12:13 +0200)]
arm: marvell: Extract kirkwood gpio functions into new common file gpio.c

This makes is possible to use those gpio functions from other MVEBU SoC's as well.

Signed-off-by: Stefan Roese <sr@denx.de>
Tested-by: Luka Perkov <luka@openwrt.org>
Acked-by: Prafulla Wadaskar <prafulla@marvell.com>
10 years agospi: kirkwood_spi.c: Compile MPP (pin-mux) only for kirkwood SoC's
Stefan Roese [Wed, 22 Oct 2014 10:13:10 +0000 (12:13 +0200)]
spi: kirkwood_spi.c: Compile MPP (pin-mux) only for kirkwood SoC's

Compile the pin multiplexing only on Kirkwood platforms. As the
Armada XP doesn't need it.

Signed-off-by: Stefan Roese <sr@denx.de>
Reviewed-by: Jagannadha Sutradharudu Teki <jaganna@xilinx.com>
Tested-by: Luka Perkov <luka@openwrt.org>
Acked-by: Prafulla Wadaskar <prafulla@marvell.com>
10 years agoarm: mvebu: Add common mbus functions to use on Marvell SoCs
Stefan Roese [Wed, 22 Oct 2014 10:13:09 +0000 (12:13 +0200)]
arm: mvebu: Add common mbus functions to use on Marvell SoCs

These mbus functions are ported from Barebox. The Barebox version is
ported from Linux. These functions will be first used by the upcoming
Armada XP support. Later other Marvell SoC's will be adopted to use
these functions as well (Kirkwood, Orion).

Signed-off-by: Stefan Roese <sr@denx.de>
Tested-by: Luka Perkov <luka@openwrt.org>
10 years agoarm: marvell: Rework timer.c to make it usable for other MVEBU platforms
Stefan Roese [Wed, 22 Oct 2014 10:13:08 +0000 (12:13 +0200)]
arm: marvell: Rework timer.c to make it usable for other MVEBU platforms

This patch does the following:
- Rename defines and registers to not use kirkwood
- Remove unused defines
- Use clrsetbits() accessor functions
- Coding style cleanup
- Clear 25MHZ bit in timer controller register init for Armada XP

There is no functional change for kirkwood. At least not intentionally.

This will be used by the upcoming Armada XP support.

Signed-off-by: Stefan Roese <sr@denx.de>
Tested-by: Luka Perkov <luka@openwrt.org>
Acked-by: Prafulla Wadaskar <prafulla@marvell.com>
10 years agoarm: marvell: Move arch-kirkwood/spi.h to arch-mvebu/spi.h
Stefan Roese [Wed, 22 Oct 2014 10:13:07 +0000 (12:13 +0200)]
arm: marvell: Move arch-kirkwood/spi.h to arch-mvebu/spi.h

This move makes it possible to use this kirkwood SPI driver from other
MVEBU platforms as well. This will be used by the upcoming Armada XP
support.

Signed-off-by: Stefan Roese <sr@denx.de>
Reviewed-by: Jagannadha Sutradharudu Teki <jaganna@xilinx.com>
Tested-by: Luka Perkov <luka@openwrt.org>
Acked-by: Prafulla Wadaskar <prafulla@marvell.com>
10 years agoarm: marvell: Move arch/kirkwood.h to arch/soc.h
Stefan Roese [Wed, 22 Oct 2014 10:13:06 +0000 (12:13 +0200)]
arm: marvell: Move arch/kirkwood.h to arch/soc.h

This move makes is possible to use this header not only from kirkwood
platforms but from all Marvell mvebu platforms.

Signed-off-by: Stefan Roese <sr@denx.de>
Tested-by: Luka Perkov <luka@openwrt.org>
Acked-by: Prafulla Wadaskar <prafulla@marvell.com>
10 years agoarm: kirkwood: Move some SoC files into new arch/arm/mvebu-common
Stefan Roese [Wed, 22 Oct 2014 10:13:05 +0000 (12:13 +0200)]
arm: kirkwood: Move some SoC files into new arch/arm/mvebu-common

By moving some kirkwood files into a Marvell common directory, those files
can be used by other Marvell platforms as well. The name mvebu is taken
from the Linux kernel source tree. It has been chosen there to represent
the SoC's from the Marvell EBU (Engineering Business Unit). Those SoC's
currently are:

Armada 370/375/XP, Dove, mv78xx0, Kirkwood, Orion5x

This will be used by the upcoming Armada XP (MV78460) platform support.

