Jan reported this:

===
After enabling CONFIG_I2C_DEBUG_CORE my system was broken
(no network, console login not possible). System log was
flooded with the this message:

 ...
[  608.052077] rtc-ds1307 0-0068: uevent
[  608.052500] rtc-ds1307 0-0068: uevent
[  608.052925] rtc-ds1307 0-0068: uevent
 ...

The culprit is the dev_dbg printk in the i2c uevent handler.
If this is activated (for instance by CONFIG_I2C_DEBUG_CORE)
it results in an endless loop with systemd-journald.

This happens if user-space scans the system log and reads the uevent
file to get information about a newly created device, which seems fair
use to me. Unfortunately reading the "uevent" file uses the same
function that runs for creating the uevent for a new device,
generating the next syslog entry.

Ideally user-space would implement a recursion detection and
after reading the same device file for the 1000th time call it a
day, but nevertheless I think we should avoid this problem by
removing the debug print completely or using another print variant.

The same problem seems to be reported here:
https://bugs.freedesktop.org/show_bug.cgi?id=76886
===

His patch converted the message to pr_debug, but I think the debug can
simply go. We have other means to see code paths these days. This
enables us to clean up the function some more while we are here.
Reported-by: NJan Glauber <jglauber@cavium.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>
Acked-by: NAlexander Sverdlin <alexander.sverdlin@nokia.com>
Tested-by: NJan Glauber <jglauber@cavium.com>

Mark the i2c bus as registered right after the the bus_register call,
not at the end of init. Otherwise, we can't register our own dummy
driver.
Reported-by: NThierry Reding <thierry.reding@gmail.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>
Fixes: 95026658 ("i2c: do not use internal data from driver core")

The variable p is a data structure which is used by the driver core
internally and it is not expected that busses will be directly accessing
these driver core internal only data.
Signed-off-by: NSudip Mukherjee <sudip.mukherjee@codethink.co.uk>
Acked-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>
[wsa: removed the unlikely()]
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

I request this for drivers, so the core should adhere to sorted includes as
well.
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

The adapter device is a logical device. Because of that, it already uses
pm_runtime_no_callbacks() in the core. To ensure proper propagation from
the children (i2c devices) to the parent of the adapter (the HW device),
make sure RuntimePM is enabled in any case.
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

This function used to be DT only, so it lived inside a CONFIG_OF block.
Now it uses device attributes and must be moved outside of it. No
further code changes, only one whitespace improvement.
Reported-by: NJim Davis <jim.epost@gmail.com>
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Inspired from the i2c-rk3x driver (thanks guys!) but refactored and
extended. See built-in docs for further information.
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

This patch fixes obvious copy-past error in wake up irq parsing
code which leads to the fact that dev_pm_set_wake_irq() will
be called with wrong IRQ number when "wakeup" IRQ is not
defined in DT.

Fixes: 3fffd128 ("i2c: allow specifying separate wakeup interrupt in device tree")
Signed-off-by: NGrygorii Strashko <grygorii.strashko@ti.com>
Acked-by: NDmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>
Cc: <stable@vger.kernel.org> # v4.3

Although I2C mux devices are easily enumerated using ACPI (_HID/_CID or
device property compatible string match), enumerating I2C client devices
connected through an I2C mux needs a little extra work.

This change implements a method for describing an I2C device hierarchy that
includes mux devices by using an ACPI Device() for each mux channel along
with an _ADR to set the channel number for the device.  See
Documentation/acpi/i2c-muxes.txt for a simple example.

To make this work the ismt, i801, and designware pci/platform devs now
share an ACPI companion with their I2C adapter dev similar to how it's done
in OF.  This is done on the assumption that power management functions will
not be called directly on the I2C dev that is sharing the ACPI node.
Acked-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Tested-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: NDustin Byford <dustin@cumulusnetworks.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

The way we currently scan I2C devices behind an I2C host controller does not
work in cases where the I2C device in question is not declared directly below
the host controller ACPI node.

This is perfectly legal according the ACPI 6.0 specification and some existing
systems are doing this.

To be able to enumerate all devices which are connected to a certain I2C host
controller we need to rework the current I2C scanning routine a bit. Instead of
scanning directly below the host controller we scan the whole ACPI namespace
for present devices with valid I2cSerialBus() connection pointing to the host
controller in question.
Signed-off-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Tested-by: NDustin Byford <dustin@cumulusnetworks.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

A change of return status was introduced in commit 3fffd128
("i2c: allow specifying separate wakeup interrupt in device tree")

The commit prevents the defer status being passed up the call stack
appropriately when dev_pm_domain_attach returns -EPROBE_DEFER.

