GitHub/LineageOS/android_kernel_motorola_exynos9610.git
8 years agogreybus: interface: fix interface-registration race
Johan Hovold [Mon, 7 Dec 2015 14:05:45 +0000 (15:05 +0100)]
greybus: interface: fix interface-registration race

Fix race with user space when registering the interface.

The interface was registered before having been fully initialised,
something which could lead to user space accessing not-yet-initialised
attribute values (e.g. zero vendor and product ids or empty vendor and
product strings).

Note that this is also needed to be able to let attribute visibility
depend on manifest data (e.g. interface unlock).

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: interface: separate manifest parsing from bundle registration
Johan Hovold [Mon, 7 Dec 2015 14:05:44 +0000 (15:05 +0100)]
greybus: interface: separate manifest parsing from bundle registration

Separate manifest parsing, including bundle and connection creation,
from bundle registration.

Note that this is also needed to allow the interface to not be
registered until the manifest has been parsed.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: bundle: separate bundle creation and registration
Johan Hovold [Mon, 7 Dec 2015 14:05:43 +0000 (15:05 +0100)]
greybus: bundle: separate bundle creation and registration

Separate bundle creation and registration.

Note that the bundle connections still needs to be initialised post
registration as protocol drivers create child devices to the bundle.

This will ultimately allow connection structures to be created while
parsing manifests, but the connections to not be enabled until a driver
is bound.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: bundle: separate connection disabling and destruction
Johan Hovold [Mon, 7 Dec 2015 14:05:42 +0000 (15:05 +0100)]
greybus: bundle: separate connection disabling and destruction

Separate bundle-connection disabling and destruction, and destroy the
connections along with the bundle.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: bundle: use dev_err for bundle-creation errors
Johan Hovold [Mon, 7 Dec 2015 14:05:41 +0000 (15:05 +0100)]
greybus: bundle: use dev_err for bundle-creation errors

Use dev_err to report duplicate bundle ids when creating a bundle.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: bundle: kill the bundle lock
Johan Hovold [Mon, 7 Dec 2015 14:05:40 +0000 (15:05 +0100)]
greybus: bundle: kill the bundle lock

Kill the bundle lock, which looked like it protected the interface
bundle lists but really did not as lock-less manipulations were still
made.

No locking for the interface bundle list is in fact needed as bundles
are created along with the interface, and the list is only used to check
for duplicate bundle ids when parsing the manifest and to destroy the
bundles when removing the interface itself.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: interface: remove defensive WARN_ON
Johan Hovold [Mon, 7 Dec 2015 14:05:39 +0000 (15:05 +0100)]
greybus: interface: remove defensive WARN_ON

Remove defensive WARN_ON testing for a NULL-interface when removing an
interface.

Every call site has just dereferenced the interface.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: use decimal notation for interfaces, bundles and cports
Johan Hovold [Mon, 7 Dec 2015 14:05:38 +0000 (15:05 +0100)]
greybus: use decimal notation for interfaces, bundles and cports

Fix up the last few places where hexadecimal rather than decimal
notation was used for interface, bundle and cport ids.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: create svc along with host device
Johan Hovold [Mon, 7 Dec 2015 14:05:37 +0000 (15:05 +0100)]
greybus: svc: create svc along with host device

Create svc device along with host-device and move the svc-connection to
the svc structure.

The svc connection is enabled when registering the host device, but as
the SVC protocol is currently driven by the SVC, we need to defer
registration of the svc device to the connection request handler.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: hd: make svc-connection life time coincide with host-device
Johan Hovold [Mon, 7 Dec 2015 14:05:36 +0000 (15:05 +0100)]
greybus: hd: make svc-connection life time coincide with host-device

Create the svc-connection as part of the host device, and destroy it in
the host-device destructor.

The svc-connection is enabled when registering the host device, and
disabled when the host device is deregistered.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: hd: initialise device last
Johan Hovold [Mon, 7 Dec 2015 14:05:35 +0000 (15:05 +0100)]
greybus: hd: initialise device last

Initialise the struct device last when creating the host device.