Signed-off-by: Stefan Roese <sr@denx.de>
Tested-by: Luka Perkov <luka@openwrt.org>
Acked-by: Prafulla Wadaskar <prafulla@marvell.com>
10 years agoMerge git://git.denx.de/u-boot-x86
Tom Rini [Thu, 23 Oct 2014 10:54:03 +0000 (06:54 -0400)]
Merge git://git.denx.de/u-boot-x86

10 years agoMerge git://git.denx.de/u-boot-fdt
Tom Rini [Thu, 23 Oct 2014 10:51:46 +0000 (06:51 -0400)]
Merge git://git.denx.de/u-boot-fdt

10 years agonet: Display the size when tftpboot finishes
Simon Glass [Fri, 10 Oct 2014 13:30:21 +0000 (07:30 -0600)]
net: Display the size when tftpboot finishes

If we know the file size, display it after loading the file.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agox86: Enable FIT, ELF on coreboot
Simon Glass [Fri, 10 Oct 2014 13:30:20 +0000 (07:30 -0600)]
x86: Enable FIT, ELF on coreboot

Enable FIT support and the bootelf command. Also change the default load
address to somewhere other than the normal load address of the kernel,
to allow for decompression without overwriting the original file.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agox86: config: link: Display the board model on the screen
Simon Glass [Fri, 10 Oct 2014 13:30:19 +0000 (07:30 -0600)]
x86: config: link: Display the board model on the screen

To get a display in U-Boot on link you must either build a coreboot that
always sets it up, or use Esc-Refresh-Power to reset the machine.

When we do have a display, it is nice to display the model at the top, so
enable this option.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agox86: link: Tidy up the command lines options
Simon Glass [Fri, 10 Oct 2014 13:30:18 +0000 (07:30 -0600)]
x86: link: Tidy up the command lines options

We may as well use hush. The auto-complete option was incorrect so this was
not enabled. Also expand the command line size a little and go back to the
default prompt since "boot>" doesn't seem any more useful.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodoc: Remove note about auto-complete not working with hush
Simon Glass [Fri, 10 Oct 2014 13:30:17 +0000 (07:30 -0600)]
doc: Remove note about auto-complete not working with hush

It does seem to work (tested on link), so update the docs.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agox86: cros_ec: Enable cros_ec for link
Simon Glass [Fri, 10 Oct 2014 13:30:16 +0000 (07:30 -0600)]
x86: cros_ec: Enable cros_ec for link

Add defines to enable the Chrome OS EC interface and set it up on init.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agox86: cros_ec: Update LPC driver for new cros_ec header
Simon Glass [Fri, 10 Oct 2014 13:30:15 +0000 (07:30 -0600)]
x86: cros_ec: Update LPC driver for new cros_ec header

There was a minor rename of one of the defines, so update the driver.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agox86: dts: Add device tree compatible string for Intel IPC
Simon Glass [Fri, 10 Oct 2014 13:30:14 +0000 (07:30 -0600)]
x86: dts: Add device tree compatible string for Intel IPC

Add this to the table so that it can be recognised.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agox86: Add device tree information for Chrome OS EC
Simon Glass [Fri, 10 Oct 2014 13:30:13 +0000 (07:30 -0600)]
x86: Add device tree information for Chrome OS EC

Add the required node describing how to find the EC on link.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agox86: config: Enable dhcp on link
Simon Glass [Fri, 10 Oct 2014 13:30:12 +0000 (07:30 -0600)]
x86: config: Enable dhcp on link

The dhcp option is required to get bootp to work on the Chromebook Pixel,
so enable it.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agofdt: Add a subnodes iterator macro
Thierry Reding [Tue, 26 Aug 2014 15:33:55 +0000 (17:33 +0200)]
fdt: Add a subnodes iterator macro

The fdt_for_each_subnode() iterator macro provided by this patch can be
used to iterate over a device tree node's subnodes. At each iteration a
loop variable will be set to the next subnode.

Acked-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Thierry Reding <treding@nvidia.com>
10 years agofdt: Add a function to return PCI BDF triplet
Thierry Reding [Tue, 26 Aug 2014 15:33:54 +0000 (17:33 +0200)]
fdt: Add a function to return PCI BDF triplet

The fdtdec_pci_get_bdf() function returns the bus, device, function
triplet of a PCI device by parsing the "reg" property according to the
PCI device tree binding.

Acked-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Thierry Reding <treding@nvidia.com>
10 years agofdt: Add resource parsing functions
Thierry Reding [Tue, 26 Aug 2014 15:33:53 +0000 (17:33 +0200)]
fdt: Add resource parsing functions

Add the fdt_get_resource() and fdt_get_named_resource() functions which
can be used to parse resources (memory regions) from an FDT. A helper to
compute the size of a region is also provided.

Signed-off-by: Thierry Reding <treding@nvidia.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agofdt: Add functions to retrieve strings
Thierry Reding [Tue, 26 Aug 2014 15:33:52 +0000 (17:33 +0200)]
fdt: Add functions to retrieve strings

Given a device tree node, a property name and an index, the new function
fdt_get_string_index() will return in an output argument a pointer to
the index'th string in the property's value.

The fdt_get_string() is a shortcut for the above with the index being 0.

