greybus: legacy: remove unimplemented classes from id table

Remove the unimplemented display, sensor, and svc classes from the
device-id table.

As Viresh noted the SVC protocol is special and having an SVC class
doesn't really make sense at all. Either way, the SVC protocol is
already implemented by core.

Testing Done: Compiled

Reported-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: arche-platform: reduce wait between WAKE_OUT checks

SVC WAKE_OUT loop is estimated 400ms during which wake_detect line
is pulled low for AP detection.  On AP side we have 500ms delay
between checks.  To avoid timing issues, reduce delay between
checks and raise total # of checks so that overall time for sequence
is the same.

Testing Done:
- Used for DB3.5/EVT1.5 hardware during bringup
- Regression tested on DB3.1+ES2, DB3.1+ES3

Signed-off-by: Michael Scott <michael.scott@linaro.org>
Reviewed-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Tested-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: arche-platform: assert wake-detect to complete WAKE_OUT event

After SVC generates WAKE_OUT (pulling wake-detect pin low) and APB is
brought out of reset, we need to assert wake-detect again to complete
SVC WAKE_OUT logic.

Testing Done:
- Used for DB3.5/EVT1.5 hardware during bringup
- Regression tested on DB3.1+ES2, DB3.1+ES3

Signed-off-by: Michael Scott <michael.scott@linaro.org>
Reviewed-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Tested-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: arche-platform: bring SVC out of reset later in probe

Move SVC deassert reset to after wake-detect pin has been pulled
low in probe.  Otherwise, SVC may trigger WAKE_OUT based on a
default high signal.

Testing Done:
- Used for DB3.5/EVT1.5 hardware during bringup
- Regression tested on DB3.1+ES2, DB3.1+ES3

Signed-off-by: Michael Scott <michael.scott@linaro.org>
Reviewed-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Tested-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: arche-platform: ensure wake-detect pin is deasserted

During DB3.5 bringup, it was noted that wake_detect signal was not
properly generating SVC edge IRQ.  To ensure signal goes from low
to high correctly, let's bring signal low (regardless of default
pin state).

Testing Done:
- Used for DB3.5/EVT1.5 hardware during bringup
- Regression tested on DB3.1+ES2, DB3.1+ES3

Signed-off-by: Michael Scott <michael.scott@linaro.org>
Reviewed-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Tested-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: hd: fix host-device-removal race

Make sure to tear down the svc and flush any on-going hotplug processing
before removing the remaining interfaces.

This fixes crashes due to host-device removal racing with svc
hotplug/unplug processing (e.g. at "UniPro restart").

Testing Done:
Verified that this fixes crashes reproducible on SDB when unloading the
host-device driver module while generating hotplug/unplug events.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Rename Audio class and remove the unused one

There should be a single class macro for Audio and two protocol macros.
Rename class with value 0x12 as GREYBUS_CLASS_AUDIO and remove the other
unused class GREYBUS_CLASS_AUDIO_DATA.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio_codec: convert to bundle driver

Convert the legacy audio management and data protocol drivers to a
bundle driver.

The Audio bundle driver can support a single management and any number
of data cports, and so we expect multiple data cports to be present for
the bundle during initialization.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio_codec: Free gccodec on dia probe failure

We aren't freeing the codec, that we allocated before failing to probe
the connection. Free it.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio_codec: Free gccodec on codec probe failure

We aren't freeing the codec, that we allocated before failing to probe
the connection. Free it.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: legacy: fix a null pointer dereference

When gb_protocol_get() fails in legacy_connection_create(), we end up
bailing-out before assigning lc->connection and lc->protocol. Calling
legacy_connection_destroy() in that case results in a null pointer
dereference.

Check if lc->connection is not null before freeing it.

Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: loopback_test: make output to csv file a parameter option

Its useful to get a CSV output on stdout for test frameworks to read
and parse the results. However, a csv file is not always needed.

Add the -z option to  create/append a csv file only when the user
asks for it.

Signed-off-by: Axel Haslam <ahaslam@baylibre.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: svc watchdog: allow it to be enabled/disabled from userspace

Sometimes you want to disable the SVC watchdog without having to patch
your kernel, so provide a sysfs file to do this.  This can let us do
power measurements, and let the firmware developers not go crazy
worrying that the kernel is going to reset their chips with no notice.

Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: camera: only build module against msm kernel

The camera driver currently does not build against anything other than
the msm kernel, due to cross-dependancies, so enable that here so that
we can build against other kernels without failing the build.

Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Tested-by: Marti Bolivar <mbolivar@leaflabs.com>
Acked-by: Laurent Pinchart <laurent.pinchart@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: svc: double the ping delay timeout

Sometimes the ping response comes back _right_ after we timed out, as
the svc got its act together and squeaked out the ack, yet we miss it
and reset the whole bus.  Double the delay to hopefully give the svc a
little more of a chance to fix itself.  Odds are, it's still in trouble,
but we can just hold off resetting it for a bit more...

Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: arche-platform: Disable clock as part of driver remove

As part of driver remove (cleanup) function, disable the clock for both
SVC, APB1 & APB2.

Testing Done:
Tested on EVT1 platform with Connect=>disconnect=>connect
iteration, almost close to 100 iterations have passed (demo branch).
And also tested with kernel-only build.

Signed-off-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: camera: Stream config change unipro speed

Unipro network speed was increased at camera initialization time and
never slowed down.
This unnecessary drains power during the entire time camera module is
plugged in.
Increasing/decreasing unipro link speed before issuing stream
configuration request to camera module prevents this from happening.

Signed-off-by: Jacopo Mondi <jacopo.mondi@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: svc: return proper error code on watchdog create error

When the watchdog is not created properly, we need to propagate the
error for this and not just return success.

Reported-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: svc: add a "watchdog" to check the network health

Now that we have a svc ping command, let's add a watchdog to call it
every so often (1 second at the moment.)  If it finds something went
wrong, post a stern message to the kernel log and call:
start unipro_reset
to reset the whole greybus hardware subsystem.

Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Reviewed-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: lsgb: Minor tweaks

Make it executable on an Android system
Change layout to be show heirachy a bit better.

Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: camera: add semiplanar and planar formats

This change adds missing planar and semiplanar
formats from gb specification.

Mbus to Gb format map:
V4L2_MBUS_FMT_NV12_1x8 -> 0x12
V4L2_MBUS_FMT_NV21_1x8 -> 0x13
V4L2_MBUS_FMT_YV12_1x8 -> 0x16
V4L2_MBUS_FMT_YU12_1x8 -> 0x17

Change depends on:
"media: add new mediabus format enums for ara camera"

Signed-off-by: Gjorgji Rosikopulos <grosikopulos@mm-sol.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: svc: Expose and retain VID/PID received from bootrom for ES2

ES2 chips doesn't have efuses for storing module's vendor_id and
product_id and so we have hacked bootrom earlier using firmware
protocol, so that VID/PID can be received for fetching firmware
packages.

Another requirement is to expose them to sysfs, so that modules can be
identified properly.

That can be easily solved by updating interface's VID/PID, when fetched
using firmware protocol and later reusing them while the module switches
its identity from bootrom to firmware.

Do that only if the module is ES2 and the VID/PID sent during hotplug
are both 0.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: lsgb: add lsgb shell script

Horrid hack, but a start at making it easier for people to see what is
currently connected without having to deal with poking around in sysfs
directly.

Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio_codec: update codec_name as per driver->name

Originally, driver->name was not poluated from GB and thus
manually set from audio_codec driver as a hack.
This is no more required.
Another patch already removes that hack.

Now, with new driver->name as "legacy.<id-bundle.interface>"
codec is registered with different name.
So, during DAI link registration as well it needs modification.

Signed-off-by: Vaibhav Agarwal <vaibhav.agarwal@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: firmware: abort if AP_READY fails

Abort if the AP_READY request fails.

Also update the comment suggesting that this operation was a temporary
one.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: vibrator: convert to bundle driver

Convert the legacy vibrator protocol driver to a bundle driver.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: hid: convert to bundle driver

Convert the legacy HID protocol driver to a bundle driver.

This also fixes a potential crash should a (malicious) module have sent
an early request before the private data had been initialised.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: hid: clean up init error paths

Separate success and error paths more clearly.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: set request handlers at creation

Set the connection request handler at creation rather than when enabling
the connection.

This is possible now that connections are created by the drivers that
use them rather than by core at manifest parsing time.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: legacy: look up protocol at connection creation

Look up legacy protocol at connection create instead of at init.