After device_initialize(), or rather dev_set_name(), we must use
put_device to release the host device. Initialising last will allow for
a simpler release callback.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: separate connection creation and enabling
Johan Hovold [Mon, 7 Dec 2015 14:05:34 +0000 (15:05 +0100)]
greybus: connection: separate connection creation and enabling

Separate connection creation from enabling.

This will ultimately allow connection structures to be created while
parsing manifests, but the connections to not be enabled until a driver
is bound.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: unbind protocol at exit
Johan Hovold [Mon, 7 Dec 2015 14:05:33 +0000 (15:05 +0100)]
greybus: connection: unbind protocol at exit

Unbind protocol at connection exit rather than when the connection is
destroyed.

Now a protocol is only bound while a connection is enabled.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: bind protocol at init
Johan Hovold [Mon, 7 Dec 2015 14:05:32 +0000 (15:05 +0100)]
greybus: connection: bind protocol at init

Bind protocol at connection init.

This is an intermediate step in separating connection creation and
enabling.

Note that the protocol is currently still unbound when the connection is
destroyed.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: fail on missing protocol
Johan Hovold [Mon, 7 Dec 2015 14:05:31 +0000 (15:05 +0100)]
greybus: connection: fail on missing protocol

Make sure to fail properly when a protocol is missing.

This prevents the connection from being created, which is fine as we
currently never bind protocols post creation.

This is an intermediate step in moving protocol binding to
connection_init.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: bundle: kill gb_bundle_bind_protocols
Johan Hovold [Mon, 7 Dec 2015 14:05:30 +0000 (15:05 +0100)]
greybus: bundle: kill gb_bundle_bind_protocols

Remove gb_bundle_bind_protocols() that was used to iterate over all
registered bundles and bind protocols to them should a protocol become
available post bundle creation.

The protocol abstraction as a generic construct is going away in favour
of class drivers. Connections will be setup when a class driver is
probed, and driver modules can be loaded on-demand by user space based
on uevents and modalias.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: bundle: remove obsolete function prototype
Johan Hovold [Mon, 7 Dec 2015 14:05:29 +0000 (15:05 +0100)]
greybus: bundle: remove obsolete function prototype

Remove obsolete function prototype that was left after a recent code
relocation.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: add missing boot-status error message
Johan Hovold [Mon, 7 Dec 2015 14:05:28 +0000 (15:05 +0100)]
greybus: svc: add missing boot-status error message

Make sure to print an error message when aborting hotplug processing due
to failure to clear the interface boot status.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: loopback: Ensure we reset stats once and once only
Bryan O'Donoghue [Mon, 7 Dec 2015 01:59:08 +0000 (01:59 +0000)]
greybus: loopback: Ensure we reset stats once and once only

9445c54c ('greybus/loopback: drop bus aggregate calculation') removed the
aggregation of data in-kernel but instead of dropping the reset of
aggregate stastics, converted that reset into a second reset of the
connection-level stats. While this doesn't result in anything bad it's
also definitely a dumb thing to be doing, so, drop it now.

Also ensure we reset the bridge-specific tracking variables at least once.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: loopback: Convert thread delay to microseconds
Bryan O'Donoghue [Mon, 7 Dec 2015 01:59:07 +0000 (01:59 +0000)]
greybus: loopback: Convert thread delay to microseconds

Currently the loopback code allows a delay between operations specified in
milliseconds. Having added asynchronous bi-directional support to loopback
its obvious that the delay value would be far more useful specified in
microseconds than milliseconds. So, this patch makes the necessary
conversion.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: loopback: Add asynchronous bi-directional support
Bryan O'Donoghue [Mon, 7 Dec 2015 01:59:06 +0000 (01:59 +0000)]
greybus: loopback: Add asynchronous bi-directional support

A particular ask from the firmware people for some time now has been the
ability to drive multiple outstanding bi-directional operations from
loopback to loopback Interfaces. This patch implments that change.

The approach taken is to make a call to gb_operation_send() and have
loopback capture the completion callback itself, with a parallel timer to
timeout completion callbacks that take too long. The calling thread will
issue each gb_operation_send() as fast as it can within the constraints of
thread-safety.

