The syscon implementation in U-Boot is different from that in Linux.
Thus, DT files imported from Linux do not work for U-Boot.
In U-Boot driver model, each node is bound to a dedicated driver
that is the most compatible to it. This design gets along with the
concept of DT, and the syscon in Linux originally worked like that.
However, Linux commit bdb0066df96e ("mfd: syscon: Decouple syscon
interface from platform devices") changed the behavior because it is
useful to let a device bind to another driver, but still work as a
syscon provider.
That change had happened before U-Boot initially supported the syscon
driver by commit 6f98b7504f ("dm: Add support for generic system
controllers (syscon)"). So, the U-Boot's syscon works differently
from the beginning. I'd say this is mis-implementation given that
DT is not oriented to a particular project, but Linux is the canon
of DT in practice.
The problem typically arises in the combination of "syscon" and
"simple-mfd" compatibles.
In Linux, they are orthogonal, i.e., the order between "syscon" and
"simple-mfd" does not matter at all.
Assume the following compatible.
compatible = "foo,bar-syscon", "syscon", "simple-mfd";
In U-Boot, this device node is bound to the syscon driver
(driver/core/syscon-uclass.c) since the "syscon" is found to be the
most compatible. Then, syscon_get_regmap() succeeds.
However,
compatible = "foo,bar-syscon", "simple-mfd", "syscon";
does not work because this node is bound to the simple-bus driver
(drivers/core/simple-bus.c) in favor of "simple-mfd" compatible.
The compatible string "syscon" is just dismissed.
Moreover,
compatible = "foo,bar-syscon", "syscon";
works like the first case because the syscon driver populates the
child devices. This is wrong because populating children is the job
of "simple-mfd" (or "simple-bus").
This commit ports syscon_node_to_regmap() from Linux. This API
does not require the given node to be bound to a driver in any way.
Reported-by: Kunihiko Hayashi <hayashi.kunihiko@socionext.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Currently, regmap_init_mem() takes a udevice. This requires the node
has already been associated with a device. It prevents syscon/regmap
from behaving like those in Linux.
Change the first argumenet to take a device node.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
device_is_compatible() takes udevice, but there is no such a helper
that takes ofnode.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Putting zero length array at the end of struct is a common technique
to embed arbitrary length of members. There is no good reason to let
regmap_alloc_count() branch by "if (count <= 1)".
As far as I understood the code, regmap->base is an alias of
regmap->ranges[0].start, but it is not helpful but make the code
just ugly.
Rename regmap_alloc_count() to regmap_alloc() because the _count
suffix seems pointless.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
[trini: fixup cpu_info-rcar.c]
Signed-off-by: Tom Rini <trini@konsulko.com>
When U-Boot started using SPDX tags we were among the early adopters and
there weren't a lot of other examples to borrow from. So we picked the
area of the file that usually had a full license text and replaced it
with an appropriate SPDX-License-Identifier: entry. Since then, the
Linux Kernel has adopted SPDX tags and they place it as the very first
line in a file (except where shebangs are used, then it's second line)
and with slightly different comment styles than us.
In part due to community overlap, in part due to better tag visibility
and in part for other minor reasons, switch over to that style.
This commit changes all instances where we have a single declared
license in the tag as both the before and after are identical in tag
contents. There's also a few places where I found we did not have a tag
and have introduced one.
Signed-off-by: Tom Rini <trini@konsulko.com>
Commit 286ede6 ("drivers: core: Add translation in live tree case") made
dev_get_addr always use proper bus translations for addresses read from
the device tree. But this leads to problems with certain busses, e.g.
I2C busses, which run into an error during translation, and hence stop
working.
It turns out that of_translate_address() and fdt_translate_address()
stop the address translation with an error when they're asked to
translate addresses for busses where #size-cells == 0 (comment from
drivers/core/of_addr.c):
* Note: We consider that crossing any level with #size-cells == 0 to mean
* that translation is impossible (that is we are not dealing with a value
* that can be mapped to a cpu physical address). This is not really specified
* that way, but this is traditionally the way IBM at least do things
To fix this case, we check in both the live-tree and non-live tree-case,
whether the bus of the device whose address is about to be translated
has size-cell size zero. If this is the case, we just read the address
as a plain integer and return it, and only apply bus translations if the
size-cell size if greater than zero.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Signed-off-by: Martin Fuzzey <mfuzzey@parkeon.com>
Reported-by: Martin Fuzzey <mfuzzey@parkeon.com>
Fixes: 286ede6 ("drivers: core: Add translation in live tree case")
Reviewed-by: Simon Glass <sjg@chromium.org>
dm_scan_fdt_node can't work when live dt is active,
we should use dm_scan_fdt_live instead.
Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Add api for who can not get phandle from a device property.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Thomas reported U-Boot failed to build host tools if libfdt-devel
package is installed because tools include libfdt headers from
/usr/include/ instead of using internal ones.
