this is an atempt to make the export of functions typesafe.
I replaced the jumptable void ** by a struct (jt_funcs) with function pointers.
The EXPORT_FUNC macro now has 3 fixed parameters and one
variadic parameter
The first is the name of the exported function,
the rest of the parameters are used to format a functionpointer
in the jumptable,
the EXPORT_FUNC macros are expanded three times,
1. to declare the members of the struct
2. to initialize the structmember pointers
3. to call the functions in stubs.c
Signed-off-by: Martin Dorwig <dorwig@tetronik.com>
Acked-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Simon Glass <sjg@chromium.org>
(resending to the list since my tweaks are not quite trivial)
I2C is now deprecated on ARM platforms and there are no devices that use it
with the v3 protocol. We can't require v3 support if we want to support I2C.
Adjust the error handling to suit.
Signed-off-by: Simon Glass <sjg@chromium.org>
We don't want to bind devices which should never be used.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
This has moved to driver model so we don't need the fdtdec support.
Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Minkyu Kang <mk7.kang@samsung.com>
At present we go through various contortions to store the I2C's chip
address in its private data. This only exists when the chip is active so
must be set up when it is probed. Until the device is probed we don't
actually record what address it will appear on.
However, now that we can support per-child platform data, we can use that
instead. This allows us to set up the address when the child is bound,
and avoid the messy contortions.
Unfortunately this is a fairly large change and it seems to be difficult to
break it down further.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
At present we go through various contortions to store the SPI slave's chip
select in its private data. This only exists when the slave is active so
must be set up when it is probed. Until the device is probed we don't
actually know what chip select it will appear on.
However, now that we can support per-child platform data, we can use that
instead. This allows us to set up the chip select when the child is bound,
and avoid the messy contortions.
Unfortunately this is a fairly large change and it seems to be difficult to
break it down further.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present we use struct spi_slave as our device pointer in a lot of places
to avoid changing the old SPI API. At some point this will go away.
But for now, it is better if the SPI uclass sets up this pointer, rather
than relying on passing it into the device when it is probed. We can use the
new uclass child_pre_probe() method to do this.
Signed-off-by: Simon Glass <sjg@chromium.org>
Some buses need to set up their devices before they can be used. This setup
may well be common to all buses in a particular uclass. Support a common
pre-probe method for the uclass, called before any bus devices are probed.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
For buses, after a child is bound, allow the uclass to perform some
processing. This can be used to figure out the address of the child (e.g.
the chip select for SPI slaves) so that it is ready to be probed.
This avoids bus drivers having to repeat the same process, which really
should be done by the uclass, since it is common.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
This is common to all SPI drivers and specifies a structure used by the
uclass. It makes more sense to define it in the uclass.
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
In many cases the per-child private data for a device's children is defined
by the uclass rather than the individual driver. For example, a SPI bus
needs to store information about each of its children, but all SPI drivers
store the same information. It makes sense to allow the uclass to define
this data.
If the driver provides a size value for its per-child private data, then use
it. Failng that, fall back to that provided by the uclass.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
At present we try to use the 'reg' property and device tree aliases to give
devices a sequence number. The 'reg' property is often actually a memory
address, so the sequence numbers thus-obtained are not useful. It would be
better if the devices were just sequentially numbered in that case. In fact
neither I2C nor SPI use this feature, so drop it.
Some devices need us to look up an alias to number them within the uclass.
Add a flag to control this, so it is not done unless it is needed.
Adjust the tests to test this new behaviour.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
This is useful to check which uclass a device is in.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Allow parent drivers to be called when a new child is bound to them. This
allows a bus to set up information it needs for that child.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
In many cases the child platform data for a device's children is defined by
the uclass rather than the individual devices. For example, a SPI bus needs
to know the chip select and speed for each of its children. It makes sense
to allow this information to be defined the SPI uclass rather than each
individual driver.
If the device provides a size value for its child platdata, then use it.
Failng that, fall back to that provided by the uclass.
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
For buses it is common for parents to need to know the address of the child
on the bus, the bus speed to use for that child, and other information. This
can be provided in platform data attached to each child.
Add driver model support for this, including auto-allocation which can be
requested using a new property to specify the size of the data.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
When using allocated platform data, allocate it when we bind the device.
This makes it possible to fill in this information before the device is
probed.
This fits with the platform data model (when not using device tree),
since platform data exists at bind-time.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Make the error handling more standard to make it easier to build on top of
it. Also correct a bug in the error path where there is no parent.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
The root device corresponds to the root device tree node, so set this up.