Catch the PROBE_DEFER and clear up the IRQ wakeup status
Signed-off-by: NKieran Bingham <kieranbingham@gmail.com>
Fixes: 3fffd128 ("i2c: allow specifying separate wakeup interrupt in device tree")
Reviewed-by: NDmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Instead of having each i2c driver individually parse device tree data in
case it or platform supports separate wakeup interrupt, and handle
enabling and disabling wakeup interrupts in their power management
routines, let's have i2c core do that for us.

Platforms wishing to specify separate wakeup interrupt for the device
should use named interrupt syntax in their DTSes:

	interrupt-parent = <&intc1>;
	interrupts = <5 0>, <6 0>;
	interrupt-names = "irq", "wakeup";

This patch is inspired by work done by Vignesh R <vigneshr@ti.com> for
pixcir_i2c_ts driver.

Note that the original code tried to preserve any existing wakeup
settings from userspace but was not quite right in that regard:
it would preserve wakeup flag set by userspace upon driver rebinding;
but it would re-arm the wakeup flag if it was disabled by userspace.

We think that resetting the flag upon re-binding the driver is proper
behavior as the driver is responsible for setting up and handling
wakeups.
Signed-off-by: NDmitry Torokhov <dmitry.torokhov@gmail.com>
Tested-by: NVignesh R <vigneshr@ti.com>
[wsa: updated the commit message]
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

There are devices that need to handle block transactions
regardless of the capabilities exported by the adapter.
For performance reasons, they need to use i2c read blocks
if available, otherwise emulate the block transaction with word
or byte transactions.

Add support for a helper function that would read a data block
using the best transfer available: I2C_FUNC_SMBUS_READ_I2C_BLOCK,
I2C_FUNC_SMBUS_READ_WORD_DATA or I2C_FUNC_SMBUS_READ_BYTE_DATA.
Signed-off-by: NIrina Tirdea <irina.tirdea@intel.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Address collisions will be rare, but we should let the user know that
slaves have their own address space nonetheless.
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

We now have seperate address spaces for 10 bit and we-are-slave clients.
Update the sysfs device instantiation method to support these types by
accepting the address offsets that are assigned to the extra address
spaces. Update the documentation, too.
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

It is not enough to compare the plain address value, we also need to
check the flags enabling a different address space. E.g. it is valid to
have address 0x50 as a 7-bit address and 0x050 as 10-bit address on the
same bus. Same for addresses when we are the slave.
Tested-by: NAndrey Danin <danindrey@mail.ru>
Acked-by: NStephen Warren <swarren@nvidia.com>
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Check for slave and 10-bit flags when probing and mark the client when
found. Improve the address validity check, too
Tested-by: NAndrey Danin <danindrey@mail.ru>
Acked-by: NStephen Warren <swarren@nvidia.com>
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

We want to use this function with struct boardinfo soon, so let's just
pass the parameters really needed. We also extend the type of addr, so
more types can be input. Remove a superfluous dangling comment while
here.
Tested-by: NAndrey Danin <danindrey@mail.ru>
Acked-by: NStephen Warren <swarren@nvidia.com>
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

The current naming is based on the arguments of the functions and not on
what they do. Even I as the maintainer find this confusing, so let's
rename them to something more descriptive.
Tested-by: NAndrey Danin <danindrey@mail.ru>
Acked-by: NStephen Warren <swarren@nvidia.com>
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

We want a separate address range for being an I2C slave. Add an offset
of 0x1000, so it can be combined with ten bit addresses as well. Add a
separate function to create the address value, we will need it later in
other places.
Tested-by: NAndrey Danin <danindrey@mail.ru>
Acked-by: NStephen Warren <swarren@nvidia.com>
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

of_find_i2c_adapter_by_node() call requires quite often missing
put_device(), and i2c_put_adapter() releases a device locked by
i2c_get_adapter() only. In general module_put(adapter->owner) and
put_device(dev) are not interchangeable.