This commit moves the header code:
include/libfdt.h -> include/linux/libfdt.h
include/libfdt_env.h -> include/linux/libfdt_env.h
and replaces include directives:
#include <libfdt.h> -> #include <linux/libfdt.h>
#include <libfdt_env.h> -> #include <linux/libfdt_env.h>
Reported-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
We need to get ofnode from a phandle, add interface to support
both live dt and fdt.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
The Rockchip video drivers need to walk the ofnode-parrents to find
an enclosing device that has a UCLASS_DISPLAY driver bound. This
adds a ofnode_get_parent()-function that returns the parent-node.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
Reviewed-by: Anatolij Gustschin <agust@denx.de>
Each uclass has a driver name which we can use to look up the uclass. This
is useful for logging, where the uclass ID is used as the category.
Add a function to handle this, as well as a test.
Signed-off-by: Simon Glass <sjg@chromium.org>
Linux uses the properties 'assigned-clocks', 'assigned-clock-parents'
and 'assigned-clock-rates' to configure the clock subsystem for use
with various peripheral nodes.
This implements clk_set_defaults() and hooks it up with the general
device probibin in drivers/core/device.c: when a new device is probed,
clk_set_defaults() will be called for it and will process the
properties mentioned above.
Note that this functionality is designed to fail gracefully (i.e. if a
clock-driver does not implement set_parent(), we simply accept this
and ignore the error) as not to break existing board-support.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Tested-by: David Wu <david.wu@rock-chips.com>
Series-changes: 2
- Fixed David's email address.
Series-version: 2
Cover-letter:
clk: support assigned-clock, assigned-clock-parents, assigned-clock-rates
For various peripherals on Rockchip SoCs (e.g. for the Ethernet GMAC),
the parent-clock needs to be set via the DTS. This adds the required
plumbing and implements the GMAC case for the RK3399.
END
dev_read_u32_default() always returns something even when the property
is missing. So, it is impossible to do nothing in the case. One
solution is to use ofnode_read_u32() instead, but adding dev_read_u32()
will be helpful.
BTW, Linux has an equvalent function, device_property_read_u32();
it is clearer that it reads a property. I cannot understand the
behavior of dev_read_u32() from its name.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Judging from its name and parameters, device_is_compatible looks like it
is compatible with a live device tree, but it actually isn't.
Make it compatible with a live device tree.
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Mario Six <mario.six@gdsys.cc>
There are some whitespace-related style violations in read.c; fix those.
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Mario Six <mario.six@gdsys.cc>
The function dev_read_addr calls ofnode_get_addr_index in the live tree
case, which does not apply bus translations to the address read from the
device tree. This results in illegal addresses on boards that rely on
bus translations being applied.
Fix this situation by applying bus translations in the live tree case as
well.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Tested-by: Stephen Warren <swarren@nvidia.com>
This currently causes a warning in sandbox and will not do the right
thing:
drivers/core/read.c: In function ‘dev_read_addr_ptr’:
drivers/core/read.c:64:44: warning: cast to pointer from integer of
different size [-Wint-to-pointer-cast]
return (addr == FDT_ADDR_T_NONE) ? NULL : (void *)addr;
Use map_sysmem() which is the correct way to convert an address to a
pointer.
Fixes: c131c8bca8 (dm: core: add dev_read_addr_ptr())
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
The livetree codepath of ofnode_get_addr_size always used the "reg"
property for of_get_property. Use the property parameter of the function
call instead and check the return value if the property exists.
Otherwise return FDT_ADDR_T_NONE.
This was discoverd while using SPI NOR with livetree.
spi_flash_decode_fdt checks for memory-map and will not fail with
livetree even if the property does not exist.
Signed-off-by: Klaus Goger <klaus.goger@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The size variable may not be always be a mulitple of
ARCH_DMA_MINALIGN and using it to flush cache leads to cache
misaligned warnings.
Therefore, round up the size to a multiple of ARCH_DMA_MINLAIGN
when allocating private data.
Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
As we discussed before in ML, dm_dbg() causes undefined reference
error if #define DEBUG is added to users, but not drivers/core/util.c
We do not need this macro because we can use pr_debug() instead, and
it is pretty easy to enable it for the DM core by using ccflags-y.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
dev_read_string_count() is used to get the number of strings in a
stringlist.
dev_read_string_index() is used to get a string in the stringlist based on
its position in the list.
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
The dev_read_addr_ptr() mimics the behaviour of the devfdt_get_addr_ptr(),
retrieving the first address of the node's reg-property and returning
it as a pointer (or NULL on failure).
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
When using 32-bit addresses dtoc works correctly. For 64-bit addresses it
does not since it ignores the #address-cells and #size-cells properties.
Update the tool to use fdt64_t as the element type for reg properties when
either the address or size is larger than one cell. Use the correct value
so that C code can obtain the information from the device tree easily.