Signed-off-by: Thierry Reding <treding@nvidia.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agofdt: Add a function to get the index of a string
Thierry Reding [Tue, 26 Aug 2014 15:33:51 +0000 (17:33 +0200)]
fdt: Add a function to get the index of a string

Given a device tree node and a property name, the new fdt_find_string()
function will look up a given string in the string list contained in the
property's value and return its index.

Signed-off-by: Thierry Reding <treding@nvidia.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agofdt: Add a function to count strings
Thierry Reding [Tue, 26 Aug 2014 15:33:50 +0000 (17:33 +0200)]
fdt: Add a function to count strings

Given a device tree node and a property name, the fdt_count_strings()
function counts the number of strings found in the property value.

Signed-off-by: Thierry Reding <treding@nvidia.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agoMerge git://git.denx.de/u-boot-dm
Tom Rini [Wed, 22 Oct 2014 17:51:45 +0000 (13:51 -0400)]
Merge git://git.denx.de/u-boot-dm

10 years agodm: serial: Support driver model in pl01x driver
Simon Glass [Mon, 22 Sep 2014 23:30:58 +0000 (17:30 -0600)]
dm: serial: Support driver model in pl01x driver

Add driver model support in this driver, using platform data provided by
the board.

Signed-off-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
10 years agodm: serial: Tidy up the pl01x driver
Simon Glass [Mon, 22 Sep 2014 23:30:57 +0000 (17:30 -0600)]
dm: serial: Tidy up the pl01x driver

Adjust the driver so that leaf functions take a pointer to the serial port
register base. Put all the global configuration in the init function, and
use the same settings from then on.

This makes it much easier to move to driver model without duplicating the
code, since with driver model we use platform data rather than global
settings.

The driver is compiled with either the CONFIG_PL010_SERIAL or
CONFIG_PL011_SERIAL option and this determines the uart type. With driver
model this needs to come in from platform data, so create a new
CONFIG_PL01X_SERIAL config which brings in the driver, and adjust the
driver to support both peripheral variants.

Signed-off-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
10 years agodm: rpi: Convert GPIO driver to driver model
Simon Glass [Mon, 22 Sep 2014 23:30:56 +0000 (17:30 -0600)]
dm: rpi: Convert GPIO driver to driver model

Convert the BCM2835 GPIO driver to use driver model, and switch over
Raspberry Pi to use this, since it is the only board.

Signed-off-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
Acked-by: Stephen Warren <swarren@wwwdotorg.org>
10 years agodm: core: Add support for simple-bus
Simon Glass [Sat, 4 Oct 2014 17:29:37 +0000 (11:29 -0600)]
dm: core: Add support for simple-bus

Add a driver for the simple-bus nodes, which allows devices within these
nodes to be bound.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: imx: Move cm_fx6 to use driver model for serial and GPIO
Simon Glass [Thu, 2 Oct 2014 01:57:28 +0000 (19:57 -0600)]
dm: imx: Move cm_fx6 to use driver model for serial and GPIO

Now that serial and GPIO are available for iMX.6, move cm_fx6 over as an
example.

Acked-by: Igor Grinberg <grinberg@compulab.co.il>
Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Nikita Kiryanov <nikita@compulab.co.il>
10 years agodm: imx: serial: Support driver model in the MXC serial driver
Simon Glass [Thu, 2 Oct 2014 01:57:27 +0000 (19:57 -0600)]
dm: imx: serial: Support driver model in the MXC serial driver

Add driver model support with this driver. Boards which use this driver
should define platform data in their board files.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: imx: gpio: Support driver model in MXC gpio driver
Simon Glass [Thu, 2 Oct 2014 01:57:26 +0000 (19:57 -0600)]
dm: imx: gpio: Support driver model in MXC gpio driver

Add driver model support with this driver. In this case the platform data
is in the driver. It would be better to put this into an SOC-specific file,
but this is best attempted when more boards are moved over to use driver
model.

Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Igor Grinberg <grinberg@compulab.co.il>
10 years agoarm: mx6: cm_fx6: use gpio request
Nikita Kiryanov [Thu, 2 Oct 2014 14:17:24 +0000 (17:17 +0300)]
arm: mx6: cm_fx6: use gpio request

Use gpio_request for all the gpios that are utilized by various
subsystems in cm-fx6, and refactor the relevant init functions
so that all gpios are requested during board_init(), not during
subsystem init, thus avoiding the need to manage gpio ownership
each time a subsystem is initialized.

The new division of labor is:
During board_init() muxes are setup and gpios are requested.
During subsystem init gpios are toggled.

Cc: Igor Grinberg <grinberg@compulab.co.il>
Cc: Simon Glass <sjg@chromium.org>
Signed-off-by: Nikita Kiryanov <nikita@compulab.co.il>
10 years agodm: imx: i2c: Use gpio_request() to request GPIOs
Simon Glass [Thu, 2 Oct 2014 14:17:23 +0000 (17:17 +0300)]
dm: imx: i2c: Use gpio_request() to request GPIOs

GPIOs should be requested before use. Without this, driver model will
not permit the GPIO to be used.