In order to support this addition the following new sysfs entries are
created on a per-connection basis.

- async
  Zero indicates loopback should use the traditional synchronous model
  i.e. gb_operation_request_send_sync().

  Non-zero indicates loopback should use the new asynchronous model i.e.
  gb_operation_send()

- requests_completed
  This value indicates the number of requests successfully completed.

- requests_timedout
  This value indicates the number of requests which timed out.

- timeout
  The number of microseconds to give an individual asynchronous request
  before timing that request out.

- timeout_min
  Read-only attribute informs user-space of the minimum allowed timeout.

- timeout_max
  Read-only attribute informs user-space of the maximum allowed timeout.

Note requests_completed + requests_timedout should always equal
iteration_max, once iteration_count == iteration_max. Also, at this time we
support either synchronous or asynchronous operations in one set of
transactions.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: loopback: Convert cross-thread mutex to spinlock while relaxing connect...
Bryan O'Donoghue [Mon, 7 Dec 2015 01:59:05 +0000 (01:59 +0000)]
greybus: loopback: Convert cross-thread mutex to spinlock while relaxing connect locks

This patch converts the cross-thread mutex used to synchronize threads with
respect to each other to a spinlock. This is done to enable taking of locks
in the following patches while in atomic context. A small re-order of
locking in connection setup/tear-down is done to minimize the amount of
time spent in spinlock_irqsave().

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: Prefix hexadecimal values with 0x while printing them
Viresh Kumar [Fri, 4 Dec 2015 16:00:10 +0000 (21:30 +0530)]
greybus: Prefix hexadecimal values with 0x while printing them

To clearly specify the base for printed values, prefix hexadecimal
values with 0x.

Suggested-by: Johan Hovold <johan@hovoldconsulting.com>
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>
8 years agogreybus: don't use %h and %hh for printing short and char variables
Viresh Kumar [Fri, 4 Dec 2015 16:00:09 +0000 (21:30 +0530)]
greybus: don't use %h and %hh for printing short and char variables

Because the width of our fields is already known, we can use %0Nx (for
hex) to print N bytes and %u (for unsigned decimal), instead of using %h
and %hh, which isn't that readable.

This patch makes following changes:
- s/%hx/%04x
- s/%04hx/%04x
- s/%hhx/%02x
- s/%02hhx/%02x
- s/%hhu/%u
- s/%hu/%u
- s/%x/%02x for u8 value (only at a single place)

Suggested-by: Johan Hovold <johan@hovoldconsulting.com>
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>
8 years agogreybus: svc: Use -EIO instead of -EINVAL for unipro errors
Viresh Kumar [Fri, 4 Dec 2015 16:00:08 +0000 (21:30 +0530)]
greybus: svc: Use -EIO instead of -EINVAL for unipro errors

-EIO fits better here, rather than -EINVAL as the arguments to the
routine itself are valid, just that we failed while doing unipro
transfers.

Suggested-by: Johan Hovold <johan@hovoldconsulting.com>
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>
8 years agogreybus: Documentation: Arrange entries in alphabetical order
Viresh Kumar [Fri, 4 Dec 2015 16:00:07 +0000 (21:30 +0530)]
greybus: Documentation: Arrange entries in alphabetical order

The order of entries in sysfs-bus-greybus file doesn't match the order
files/directories in sysfs on a real board. More specifically, N-svc
comes at last and ap_interface_id comes before endo_id within the svc.

Fix that.

Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: core: add bundle id to bundle uevents
Johan Hovold [Fri, 4 Dec 2015 09:44:24 +0000 (10:44 +0100)]
greybus: core: add bundle id to bundle uevents

Add the bundle id to bundle uevents.

This is needed to identify bundles that are being removed (e.g. at
hot-unplug).

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: core: add interface id to interface and bundle uevents
Johan Hovold [Thu, 3 Dec 2015 18:18:04 +0000 (19:18 +0100)]
greybus: core: add interface id to interface and bundle uevents

Add the interface id to interface and bundle uevents.

This is needed to identify interfaces that are being removed (e.g. at
hot-unplug).

