i2c-host updates for v6.15

Refactoring and cleanups
 - octeon, cadence, i801, pasemi, mlxbf, bcm-iproc: general
   refactorings
 - octeon: remove 10-bit address support

Improvements
 - amd-asf: improved error handling
 - designware: use guard(mutex)
 - amd-asf, designware: update naming to follow latest specs
 - cadence: fix cleanup path in probe
 - i801: use MMIO and I/O mapping helpers to access registers
 - pxa: handle error after clk_prepare_enable

New features
 - added i2c_10bit_addr_*_from_msg() and updated multiple drivers
 - omap: added multiplexer state handling
 - qcom-geni: update frequency configuration
 - qup: introduce DMA usage policy

New hardware support
 - exynos: add support for Samsung exynos7870
 - k1: add support for spacemit k1 (new driver)
 - imx: add support for i.mx94 lpi2c
 - rk3x: add support for rk3562

Multiplexers
 - ltc4306, reg: fix assignment in platform_driver structure

There are already a lot of drivers that have been using
i2c_8bit_addr_from_msg() for 7-bit addresses, now it's time
to have the similar for 10-bit addresses.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20250213141045.2716943-2-andriy.shevchenko@linux.intel.com
Signed-off-by: Andi Shyti <andi.shyti@kernel.org>

i2c_of_match_device() is not used anymore outside of I²C framework,
unexport it.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Vladimir Zapolskiy <vladimir.zapolskiy@linaro.org>
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

This reverts commit 56a50667cbcfaf95eea9128d5676af94e54b51a8. Mux
handling is not sufficiently implemented. It needs more time.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

This reverts commit 3cfe39b3a845593a485ab1c716615979004ef9f6. Mux
handling is not sufficiently implemented. It needs more time.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

When adding the new debugfs entry, its kdoc equivalent was forgotten.
Add it now.

Fixes: d06905d68610 ("i2c: add core-managed per-client directory in debugfs")
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Closes: https://lore.kernel.org/r/20250115163146.6c48f066@canb.auug.org.au
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

More and more I2C client drivers use debugfs entries and currently they
need to manage a subdir for their files on their own. This means
inconsistent naming for these subdirs and they are scattered all over
the debugfs-tree as well. Not to mention the duplicated code.

Let the I2C core provide and maintain a proper directory per client.

Note: It was considered to save the additional pointer in 'struct
i2c_client' and only provide a subdir when requested via a helper
function. When sketching this approach, more and more corner cases
appeared, though, so the current solution with its simple and unabiguous
code was chosen.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>

Similar to the list of auto-detected clients, we can also replace the
list of userspace-created clients with flagging such client devices.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
[wsa: fixed description of the new flag; reordered new code in
'device_store' to have single exit point; fixed whitespace errors;
 folded cleanup patch into this one]
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

So far a list is used to track auto-detected clients per driver.
The same functionality can be achieved much simpler by flagging
auto-detected clients.

Two notes regarding the usage of driver_for_each_device:
In our case it can't fail, however the function is annotated __must_check.
So a little workaround is needed to avoid a compiler warning.
Then we may remove nodes from the list over which we iterate.
This is safe, see the explanation at the beginning of lib/klist.c.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
[wsa: fixed description of the new flag]
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

The DesignWare and the Renesas I2C drivers have received most of
the changes in this pull request.

The first has has undergone through a series of cleanups that
have been sent to the mailing list a year ago for the first time
and finally get merged in this pull request. They are many, from
typos (e.g. i2/i2c), to cosmetics, to refactoring (e.g. move
inline functions to librarieas) and many others.

Besides that, all the DesignWare Kconfig options have been
grouped under the I2C_DESIGNWARE_CORE and this required some
adaptation in many of the kernel configuration files for
different arm and mips boards.

Follows the list of the rest of the changes grouped by type of
change.

Cleanups
--------
The Qualcomm Geni platform improves the exit path in the runtime
resume function.

The Intel LJCA driver loses "target_addr" parameter in
ljca_i2c_stop() because it was unused.

The MediaTek controller intializes the restart_flag in the
transfer function using the ternary conditional operator ("? :")
instead of initializing it in different parts.

Constified a few global data structures in the virtio driver.

The Renesas driver simplifies the bus speed handling in the init
function making it more readable.

Improved an if/else statement in probe function of the Renesas
R-Car driver.

The iMX/MXC driver switches to using the RUNTIME_PM_OPS() instead
of SET_RUNTIME_PM_OPS().

Still in the iMX/MXC driver a comma ',' has been replaced by a
semicolon ';', while in different drivers the ',' has been
removed from the '{ }' delimiters.

Finally three devm_clk_get_enabled() have been used to simplify
the devm_clk_get/clk_prepare_enable tuple in the Renesas EMEV2,
Ingenic and MPC drivers.

Refactors
---------
The Nuvoton fixes a potential out of boundary array access. This
is not a bug fix because the issue could never occur due to
hardware not having the properties listed in the array. The
change makes the driver more future proof and, at the same time,
silences code analyzers.

Improvements
------------
The Renesas I2C (riic) driver undergoes several patches improving
the runtime power management handling.

The Intel i801 driver uses a more descriptive adapter's name to
show the presence of the IDF feature.

In the Intel Denverton (ismt) adapter the pending transactions
are killed when irq's can't complete their handling, triggering a
timeout. This could have been considered as a bug fix, but
because, standing to Vasily, it's very sporadic, I preferred
considering the patch rather as an improvement.

New Feature
-----------
The Renesas I2C (riic) driver now supports the fast mode plus.

New support
-----------
Added support for:

    - Renesas R9A08G045
    - Rockchip RK3576
    - KEBA I2C
    - Theobroma Systems Mule Multiplexer.

The Keba comes with a new driver, i2c-keba.c.
The Mule is an i2c multiplexer and it also comes with a new
driver, mux/i2c-mux-mule.c.

Core patch
----------
This pull request includes also a patch in the I2C framework, in
i2c-core-base.c where the runtime PM functions have been replaced
in order to allow to be accessed during the device add.

Devicetree
----------
Some cleanups in the devicetree, as well. nVidia and Qualcomm
bindings improve their "if:then:" blocks. While the aspeed
binding loses the "multi-master" property because it was
redundant.

The i2c-sprd binding has been converted to YAML.