Note that we can now look up the protocol before even creating the
actual connection. This is needed to be able to set a connection request
handler when creating the connection rather than when enabling it.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: remove legacy protocol id from core

Remove legacy protocol-id field that is only used by the legacy-protocol
driver from the connection structure.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: legacy: add initialized flag

Add initialized flag and use instead of the connection protocol pointer
to determine when the legacy connection has been initialised.

This is a step in moving legacy connection binding to
connection-creation time.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: legacy: refactor legacy-connection handling

Add wrapper structure for legacy connections and add helpers to create
and destroy legacy connections.

This is a step in removing legacy connection fields from the core
structures.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: return error-valued pointer on creation errors

Return an ERR_PTR on errors when creating connections.

This allows driver probe to fail with a more informative error message
as not all connection creation errors are due to memory exhaustion.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: rename connection-create interface

Drop the _dynamic suffix from the exported interface that drivers use to
create connections.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: simplify connection-creation interface

Simplify the exported generic interface for creating dynamic
connections.

Since all driver connections will have a bundle we can drop the
redundant interface parameter.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: drop the legacy protocol-id parameter

The protocol-id parameter is only used by the legacy driver so remove
it from the generic interface and move it to the legacy driver until it
can be removed completely.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: add helper to create control connections

Add dedicated helper to create control connections.

This will allow us to simplify the generic (dynamic) interface.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: drop protocol parameter from static interface

Drop legacy protocol parameter from static connection interface.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: manifest: check for duplicate CPort descriptors when parsing

Now that connection creation has been separated from interface
initialisation, we should explicitly check for duplicate CPort
descriptors when parsing the manifest.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: core: defer connection creation to driver probe

Defer connection creation to bundle driver probe instead of creating
them when initialising the interface and parsing the manifest.

Store copies of the CPorts descriptors in the bundle for the drivers to
use, and update the legacy driver.

This is needed for drivers that need more control over host-device
resource management, for example, when a protocol needs to use a
dedicated host CPort for traffic offloading (e.g. camera data).

This also avoids allocating host CPorts for bundles that are not bound
to a driver or for remote CPorts that a driver does not need.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: add bundle class to the bundle uevent

When bundles are added and then removed, we have a race where we go to
read the sysfs file, but it is now for a different bundle than the
uevent was originally for.  So add the bundle class to the uevent so we
"know" what the correct bundle class was.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: svc: add ping command

This implements the SVC "ping" command.  It's tiny and simple, but we
need something like this in order for us to "know" if all is working
well.

Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: firmware: log the name of the firmware being requested

People are getting confused as to what the firmware file name is for
their module, so log it to make it easier to find modules with "blank"
vid/pid values and let people know what to name their firmware files.

Because we are now logging this, turn the debug message after the
firmware request into an error message if something fails.

Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Reviewed-by: Michael Scott <michael.scott@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: svc: add key event handling

Add a new input device associated with the SVC to handle key events.
This new events are transfer over a new greybus svc operation which is
unidirectional.

It was selected the KEY_A for representing the KEY_ARA_BUTTON key code.

Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio_codec: Don't be tricky with the driver model

With the recent changes by Johan to the driver model of the greybus
code, the audio_codec no longer should need to provide a "dummy" driver
when registering the codec.  In fact, having it there causes the greybus
core to crash when inserting the module.  Removing it all fixes the
crash.

Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: legacy: add private driver data

Allocate private data structure at probe and release at disconnect.

The private data will be used to track the connections managed by the
driver, but for now only holds a count of the number of cports the
bundle has.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: make cport lookup thread-safe

Use the gb_connection_mutex when making sure a remote CPort is
available.

This is needed when moving to driver-managed connections that can be
created and destroyed at any time.

Also check for duplicate bundle-less connections.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: serialise connection creation

Serialise connection creation against concurrent creation and
destruction using a global mutex.

This is needed to prevent two drivers from attempting to create a
connection to the same interface CPort and to cope with a racing
connection destroy when moving to driver managed connections.

Note that the locking can not (easily) be made more fine-grained as not
all connections have an interface, but these are not hot paths anyway.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: fix lookup race

Fix longstanding race that could lead to a connection being destroyed
while processing an incoming message due to missing reference counting
when looking up the connection.