Alos create a new type, fdt_val_t, which is defined to either fdt32_t or
fdt64_t depending on the word size of the machine. This type corresponds
to fdt_addr_t and fdt_size_t. Unfortunately we cannot just use those types
since they are defined to phys_addr_t and phys_size_t which use
'unsigned long' in the 32-bit case, rather than 'unsigned int'.
Add tests for the four combinations of address and size values (32/32,
64/64, 32/64, 64/32). Also update existing uses for rk3399 and rk3368
which now need to use the new fdt_val_t type.
Signed-off-by: Simon Glass <sjg@chromium.org>
Suggested-by: Heiko Stuebner <heiko@sntech.de>
Reported-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Tested-by: Kever Yang <kever.yang@rock-chips.com>
Currently, all fixed-clock declared in "clocks" node in device tree
can be binded by clk_fixed_rate.c driver only if each of them have
the "simple-bus" compatible string.
This constraint has been invoked here [1].
This patch offers a solution to avoid adding "simple-bus" compatible
string to nodes that are not busses.
[1] https://patchwork.ozlabs.org/patch/558837/
Signed-off-by: Patrice Chotard <patrice.chotard@st.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Add a convenience macro to iterate over subnodes of a node. Make use of
this where appropriate in the code.
Signed-off-by: Simon Glass <sjg@chromium.org>
We can use printf() to limit the string width. Adjust the code to do this
instead of using strlcpy() which is a bit clumbsy.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Linux supports platform_get_resource_byname() to look up a resource
by name.
We want a similar helper. It is useful when a device node has named
register regions.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
We cannot call dm_scan_fdt_dev() with of-platdata since there is no device
tree. Fix this with an #if check.
Fixes: 3be9a37 (dm: syscon: scan sub-nodes of the syscon node)
Signed-off-by: Simon Glass <sjg@chromium.org>
This change introduces TPL variants of the REGMAP and SYSCON config
options (i.e. TPL_REGMAP and TPL_SYSCON in analogy to SPL_REGMAP and
SPL_SYSCON) in preparation of a finer-grained feature selection for
building feature-rich TPL variants.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
SPL_REGMAP and SPL_SYSCON were marked as depending on DM, when a
stricter dependency of SPL_DM was possible. This commit makes the
prereq more specific.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
This function is usefull to get phandle number contained
in a property list.
For example, this allows to allocate the right amount
of memory to keep clock's reference contained into the
"clocks" property.
To implement it, either of_count_phandle_with_args() or
fdtdec_parse_phandle_with_args() are used respectively
for live tree and flat tree.
By passing index = -1, these 2 functions returns the
number of phandle contained into the property list.
Add also the dev_count_phandle_with_args() based on
ofnode_count_phandle_with_args()
Signed-off-by: Patrice Chotard <patrice.chotard@st.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
At present this function only supports 32-bit (single-cell) values. Update
it to support two-cell values also.
Signed-off-by: Simon Glass <sjg@chromium.org>
Tested-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Tested-on: Beaver, Jetson-TK1
Tested-by: Stephen Warren <swarren@nvidia.com>
We sometimes need to read a resource from an arbitrary node. In any case
for consistency we should not put the live-tree switching code in
a dev_read_...() function. Update this to suit.
Signed-off-by: Simon Glass <sjg@chromium.org>
Tested-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Tested-on: Beaver, Jetson-TK1
Tested-by: Stephen Warren <swarren@nvidia.com>
DT property values can be strings as well as integers. This is why
of_get_property/fdt_getprop returns an opaque pointer.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Simon Glass <sjg@chromium.org>
Some platforms have very limited SRAM to run SPL code, so there may
not be the same amount space for a malloc pool before relocation in
the SPL stage as the normal U-Boot stage.
Make SPL and (the full) U-Boot stage use independent SYS_MALLOC_F_LEN,
so the size of pre-relocation malloc pool can be configured memory
space independently.
Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
[fixed up commit-message:]
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
DTB is encoded in big endian. When we retrieve property values,
we need to use fdt32_to_cpu (aka be32_to_cpu) for endian conversion.
This is a bit error-prone, but sparse is useful to detect endian
mismatch.
We need to use (fdt32_t *) instead of (u32 *) for a pointer of a
property value. Otherwise sparse warns "cast to restricted __be32".
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Include <dm/util.h> to fix sparse warnings:
symbol 'dm_dump_all' was not declared. Should it be static?
symbol 'dm_dump_uclass' was not declared. Should it be static?
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
Reuse ofnode_get_property() to simplify the implementation.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Simon Glass <sjg@chromium.org>
This function returns the pointer to the value of a node property.
The current name ofnode_read_prop() is confusing. Follow the naming
of_get_property() from Linux.
The return type (const u32 *) is wrong. DT property values can be
strings as well as integers. This is why of_get_property/fdt_getprop
returns an opaque pointer.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Simon Glass <sjg@chromium.org>
The code inside the if-block is the same as of_get_property().
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Simon Glass <sjg@chromium.org>