Cc: Igor Grinberg <grinberg@compulab.co.il>
Signed-off-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Nikita Kiryanov <nikita@compulab.co.il>
Acked-by: Igor Grinberg <grinberg@compulab.co.il>
10 years agoimx: Add error checking to setup_i2c()
Simon Glass [Thu, 2 Oct 2014 01:57:24 +0000 (19:57 -0600)]
imx: Add error checking to setup_i2c()

Since this function can fail, check its return value.

Signed-off-by: Simon Glass <sjg@chromium.org>
Tested-by: Nikita Kiryanov <nikita@compulab.co.il>
10 years agodm: serial: Put common code into separate functions
Simon Glass [Thu, 2 Oct 2014 01:57:23 +0000 (19:57 -0600)]
dm: serial: Put common code into separate functions

Avoid duplicating the code which deals with getc() and putc(). It is fairly
simple, but may expand later.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agoinitcall: Display error number when an error occurs
Simon Glass [Thu, 2 Oct 2014 01:57:22 +0000 (19:57 -0600)]
initcall: Display error number when an error occurs

Now that some initcall functions return a useful error number, display it
when something goes wrong.

Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Igor Grinberg <grinberg@compulab.co.il>
10 years agodm: core: Allow a list of devices to be declared in one step
Simon Glass [Thu, 2 Oct 2014 01:57:21 +0000 (19:57 -0600)]
dm: core: Allow a list of devices to be declared in one step

The U_BOOT_DEVICE macro allows the declaration of a single U-Boot device.
Add an equivalent macro to declare an array of devices, for convenience.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: linker_lists: Add a way to declare multiple objects
Simon Glass [Thu, 2 Oct 2014 01:57:20 +0000 (19:57 -0600)]
dm: linker_lists: Add a way to declare multiple objects

The existing ll_entry_declare() permits a single element of the list to
be added to a linker list. Sometimes we want to add several objects at
once. To avoid lots of messy declarations, add a macro to support this.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: cros_ec: Move cros_ec_spi to driver model
Simon Glass [Tue, 14 Oct 2014 05:42:16 +0000 (23:42 -0600)]
dm: exynos: cros_ec: Move cros_ec_spi to driver model

Adjust this driver to use driver model and move smdk5420 boards over to
use it.

Acked-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: sandbox: cros_ec: Move sandbox cros_ec to driver module
Simon Glass [Tue, 14 Oct 2014 05:42:15 +0000 (23:42 -0600)]
dm: sandbox: cros_ec: Move sandbox cros_ec to driver module

Adjust the sandbox cros_ec emulation driver to work with driver model, and
switch over to driver model for sandbox cros_ec.

Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: cros_ec: Add support for driver model
Simon Glass [Tue, 14 Oct 2014 05:42:14 +0000 (23:42 -0600)]
dm: cros_ec: Add support for driver model

Add support for driver model if enabled. This involves minimal changes
to the code, mostly just plumbing around the edges.

Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: tegra: spi: Convert to driver model
Simon Glass [Tue, 14 Oct 2014 05:42:13 +0000 (23:42 -0600)]
dm: tegra: spi: Convert to driver model

This converts the Tegra SPI drivers to use driver model. This is tested
on:

- Tegra20 - trimslice
- Tegra30 - beaver
- Tegra124 - dalmore

(not tested on Tegra124)

Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: tegra: dts: Add aliases for spi on tegra30 boards
Simon Glass [Tue, 14 Oct 2014 05:42:12 +0000 (23:42 -0600)]
dm: tegra: dts: Add aliases for spi on tegra30 boards

All boards with a SPI interface have a suitable spi alias except the tegra30
boards. Add these missing aliases.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: sf: Add tests for SPI flash
Simon Glass [Tue, 14 Oct 2014 05:42:11 +0000 (23:42 -0600)]
dm: sf: Add tests for SPI flash

Add a simple test for SPI that uses SPI flash. It operates by creating a
SPI flash file and using the 'sf test' command to test that all
operations work correctly.

Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: spi: Add tests
Simon Glass [Tue, 14 Oct 2014 05:42:10 +0000 (23:42 -0600)]
dm: spi: Add tests

These tests use SPI flash (and the sandbox emulation) to operate.

Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: exynos: config: Use driver model for SPI flash
Simon Glass [Tue, 14 Oct 2014 05:42:09 +0000 (23:42 -0600)]
dm: exynos: config: Use driver model for SPI flash

Use driver model for exynos5 board SPI flash.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: sf: sandbox: Convert SPI flash driver to driver model
Simon Glass [Tue, 14 Oct 2014 05:42:08 +0000 (23:42 -0600)]
dm: sf: sandbox: Convert SPI flash driver to driver model

Convert sandbox's spi flash emulation driver to use driver model.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: Convert spi_flash_probe() and 'sf probe' to use driver model
Simon Glass [Tue, 14 Oct 2014 05:42:07 +0000 (23:42 -0600)]
dm: Convert spi_flash_probe() and 'sf probe' to use driver model

We want the SPI flash probing feature to operate as a standard driver.
Add a driver for the basic probing feature used by most boards. This
will be activated by device_probe() as with any other driver.

The 'sf probe' command currently keeps track of the SPI slave that it
last used. This doesn't work with driver model, since some other driver
or system may have probed the device and have access to it too. On the
other hand, if we try to probe a device twice the second probe is a nop
with driver model.

Fix this by searching for the matching device, removing it, and then
probing it again. This should work as expected regardless of other device
activity.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: sf: Add a uclass for SPI flash
Simon Glass [Tue, 14 Oct 2014 05:42:06 +0000 (23:42 -0600)]
dm: sf: Add a uclass for SPI flash

Add a driver model uclass for SPI flash which supports the common
operations (read, write, erase). Since we must keep support for the
non-dm interface, some modification of the spi_flash header is required.

CONFIG_DM_SPI_FLASH is used to enable driver model for SPI flash.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agospi: Use error return value in sf_ops
Simon Glass [Tue, 14 Oct 2014 05:42:05 +0000 (23:42 -0600)]
spi: Use error return value in sf_ops

Adjust spi_flash_probe_slave() to return an error value instead of a
pointer so we get the correct error return.

Have the caller allocate memory for spi_flash to simplify error handling,
and also so that driver model can use its existing allocated memory.

Add a spi.h include in the sf_params file.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agosf: Tidy up public and private header files
Simon Glass [Tue, 14 Oct 2014 05:42:04 +0000 (23:42 -0600)]
sf: Tidy up public and private header files

Since spi_flash.h is supposed to be the public API for SPI flash, move
private things to sf_internal.h. Also tidy up a few comment nits.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agoexynos: universal_c210: Move to driver model soft_spi
Simon Glass [Tue, 14 Oct 2014 05:42:03 +0000 (23:42 -0600)]
exynos: universal_c210: Move to driver model soft_spi

Adjust this board to use the driver model soft_spi implementation.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: spi: Add documentation on how to convert over SPI drivers
Simon Glass [Tue, 14 Oct 2014 05:42:02 +0000 (23:42 -0600)]
dm: spi: Add documentation on how to convert over SPI drivers

This README is intended to help maintainers move their SPI drivers over to
driver model. It works through the required steps with an example.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: exynos: Convert SPI to driver model
Simon Glass [Tue, 14 Oct 2014 05:42:01 +0000 (23:42 -0600)]
dm: exynos: Convert SPI to driver model

Move the exynos SPI driver over to driver model. This removes quite a bit
of boilerplate from the driver, although it adds some for driver model.

A few device tree additions are needed to make the SPI flash available.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: spi: Add soft_spi implementation
Simon Glass [Tue, 14 Oct 2014 05:42:00 +0000 (23:42 -0600)]
dm: spi: Add soft_spi implementation

Add a new implementation of soft_spi that uses device tree to specify the
GPIOs. This will replace soft_spi_legacy for boards which use driver model.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: spi: Remove SPI_INIT feature
Simon Glass [Tue, 14 Oct 2014 05:41:59 +0000 (23:41 -0600)]
dm: spi: Remove SPI_INIT feature

This feature provides for init of a single SPI port for the soft SPI
feature. It is not really compatible with driver model since it assumes a
single SPI port. Also, inserting SPI init into the driver by means of
a #define is not very nice.

This feature is not used by any active boards, so let's remove it.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: spi: Rename soft_spi.c to soft_spi_legacy.c
Simon Glass [Tue, 14 Oct 2014 05:41:58 +0000 (23:41 -0600)]
dm: spi: Rename soft_spi.c to soft_spi_legacy.c

Reserve the 'normal' name for use by driver model, and rename the old
driver so that it is clear that it is for 'legacy' drivers only.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: sandbox: spi: Move to driver model
Simon Glass [Tue, 14 Oct 2014 05:41:57 +0000 (23:41 -0600)]
dm: sandbox: spi: Move to driver model

Adjust the sandbox SPI driver to support driver model and move sandbox over
to driver model for SPI.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: spi: Adjust cmd_spi to work with driver model
Simon Glass [Tue, 14 Oct 2014 05:41:56 +0000 (23:41 -0600)]
dm: spi: Adjust cmd_spi to work with driver model

Driver model uses a different way to find the SPI bus and slave from the
numbered devices given on the command line. Adjust the code to suit.