Also add a few notes to the documentation.
Signed-off-by: Simon Glass <sjg@chromium.org>
Rather than assuming that the chip offset length is 1, allow it to be
provided. This allows chips that don't use the default offset length to
be used (at present they are only supported by the command line 'i2c'
command which sets the offset length explicitly).
Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Heiko Schocher <hs@denx.de>
For boards which use multiple I2C devices, or for SOCs which support
multiple boards, we might want to convert these to driver model at different
times. At present this is difficult because we need to either use
CONFIG_DM_I2C for a board or not.
Add a compatibility layer which implements the old API, thus allowing a
board to move to driver model for I2C without requiring that everything it
uses is moved in the same commit.
Signed-off-by: Simon Glass <sjg@chromium.org>
Add a dm_ prefix to driver model I2C functions so that we can keep the old
ones around.
This is a little unfortunate, but on reflection it is too difficult to
change the API. We can undo this rename when most boards and drivers are
converted to use driver model for I2C.
Signed-off-by: Simon Glass <sjg@chromium.org>
Now that we support device tree GPIO bindings directly in the driver model
GPIO uclass we can remove these functions.
Signed-off-by: Simon Glass <sjg@chromium.org>
U-Boot now supports using GPIOs using bank phandles instead of global
numbers. Update the exynos device tree files to use this.
Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Minkyu Kang <mk7.kang@samsung.com>
This deals with the polarity bit. It also changes the GPIO devices so that
the correct device tree node is linked to each one. This allows us to use
the new uclass phandle functionality to implement a proper GPIO binding.
Signed-off-by: Simon Glass <sjg@chromium.org>
Add a new 'demo light' command which uses GPIOs to control imaginary lights.
Each light is assigned a bit number in the overall value. This provides an
example driver for using the new GPIO API.
Signed-off-by: Simon Glass <sjg@chromium.org>
Add a deprecation notice to each function so that it is more obvious that we
are moving GPIOs to driver model.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present U-Boot sort-of supports the standard way of reading GPIOs from
device tree nodes, but the support is incomplete, a bit clunky and only
works for GPIO bindings where #gpio-cells is 2.
Add new functions to request GPIOs, taking full account of the device
tree binding. These permit requesting a GPIO with a simple call like:
gpio_request_by_name(dev, "cd-gpios", 0, &desc, GPIOD_IS_IN);
This will request the GPIO, looking at the device's node which might be
this, for example:
cd-gpios = <&gpio TEGRA_GPIO(B, 3) GPIO_ACTIVE_LOW>;
The GPIO will be set to input mode in this case and polarity will be
honoured by the GPIO calls.
It is also possible to request and free a list of GPIOs.
Signed-off-by: Simon Glass <sjg@chromium.org>
Only the GPIO driver knows about the full GPIO device tree binding used by
a device. Add a method to allow the driver to provide this information to the
uclass, including the GPIO offset within the device and flags such as the
polarity.
Signed-off-by: Simon Glass <sjg@chromium.org>
So far driver model's GPIO uclass just implements the existing GPIO API.
This has some limitations:
- it requires manual device tree munging to support GPIOs in device tree
(fdtdec_get_gpio() and friends)
- it does not understand polarity
- it is somewhat slower since we must scan for the GPIO device each time
- Global GPIO numbering can change if other GPIO drivers are probed
- it requires extra steps to set the GPIO direction and value
The new functions have a dm_ prefix where necessary to avoid name conflicts
but we can remove that when it is no-longer needed. The new struct gpio_desc
holds all required information about the GPIO. For now this is intended to
be stored by the client requesting the GPIO, but in future it might be
brought into the uclass in some way.
With these changes the old GPIO API still works, and uses the driver model
API underneath.
Signed-off-by: Simon Glass <sjg@chromium.org>
For GPIOs and other functions we want to look up a phandle and then decode
a list of arguments for that phandle. Each phandle can have a different
number of arguments, specified by a property in the target node. This is
the "#gpio-cells" property for GPIOs.
Add a function to provide this feature, taken modified from Linux 3.18.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present the tegra GPIO driver does not fully support the existing device
tree binding, but add the binding file to cover the existing partial support.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present the exynos GPIO driver does not fully support the existing device
tree binding, but add the binding file to cover the existing partial support.
Signed-off-by: Simon Glass <sjg@chromium.org>
Martin Dorwig