This is a common error reproduction scenario as a result of the
misusage described above (for clearness this is run on iMX6 platform
with HDMI and I2C bus drivers compiled as kernel modules):

    root@mx6q:~# lsmod | grep i2c
    i2c_imx                10213  0
    root@mx6q:~# lsmod | grep dw_hdmi_imx
    dw_hdmi_imx             3631  0
    dw_hdmi                11846  1 dw_hdmi_imx
    imxdrm                  8674  3 dw_hdmi_imx,imx_ipuv3_crtc,imx_ldb
    drm_kms_helper        113765  5 dw_hdmi,imxdrm,imx_ipuv3_crtc,imx_ldb
    root@mx6q:~# rmmod dw_hdmi_imx
    root@mx6q:~# lsmod | grep i2c
    i2c_imx                10213  -1

                                 ^^^^^

    root@mx6q:~# rmmod i2c_imx
    rmmod: ERROR: Module i2c_imx is in use

To fix existing users of these interfaces and to avoid any further
confusion and misusage in future, add one more interface
of_get_i2c_adapter_by_node(), it is similar to i2c_get_adapter() in
sense that an I2C bus device driver found and locked by user can be
correctly unlocked by i2c_put_adapter().
Signed-off-by: NVladimir Zapolskiy <vladimir_zapolskiy@mentor.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

In addition to module_get()/module_put() add get_device()/put_device()
calls into i2c_get_adapter()/i2c_put_adapter() exported
interfaces. This is done to lock I2C bus device, if it is in use by a
client.
Signed-off-by: NVladimir Zapolskiy <vladimir_zapolskiy@mentor.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

If of_find_i2c_device_by_node() or of_find_i2c_adapter_by_node() find
a device by node, but its type does not match, a reference to that
device is still held. This change fixes the problem.
Signed-off-by: NVladimir Zapolskiy <vladimir_zapolskiy@mentor.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Using set_scl may be ineffective before calling the driver specific
prepare_recovery callback, which might change into a test mode. So
instead of setting SCL in i2c_generic_scl_recovery, move it to
i2c_generic_recovery (after the optional prepare_recovery).
Signed-off-by: NJan Luebbe <jlu@pengutronix.de>
Acked-by: NAlexander Sverdlin <alexander.sverdlin@nokia.com>
Tested-by: NAlexander Sverdlin <alexander.sverdlin@gmail.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Mark (and unmark) device nodes with the POPULATE flag as appropriate.
This is required to avoid multi probing when using I2C and device
overlays containing a mux.
This patch is also more careful with the release of the adapter device
which caused a deadlock with muxes, and does not break the build
on !OF since the node flag accessors are not defined then.
Signed-off-by: NPantelis Antoniou <pantelis.antoniou@konsulko.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

sizeof(struct i2c_client) is 1088 bytes on a CONFIG_X86_64=y build and
produces following warning when CONFIG_FRAME_WARN is set to 1024:

drivers/i2c/i2c-core.c: In function ‘acpi_i2c_space_handler’:
drivers/i2c/i2c-core.c:367:1: warning: the frame size of 1152 bytes is
larger than 1024 bytes [-Wframe-larger-than=]

This is not critical given that kernel stack is 16 kB on x86_64 but lets
reduce the stack usage by allocating the struct i2c_client from the heap.
Signed-off-by: NJarkko Nikula <jarkko.nikula@linux.intel.com>
Acked-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Signed-off-by: NShailendra Verma <shailendra.capricorn@gmail.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

int is vague, let's simply use the type of the variable in question.
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: NSimon Horman <horms+renesas@verge.net.au>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Inform users what went wrong from the core, so drivers don't have to do
it.
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Acked-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Following what DT already does. If the device does not have ACPI Interrupt
resource but instead it has one or more GpioInt resources listed below it,
we take the first GpioInt resource, convert it to suitable Linux IRQ number
and pass it to the driver instead.

This makes drivers simpler because the don't need to care about GPIOs at
all if only thing they need is interrupt.
Signed-off-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: NWolfram Sang <wsa@the-dreams.de>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

This is the convention used in most parts of the kernel including DT
counterpart of I2C slave enumeration. To make things consistent do the same
for ACPI I2C slave enumeration path as well.
Signed-off-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: NWolfram Sang <wsa@the-dreams.de>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Commit 523c5b89 ("i2c: Remove support for legacy PM") removed the PM
ops from the bus type, which causes the pm operations on the s3c2410
adapter device to fail (-ENOSUPP in rpm_callback). The adapter device
doesn't get bound to a driver and as such can't have its own pm_runtime
callbacks. Previously this was fine as the bus callbacks would have been
used, but now this can cause devices which use PM runtime and are
attached over I2C to fail to resume.