We use a generic SPI device, and attach it to the SPI bus before performing
the transaction.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: Add spi.h header to a few files
Simon Glass [Tue, 14 Oct 2014 05:41:55 +0000 (23:41 -0600)]
dm: Add spi.h header to a few files

Some files are using SPI functions but not explitly including the SPI
header file. Fix this, since driver model needs it.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: Remove spi_init() from board_r.c when using driver model
Simon Glass [Tue, 14 Oct 2014 05:41:54 +0000 (23:41 -0600)]
dm: Remove spi_init() from board_r.c when using driver model

Driver model does its own init, so we don't need this.

There is still a call in board_f.c but it is only enabled by CONFIG_HARD_SPI.
It is easy enough to disable that option when converting boards which use
it to driver model.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: sandbox: Add a SPI emulation uclass
Simon Glass [Tue, 14 Oct 2014 05:41:53 +0000 (23:41 -0600)]
dm: sandbox: Add a SPI emulation uclass

U-Boot includes a SPI emulation driver already but it is not explicit, and
is hidden in the SPI flash code.

Conceptually with sandbox's SPI implementation we have a layer which
creates SPI bus transitions and a layer which interprets them, currently
only for SPI flash. The latter is actually an emulation, and it should be
possible to add more than one emulation - not just SPI flash.

Add a SPI emulation uclass so that other emulations can be plugged in to
support different types of emulated devices on difference buses/chip
selects.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: spi: Add a uclass for SPI
Simon Glass [Tue, 14 Oct 2014 05:41:52 +0000 (23:41 -0600)]
dm: spi: Add a uclass for SPI

Add a uclass which provides access to SPI buses and includes operations
required by SPI.

For a time driver model will need to co-exist with the legacy SPI interface
so some parts of the header file are changed depending on which is in use.
The exports are adjusted also since some functions are not available with
driver model.

Boards must define CONFIG_DM_SPI to use driver model for SPI.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
(Discussed some follow-up comments which will address in future add-ons)

10 years agodm: core: Add a clarifying comment on struct udevice's seq member
Simon Glass [Tue, 14 Oct 2014 05:41:51 +0000 (23:41 -0600)]
dm: core: Add a clarifying comment on struct udevice's seq member

The sequence number is unique within the uclass, so state this clearly.

Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: core: Allow parents to pass data to children during probe
Simon Glass [Tue, 14 Oct 2014 05:41:50 +0000 (23:41 -0600)]
dm: core: Allow parents to pass data to children during probe

Buses sometimes want to pass data to their children when they are probed.
For example, a SPI bus may want to tell the slave device about the chip
select it is connected to.

Add a new function to permit the parent data to be supplied to the child.

Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agodm: core: Add functions for iterating through device children
Simon Glass [Tue, 14 Oct 2014 05:41:49 +0000 (23:41 -0600)]
dm: core: Add functions for iterating through device children

Buses need to iterate through their children in some situations. Add a few
functions to make this easy.

Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
10 years agosandbox: dts: Add a SPI device and cros_ec device
Simon Glass [Tue, 14 Oct 2014 05:41:48 +0000 (23:41 -0600)]
sandbox: dts: Add a SPI device and cros_ec device

Add a SPI device which can be used for testing SPI flash features in
sandbox.

Also add a cros_ec device since with driver model the Chrome OS EC
emulation will not otherwise be available.

Reviewed-by: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: Move serial to driver model
Simon Glass [Sun, 14 Sep 2014 22:36:17 +0000 (16:36 -0600)]
dm: exynos: Move serial to driver model

Change the Exynos serial driver to work with driver model and switch over
all relevant boards to use it.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: Mark exynos5 console as pre-reloc
Simon Glass [Sun, 14 Sep 2014 22:36:16 +0000 (16:36 -0600)]
dm: exynos: Mark exynos5 console as pre-reloc

We will need the console before relocation, so mark it that way.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: gpio: Convert to driver model
Simon Glass [Tue, 21 Oct 2014 01:48:40 +0000 (19:48 -0600)]
dm: exynos: gpio: Convert to driver model

Convert the exynos GPIO driver to driver model. This implements the generic
GPIO interface but not the extra Exynos-specific functions.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: Make sure that GPIOs are requested
Simon Glass [Tue, 21 Oct 2014 01:48:39 +0000 (19:48 -0600)]
dm: exynos: Make sure that GPIOs are requested

With driver model GPIOs must be requested before use. Make sure this is
done correctly.

(Note that the soft SPI part of universal is omitted, since this driver
is about to be replaced with a driver-model-aware version)

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: Tidy up GPIO defines
Simon Glass [Tue, 21 Oct 2014 01:48:38 +0000 (19:48 -0600)]
dm: exynos: Tidy up GPIO defines

The defines at the top of the GPIO driver use single-character names for
parameters which are not very descriptive.

Improve these to use descriptive parameter names.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: Tidy up GPIO headers
Simon Glass [Tue, 21 Oct 2014 01:48:37 +0000 (19:48 -0600)]
dm: exynos: Tidy up GPIO headers

The wrong header is being included, thus requiring the code to re-declare
the generic GPIO interface in each GPIO header.