Add helpers to manipulate the connection kref.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: drop the connection_mutex

Drop the useless connection_mutex that did not, and can not be used to
prevent connection lookup races in atomic context and therefore serves
no purpose.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: destroy workqueue at unregister

Destroy the work queue when the connection is destroyed/deregistered
instead of when the last reference is dropped.

The work queue is not needed once the connection has been deregistered,
and no operations will ever be added to it again (handled by checking
connection->state) even if the connection object may not be deallocated
until the final reference is dropped.

The work-queue name is set based on the host-device name and host-device
cport id, something which guarantees a unique name. This would no longer
be true if the work queue was not destroyed at connection deregistration
as a new connection could then be created for that very same host cport.

This is not necessarily a problem unless some work queue features are
used that require unique work-queue names, but let's try to be well
behaved.

Also update an obsolete comment and make explicit that a connection must
be disabled before being destroyed.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: remove WARN_ON from destroy

Remove WARN_ON from connection destroy testing for a NULL-connection
when destroying a connection.

This will allow for simpler driver code when releasing driver-managed
connections.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: control: add bundle-version operation

Add bundle-version operation to fetch the version of the bundle class.

Retrieve the bundle version of all bundles when initialising the
interface in case the control-protocol version is greater than 0.1.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: remove broken protocol-version handling

Remove the broken legacy protocol-version handling from core and move it
to the legacy driver instead.

Note that version handling has always been broken as legacy protocols
were looked up and bound to connections before the connections were
enabled and version negotiation could take place (over that very
connection). Document that in the legacy driver as well.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: svc: store protocol version

Store the "negotiated" protocol version to use in the svc state
struct instead of the connection struct.

The generic concept of a connection version is going away in favour of
bundle-class versions.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: core: disable bundle connections on hot-unplug

Disable bundle connections in core before calling driver disconnect in
case the interface is already gone.

This avoids unnecessary timeouts on hot-unplug when a driver does I/O in
its disconnect callback.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: legacy: use protocol pointer to determine state

Use the protocol pointer to determine the legacy connection state.

This is needed to allow core disable connections when an interface has
been hot-unplugged without the legacy protocols leaking its resources.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: move legacy-protocol handling to legacy driver

Move legacy protocol and connection handling to the legacy driver.

Rename the former global functions using a common legacy_ prefix.

Note that all legacy protocols suffer from a connection initialisation
race in that the protocol-specific initialisation, which includes
allocation of protocol-specific state containers, can not happen until
*after* the connection has been enabled. This is a major flaw in the
original design that we can now finally address by converting the legacy
protocol drivers into proper bundle (class) drivers.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: svc: drop legacy-protocol dependency

Drop dependency on the legacy protocol abstraction.

Remove the now unused and last legacy-protocol flag
GB_PROTOCOL_SKIP_VERSION along with the protocol-flag feature.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: control: drop legacy-protocol dependency

Drop dependency on the legacy protocol abstraction.

Instead implement the protocol-specific version request directly, and
use the new interface for managing the control connection.

Note that the version request is being removed from most protocols, but
we need to keep the current request for the control protocol as-is
indefinitely to maintain backwards compatibility (e.g. with the ES2/ES3
bootrom).

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: greybus_protocols: remove control-protocol version

Remove control-protocol version from the exported protocol definitions
as it is an implementation detail that makes no sense to export.

Currently gbsim uses the kernel's control-protocol version definitions
directly instead of reporting the version of the protocol it actually
implements.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: remove skip-connected legacy protocol flags

Remove the legacy protocol flags that were used to suppress the
connected and disconnected events.

Instead send the connected and disconnected event for all bundle
connections and explicitly exclude static connections and control
connections.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: firmware: remove skip-disconnected flag

Remove the legacy protocol flag that was used to suppress sending the
control disconnected operation.

Instead rely on the fact that no control operations will be sent by
core to an interface that has been removed (e.g. after having received a
new hotplug event after the bootrom has jumped to the new image).

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: core: add defensive connection disable post disconnect

Bundle drivers *must* disable their connections in the disconnect
callback, but add a defensive test and warn about buggy drivers
nonetheless.

Note that bundle drivers would generally release their state containers
in disconnect so a failure stop I/O could potentially lead to
use-after-free bugs in any late operation completion callbacks.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: core: disable incoming operations pre disconnect

Disable and flush incoming operations before calling driver disconnect.