This commit fixes this issue by marking all adapter devices with
pm_runtime_no_callbacks, since they can't have any.
Signed-off-by: NCharles Keepax <ckeepax@opensource.wolfsonmicro.com>
Acked-by: NBeata Michalska <b.michalska@samsung.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>
Fixes: 523c5b89
Cc: stable@kernel.org

Current -next fails to link an ARM allmodconfig because drivers that use
the core recovery functions can be built as modules but those functions
are not exported:

ERROR: "i2c_generic_gpio_recovery" [drivers/i2c/busses/i2c-davinci.ko] undefined!
ERROR: "i2c_generic_scl_recovery" [drivers/i2c/busses/i2c-davinci.ko] undefined!
ERROR: "i2c_recover_bus" [drivers/i2c/busses/i2c-davinci.ko] undefined!

Add exports to fix this.

Fixes: 5f9296ba (i2c: Add bus recovery infrastructure)
Signed-off-by: NMark Brown <broonie@kernel.org>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

This patch changes type of input parameter for
prepare/unprepare_recovery() callbacks from struct i2c_bus_recovery_info
* to struct i2c_adapter *. This allows to simplify implementation of
these callbacks and avoid type conversations from i2c_bus_recovery_info
to i2c_adapter. The i2c_bus_recovery_info can be simply retrieved from
struct i2c_adapter which contains pointer on it. There are no users
currently, so this is safe to do.
Acked-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: NAlexander Sverdlin <alexander.sverdlin@nokia.com>
Signed-off-by: NGrygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Make sure dynamic ids do not interfere with fixed ones and let them
start after the highest fixed id. This patch might cause different
bus-numbers than before for dynamic ids, however it fixes a bug. Assume:

- fixed id0 defers probe
- fixed id1 succeeds and registers a muxed bus with dynamic id
- muxed bus gets id0
- fixed id0 wants to probe again, but its fixed id is gone now
- fixed id0 probe fails

With this patch, the fixed ids are always reserved in the DT case.
For legacy board init, we already have a mechanism like this in
i2c_register_board_info().
Reported-by: NBob Feretich <bob.feretich@rafresearch.com>
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Now that we have struct fwnode_handle, we can use that to point to
ACPI companions from struct device objects instead of pointing to
struct acpi_device directly.

There are two benefits from that.  First, the somewhat ugly and
hackish struct acpi_dev_node can be dropped and, second, the same
struct fwnode_handle pointer can be used in the future to point
to other (non-ACPI) firmware device node types.
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: NGrant Likely <grant.likely@linaro.org>

Let the core do the checks if HW quirks prevent a transfer. Saves code
from drivers and adds consistency.
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>
Tested-by: NRay Jui <rjui@broadcom.com>
Tested-by: NIvan T. Ivanov <iivanov@mm-sol.com>
Tested-by: NNeelesh Gupta <neelegup@linux.vnet.ibm.com>
Tested-By: NLudovic Desroches <ludovic.desroches@atmel.com>

This reverts commit e4df3a0b
("i2c: core: Dispose OF IRQ mapping at client removal time")

Calling irq_dispose_mapping() will destroy the mapping and disassociate
the IRQ from the IRQ chip to which it belongs. Keeping it is OK, because
existent mappings are reused properly.

Also, this commit breaks drivers using devm* for IRQ management on
OF-based systems because devm* cleanup happens in device code, after
bus's remove() method returns.
Signed-off-by: NJakub Kicinski <kubakici@wp.pl>
Reported-by: NSébastien Szymanski <sebastien.szymanski@armadeus.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: NDmitry Torokhov <dmitry.torokhov@gmail.com>
[wsa: updated the commit message with findings fromt the other bug report]
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>
Cc: stable@kernel.org
Fixes: e4df3a0b

There was quite some confusion why this completion is there and if it is
still necessary. Sadly, it is. However, let's improve the comments and
share what we rediscovered.
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Make the slave support depend on CONFIG_I2C_SLAVE. Otherwise it gets
included unconditionally, even when it is not needed.

I2C bus drivers which implement slave support must select
I2C_SLAVE.
Signed-off-by: NJean Delvare <jdelvare@suse.de>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>