Fix this.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: Move s5p_goni to generic board
Simon Glass [Tue, 21 Oct 2014 01:48:36 +0000 (19:48 -0600)]
dm: exynos: Move s5p_goni to generic board

The generic board deadline is approaching, and we need this feature to
enable driver model. Enable CONFIG_SYS_GENERIC_BOARD for s5p_goni.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: Move smdkc100 to generic board
Simon Glass [Tue, 21 Oct 2014 01:48:35 +0000 (19:48 -0600)]
dm: exynos: Move smdkc100 to generic board

The generic board deadline is approaching, and we need this feature to
enable driver model. Enable CONFIG_SYS_GENERIC_BOARD for smdkc100.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: Add pinctrl settings for s5p_goni
Simon Glass [Tue, 21 Oct 2014 01:48:34 +0000 (19:48 -0600)]
dm: exynos: Add pinctrl settings for s5p_goni

These describe the GPIOs in enough detail for U-Boot's GPIO driver to
operate.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: Add pinctrl settings for smdkc100
Simon Glass [Tue, 21 Oct 2014 01:48:33 +0000 (19:48 -0600)]
dm: exynos: Add pinctrl settings for smdkc100

These describe the GPIOs in enough detail for U-Boot's GPIO driver to
operate.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: dts: Adjust device tree files for U-Boot
Simon Glass [Tue, 21 Oct 2014 01:48:32 +0000 (19:48 -0600)]
dm: exynos: dts: Adjust device tree files for U-Boot

The pinctrl bindings used by Linux are an incomplete description of the
hardware. It is possible in most cases to determine the register address
of each, but not in all cases. By adding an additional property we can
fix this, and avoid adding a table to U-Boot for every single Exynos
SOC.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: dts: Remove unused pinctrl information to save space
Simon Glass [Tue, 21 Oct 2014 01:48:31 +0000 (19:48 -0600)]
dm: exynos: dts: Remove unused pinctrl information to save space

We don't include the pinctrl functions for U-Boot as they use up quite
a bit of space and are not used.

We could instead perhaps eliminate this material with fdtgrep, but so far
this tool has not made it to upstream.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: Bring in pinctrl dts files from Linux kernel
Simon Glass [Tue, 21 Oct 2014 01:48:30 +0000 (19:48 -0600)]
dm: exynos: Bring in pinctrl dts files from Linux kernel

Bring in required device tree files for pinctrl from Linux v3.14. These
are initially unchanged and have a number of pieces not needed by U-Boot.

Note that exynos5420 is renamed to exynos54xx here since we want to
support exynos5422 also.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodm: exynos: dts: Convert /include/ to #include
Simon Glass [Tue, 21 Oct 2014 01:48:29 +0000 (19:48 -0600)]
dm: exynos: dts: Convert /include/ to #include

We should be consistent about this. The kernel has moved to #include
which breaks error reporting to some extent but does allow us to include
binding files.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agoarm: goni: add i2c_init_board()
Robert Baldyga [Mon, 6 Oct 2014 12:33:11 +0000 (14:33 +0200)]
arm: goni: add i2c_init_board()

Add proper initialization of GPIO pins used by software i2c.

Signed-off-by: Robert Baldyga <r.baldyga@samsung.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agodm: add of_match_ptr() macro
Masahiro Yamada [Tue, 7 Oct 2014 05:51:31 +0000 (14:51 +0900)]
dm: add of_match_ptr() macro

The driver model supports two ways for passing device parameters;
Device Tree and platform_data (board file).
Each driver should generally support both of them because some
popular IPs are used on various platforms.

Assume the following scenario:
  - The driver Foo is used on SoC Bar and SoC Baz
  - The SoC Bar uses Device Tree control (CONFIG_OF_CONTROL=y)
  - The SoC Baz does not support Device Tree; uses a board file

In this situation, the device driver Foo should work with/without
the device tree control.  The driver should have .of_match and
.ofdata_to_platdata members for SoC Bar, while they are meaningless
for SoC Baz; therefore those device-tree control code should go
inside #ifdef CONFIG_OF_CONTROL.