Bundle drivers are still responsible for disabling their connections
in their disconnect callback.

Note that specifically the legacy protocols must have incoming
operations disabled when their connection_exit callback is called as
that is where their state is deallocated.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: add helper to disable incoming operations

Add helper to disable and flush incoming operations.

This is intended to be used by core to flush any incoming requests
before calling driver disconnect, but could potentially later be
exported for driver use as well.

Note that we currently flush all incoming operation and allow the
request handlers to run, but cancel any responses sent. This may need to
be refined later.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: add unidirectional enabled state

Add a new connection state ENABLED_TX in which only outgoing operations
are allowed.

This allows drivers to query the device during probe before allocating
their state containers without having to worry about racing incoming
requests.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: make connection enable/disable thread safe

Add connection mutex to protect connection enabling and disabling.

This is needed to eventually allow drivers to manage the state of their
connections during operation (i.e. post probe and pre disconnect).

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: disable operations on enable errors

Make sure to cancel all (incoming) operations when failing to enable a
connection.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: clean up operation cancellation on disable

Move helper to cancel active operations above gb_connection_enable and
simplify locking.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: always cancel operations on connection disable

Always cancel all operations on connection disable and remove the now
unused DESTROYING state.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: add per-connection request handlers

Add a connection request-handler field to struct gb_connection that is
set when the connection is enabled.

This is a step towards removing the legacy protocol abstraction from
core, and will also be used to implement unidirectional connection
states (e.g. only outgoing operations are allowed).

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: remove disable from destructor

Remove implicit disable of legacy connections from the destructor.

This is a step towards removing the legacy-protocol handling from core.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: rename legacy init and exit functions

Rename legacy connection init and exit functions.

This is a step towards removing the legacy-protocol handling from core.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: legacy: add legacy-protocol bundle driver

Add the first Greybus bundle driver that will be used when transitioning
from legacy Greybus protocols to bundle drivers.

The legacy-protocol driver initially binds to all current bundle classes.

In order to avoid having to update current module-loading scripts, keep
this driver internal to greybus core at least until modalias support is
added. Note that this prevents unloading any protocol drivers without
first tearing down the host device due to a circular module dependency.

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: camera: Update configure stream based on new interface

Interface has been changed.
return code will not return number of configured streams but
just error code.
Number of streams is passed as pointer and if operation result
is changed number of streams will be updated.

Flags are also used for information regarding configure stream
operation result.

Signed-off-by: Gjorgji Rosikopulos <grosikopulos@mm-sol.com>
Acked-by: Laurent Pinchart <laurent.pinchart@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: camera: Add support for configure streams flag in gb interface

Update gb interface and export flags needed for latest
protocol version. Number of streams also can be changed
based on operation result.
Caller sets input flags, end fucntion return output
flags

Input flags supported:
- GB_CAMERA_IN_FLAG_TEST - Need to be set when operation
is not actually applied.

Output flags supported:
- GB_CAMERA_OUT_FLAG_ADJUSTED - This is result of the operation
if this flag is set, result is adjusted and operation
need to be repeat.

Signed-off-by: Gjorgji Rosikopulos <grosikopulos@mm-sol.com>
Acked-by: Laurent Pinchart <laurent.pinchart@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: firmware: replace colons with underscore in firmware file request

Due to some issues with handling colons in Android (and possibly
future exposure to other problems) we should remove the colons from
the firmware file request.  Replacing them with underscores.

Specifically, we copy firmware into the Android build using a line
similar to this:
PRODUCT_COPY_FILES += \
    source-repo/ara:00000126:00001000:00000001:00000001:02.tftf:system/etc/firmware/ara:00000126:00001000:00000001:00000001:02.tftf

There is a colon delimiter between the source and destination in the
PRODUCT_COPY_FILES format.  The greybus naming logic breaks the parsing
routine and generates an Android build break.

Signed-off-by: Michael Scott <michael.scott@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: Revert "camera: Fix backword compatibility in configure streams"

This reverts commit 48cc91e52dba9abad4c9b56f911994c65071bfc4 as the
caller should be changed instead.

Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: camera: Fix backword compatibility in configure streams

Configure streams ret code should be number of streams in
gb operation exported to HOST camera.
Until gb operation is migrated to new interface return
number of streams in ret code.

Signed-off-by: Gjorgji Rosikopulos <grosikopulos@mm-sol.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: camera: Return the result flags from the configure_streams response

And return the response num_streams field through a pointer variable
instead of using the return value of the function as both an error code
and a positive number of streams, it makes the API more consistent.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Enable codec module detection on different slots

driver_name associated with dev_name was hard coded earlier.
This limits, audio codec module to be detected on Port#5 only.

Now, driver_name is generated dynamically based on dev_name.
This enables codec module detection on any 1x2 slot.

Also, Update dev_id based on slot number, instead of bundle->id.
bundle->id is not unique for multiple modules added, thus now
using slot number for unique identification.

Signed-off-by: Vaibhav Agarwal <vaibhav.agarwal@linaro.org>
Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Cleanup GB protocol connections in case of abrupt codec removal

We need to clean up GB protocl connections, otherwise successive
codec insertions fails repeatedly.

NOTE: As per suggestion, since codec is already removed, one should
not trigger any GB command. It'll cause a delay of atleast TIMEOUT
value.
HOwever, failing to cleanup GB protocol, causes successive module
insertion to fail

Signed-off-by: Vaibhav Agarwal <vaibhav.agarwal@linaro.org>
Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: cleanup unnecessary dev_err messages

Replace unnecessary dev_err msg with dev_dbg.
Same were added during development to trace topology
parser progress.

Signed-off-by: Vaibhav Agarwal <vaibhav.agarwal@linaro.org>
Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Report uevent on GB codec module insertion/removal

GB-Audio-manager module is currently used to report uevent
to above layer in response to any codec module inserted or
removed.

Signed-off-by: Vaibhav Agarwal <vaibhav.agarwal@linaro.org>
Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Add GB Audio class protocol functionality in GB codec DAI ops

GB Audio class driver provides APIs to configure GB codec module.
This patch adds relevant operations in DAI ops callback functions to
configure codec module as per DAPM sequence triggered.

Signed-off-by: Vaibhav Agarwal <vaibhav.agarwal@linaro.org>
Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Add topology parser for GB codec

For each GB codec module inserted, DAPM widgets, kcontrols, routes
and DAIs can be fetched through greybus in a binary chunk and parsed
locally to create & populate DAPM graph for the specific module.

It is required by each codec module to populate a minimum set of
kcontrols with fixed names to support basic audio usecase.
To support advanced features of codec module, the same can be polpulated
with existing topology parser. However, to use them for different usecase
separate mechanism (may be via MSP) is required to inform userspace about
their configuration value & enable/disable sequence.

ToDos:
Currently, support for enumerated kcontrol/dapm control is hardcoded.
Need to add complete logic within the parser.

Signed-off-by: Vaibhav Agarwal <vaibhav.agarwal@linaro.org>
Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Use greybus connection device for codec registration

Use GB Audio mgmt, data protocol ids to register codec module with
GB protocol. And in response to mgmt->connection_init(), register
GB codec driver with ASoC.

Now, using msm8994  machine to register DAI link dynamically on
codec insertion.

ToDos:
- snd_soc_register_codec() uses driver->name to identify device id.
  However, for GB device, .driver{} is not yet populated by GB core.
  Thus, defining dummy structure within codec driver. This should
  come from GB core itself.
  Even existing .driver{} may cause problem in case of multiple
  modules inserted or inserted at a different slot.
- Fix logic for gbcodec->dais & gbcodec->dailinks. Current
  implementation contains some hard coded data with assumption of
  count=1.
- Evaluate definition of 'gbaudio_dailink.be_id' in case of multiple
  DAI links.

Signed-off-by: Vaibhav Agarwal <vaibhav.agarwal@linaro.org>
Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: Makefile: fix up build test for gb-audio-codec.ko

We really only want to build the module if the config is set, not all
the time like the current code does.

Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Build audio module conditionally

Added CONFIG_XXX flag check before compiling audio module.
Once we add dynamic DAI link registration from audio driver,
this check wil be required to avoid compilation failures with
other kernel revisions.

Also, renamed header file to better align with .c file name.

Signed-off-by: Vaibhav Agarwal <vaibhav.agarwal@linaro.org>
Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Add Audio Manager

This is a simple module that keeps a list of connected GB audio
modules.