The driver code will be like this:

  #ifdef CONFIG_OF_CONTROL
  static const struct udevice_id foo_of_match = {
          { .compatible = "foo_driver" },
          {},
  }

  static int foo_ofdata_to_platdata(struct udevice *dev)
  {
          ...
  }
  #endif

  U_BOOT_DRIVER(foo_driver) = {
          ...
          .of_match = of_match_ptr(foo_of_match),
          .ofdata_to_platdata = of_match_ptr(foo_ofdata_to_platdata),
          ...
  }

This idea has been borrowed from Linux.
(In Linux, this macro is defined in include/linux/of.h)

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agodm: fix include guard
Masahiro Yamada [Tue, 7 Oct 2014 05:49:38 +0000 (14:49 +0900)]
dm: fix include guard

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agodm: include <linker_lists.h> from platdata.h and uclass.h
Masahiro Yamada [Tue, 7 Oct 2014 05:49:13 +0000 (14:49 +0900)]
dm: include <linker_lists.h> from platdata.h and uclass.h

The header files include/dm/platdata.h and include/dm/uclass.h
use ll_entry_declare(); therefore they depend on
include/linker_lists.h.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agolinker_lists: include <linux/compiler.h>
Masahiro Yamada [Tue, 7 Oct 2014 05:48:22 +0000 (14:48 +0900)]
linker_lists: include <linux/compiler.h>

The header file include/linker_lists.h uses __aligned();
therefore it depends on include/linux/compiler.h

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agodm: simplify the loop in lists_driver_lookup_name()
Masahiro Yamada [Sun, 28 Sep 2014 13:52:27 +0000 (22:52 +0900)]
dm: simplify the loop in lists_driver_lookup_name()

if (strncmp(name, entry->name, len))
                continue;

        /* Full match */
        if (len == strlen(entry->name))
                return entry;

is equivalent to:

        if (!strcmp(name, entry->name))
                return entry;

The latter is simpler.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Simon Glass <sjg@chromium.org>
Acked-by: Igor Grinberg <grinberg@compulab.co.il>
10 years agodm: do not check the existence of uclass operation
Masahiro Yamada [Sun, 28 Sep 2014 13:52:25 +0000 (22:52 +0900)]
dm: do not check the existence of uclass operation

The function uclass_add() checks uc_drv->ops as follows:

        if (uc_drv->ops) {
                dm_warn("No ops for uclass id %d\n", id);
                return -EINVAL;
        }

It seems odd because it warns "No ops" when uc_drv->ops has
non-NULL pointer.  (Looks opposite.)

Anyway, most of UCLASS_DRIVER entries have no .ops member.
This check makes no sense.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agodm: fix comments
Masahiro Yamada [Sun, 28 Sep 2014 13:52:24 +0000 (22:52 +0900)]
dm: fix comments

The struct udevice stands for a device, not a driver.
The driver_info.name is a driver's name, which is referenced
to bind devices.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agox86: Fix GDT limit in start16.S
Bin Meng [Thu, 16 Oct 2014 14:58:35 +0000 (22:58 +0800)]
x86: Fix GDT limit in start16.S

GDT limit should be one less than an integral multiple of eight.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agox86: Fix rom version build with CONFIG_X86_RESET_VECTOR
Bin Meng [Thu, 16 Oct 2014 14:58:20 +0000 (22:58 +0800)]
x86: Fix rom version build with CONFIG_X86_RESET_VECTOR

When building U-Boot with CONFIG_X86_RESET_VECTOR, the linking
process misses the resetvec.o and start16.o so it cannot generate
the rom version of U-Boot. The arch/x86/cpu/Makefile is updated to
pull them into the final linking process.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agox86: Support loading kernel setup from a FIT
Simon Glass [Mon, 20 Oct 2014 03:11:24 +0000 (21:11 -0600)]
x86: Support loading kernel setup from a FIT

Add a new setup@ section to the FIT which can be used to provide a setup
binary for booting Linux on x86. This makes it possible to boot x86 from
a FIT.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agodoc: Tidy up and update part of the FIT documentation
Simon Glass [Mon, 20 Oct 2014 03:11:23 +0000 (21:11 -0600)]
doc: Tidy up and update part of the FIT documentation

This uses cfg instead of conf, and img instead of image. Fix these and
update in a few other places.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agosandbox: bootm: Don't fail the architecture check
Simon Glass [Mon, 20 Oct 2014 03:11:22 +0000 (21:11 -0600)]
sandbox: bootm: Don't fail the architecture check

Since sandbox is used for testing, it should be able to 'boot' an image
from any archhitecture. This allows us to test an image by loading it in
sandbox.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agox86: Allow cmdline setup in setup_zimage() to be optional
Simon Glass [Mon, 20 Oct 2014 03:11:21 +0000 (21:11 -0600)]
x86: Allow cmdline setup in setup_zimage() to be optional

If we are passing this using the device tree then we may not want to
set this up here.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agox86: Rewrite bootm.c to make it similar to ARM
Simon Glass [Mon, 20 Oct 2014 03:11:20 +0000 (21:11 -0600)]
x86: Rewrite bootm.c to make it similar to ARM

The x86 bootm code is quite special, and geared to zimage. Adjust it
to support device tree and make it more like the ARM code, with
separate bootm stages and functions for each stage.

Create a function announce_and_cleanup() to handle printing the
"Starting kernel ..." message and put it in bootm so it is in one
place and can be used by any loading code. Also move the
board_final_cleanup() function into bootm.

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