Whenever a device is attached, an appropriate uevent is sent to
userspace:

    UDEV  [4941.803215] add      /kernel/gb_audio_manager/0 (gb_audio_manager)
    ACTION=add
    CPORT=99
    DEVICES=0x10
    DEVPATH=/kernel/gb_audio_manager/0
    NAME=naim
    PID=64
    SEQNUM=1828
    SLOT=2
    SUBSYSTEM=gb_audio_manager
    USEC_INITIALIZED=802416
    VID=128

And whenever removed:

    UDEV  [4941.836588] remove   /kernel/gb_audio_manager/0 (gb_audio_manager)
    ACTION=remove
    DEVPATH=/kernel/gb_audio_manager/0
    SEQNUM=1833
    SUBSYSTEM=gb_audio_manager
    USEC_INITIALIZED=835681

The API consists of functions for adding, removing and inspecting
added device module descriptions (struct gb_audio_module):

    int                         gb_audio_manager_add(struct gb_audio_module_descriptor *desc);
    int                         gb_audio_manager_remove(int id);
    int                         gb_audio_manager_remove_all(void);
    struct  gb_audio_module*    gb_audio_manager_get_module(int id);
    void                        gb_audio_manager_put_module(struct gb_audio_module *module);
    int                         gb_audio_manager_dump_module(int id);
    void                        gb_audio_manager_dump_all(void);

Devices can be inspected through sysfs in /sys/kernel/gb_audio_manager/{id}/*

If GB_AUDIO_MANAGER_SYSFS is exported as 'true', managing devices can be done
via the SYSFS as well. For instance:

    echo name=naim slot=2 vid=128 pid=64 cport=99 devices=0x10 > /sys/kernel/gb_audio_manager/add
    echo all > /sys/kernel/gb_audio_manager/dump
    echo 2 > /sys/kernel/gb_audio_manager/dump
    echo 2 > /sys/kernel/gb_audio_manager/remove

Signed-off-by: Svetlin Ankov <ankov_svetlin@projectara.com>
Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Add direct audio streaming control for APBridgeA

The latest audio architecture does not stream audio data over the
USB link between the AP and APBridgeA.  Instead, audio data is
streamed directly over the I2S link between the two.

To support the Greybus audio driver in setting up the I2S port and
controling packeting/depacketizing of audio data to/from Greybus
audio messages, define a set of commands and their parameters.
These commands and parameters will be placed into a request and sent
over the USB control channel to APBridgeA.  A corresponding driver
on APBridgeA will receive and process the requests.

Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Add Greybus Audio Device Class Protocol helper routines

Add helper routines to make communicating with audio modules easier.

Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Add Audio Device Class Protocol definitions

Add the macros and structures for the Greybus Audio Device
Class Protocol.

Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: audio: Use underscore in file name

For consistency with most other files in the Greybus repository,
change 'audio-codec.c' to use an underscore instead of a hyphen
in its name.

Signed-off-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: arche-apb-ctrl: Enable the clocks after regulator enable

It makes more sense to enable the clock after power is enabled
to the device, so move clock enable code after regulator_enable
and setting up BOOT_RET pin.

Signed-off-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Tested-by: Michael Scott <michael.scott@linaro.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@linaro.org>
Tested-by: Laurent Pinchart <laurent.pinchart@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: camera: HACK: Register gb camera to the HOST camera

This change implements gb camera interface and register
gb camera to the HOST driver.
This implementation need to be reworked after the demo.

Tested with db3 with two camera modules.

Signed-off-by: Gjorgji Rosikopulos <grosikopulos@mm-sol.com>
Acked-by: Laurent Pinchart <laurent.pinchart@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: camera: HACK: Export GB camera interface

Gb camera need to communicate with HOST driver.
as temporary solution there will be shared header
between gb camera HOST camera.
Both need to be in sync since gb drivers are compiled
out of the kernel tree.

Gb camera register camera operation functions when is created.
Currently unregister is not supported.

Signed-off-by: Gjorgji Rosikopulos <grosikopulos@mm-sol.com>
Acked-by: Laurent Pinchart <laurent.pinchart@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: lights: remove sync operation and work queue

In kernel v4.5 there is a change in LED api, which remove the need for
individual work queue and rename the set_sync operation to set_blocking.
This patch add the handling of this case and avoid compilation failure
for this kernel versions.

Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>