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: core: add bus id to uevents
Johan Hovold [Thu, 3 Dec 2015 18:18:02 +0000 (19:18 +0100)]
greybus: core: add bus id to uevents

Add the bus id to all greybus uevents.

This is needed to identify devices that are being removed (e.g. at
hot-unplug).

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: flush workqueue at connection exit
Johan Hovold [Thu, 3 Dec 2015 16:29:00 +0000 (17:29 +0100)]
greybus: svc: flush workqueue at connection exit

Make sure to flush the workqueue from hotplug and hotunplug events when
the svc connection is tore down.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: loopback: drop bus aggregate calculation
Bryan O'Donoghue [Thu, 3 Dec 2015 17:29:41 +0000 (17:29 +0000)]
greybus: loopback: drop bus aggregate calculation

At some point we had a statement in a Jira work item to pull out 'bus
level' data from greybus and to have messages to different interfaces be
synchronized with respect to each other. Synchronizing threads with respect
to each other is slow and it turns out we can get the same 'bus level'
stastics by making the user-space test application smarter.

That's great news for the in-kernel code since it means we can cut out a
whole lot of code to-do with calculating 'bus level' aggregate data and we
can stop forcing threads to hit a rendezvous before sending out another
loopback operation.

So this patch drops bus level aggregates in favour of doing that in
user-space. It subtracts a lot of code and cycles that in practice nobody
cares about anyway.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: loopback: Drop unused timeval variables
Bryan O'Donoghue [Thu, 3 Dec 2015 17:29:40 +0000 (17:29 +0000)]
greybus: loopback: Drop unused timeval variables

start and end aren't used and should be dropped.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: loopback: Move latency_ts initialization to transfer routine
Bryan O'Donoghue [Thu, 3 Dec 2015 17:29:39 +0000 (17:29 +0000)]
greybus: loopback: Move latency_ts initialization to transfer routine

Initializing the bridge specific latency variables is only relevant to the
transfer operation, so make it loopback-transfer specific.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: loopback: Relax locking during loopback operations
Bryan O'Donoghue [Thu, 3 Dec 2015 17:29:38 +0000 (17:29 +0000)]
greybus: loopback: Relax locking during loopback operations

Currently a per-connection mutex is held during calls to
gb_operation_send_sync. It is not necessary to hold this lock and later
patches supporting multiple-outstanding bi-directional operations need to
take the per-connection lock and the gb_dev level lock. Since gb_dev must
always be taken before per-connection locks, it is both desirable and safe
to drop the lock now.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: lights: default value for v4l2 flash controls
Rui Miguel Silva [Thu, 3 Dec 2015 16:07:49 +0000 (16:07 +0000)]
greybus: lights: default value for v4l2 flash controls

V4l2 flash will return erro ERANGE if val(which is the default value)
is not defined. Just set it to the max value reported by the module.

Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: lights: avoid channel torch double free
Rui Miguel Silva [Thu, 3 Dec 2015 16:07:48 +0000 (16:07 +0000)]
greybus: lights: avoid channel torch double free

When attaching torch to a flash we release the channel torch resources,
but afterwards we do it again when releasing all the channels.
Just free all the resource at channel release.

Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: lights: add v4l2 flash operations
Rui Miguel Silva [Thu, 3 Dec 2015 16:07:47 +0000 (16:07 +0000)]
greybus: lights: add v4l2 flash operations

We do not implement any of the v4l2 flash operations, as the default
ones are ok for now, however the init needs anything define, if not it
will return an error. So, just define it and have an error free v4l2
flash init.

Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: Documentation: Fix N-svc directory name
Viresh Kumar [Thu, 3 Dec 2015 08:24:57 +0000 (13:54 +0530)]
greybus: Documentation: Fix N-svc directory name

endo_id was present in place of the directory N-svc, fix it.

Fixes: 4f7b1833e78f ("Documentation/sysfs-bus-greybus: update the bus ABI documentation")
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>
8 years agogreybus: spi: fix transfers bigger than greybus payload
Rui Miguel Silva [Wed, 2 Dec 2015 11:12:29 +0000 (11:12 +0000)]
greybus: spi: fix transfers bigger than greybus payload

Add helper functions calculate the tx and rx size possible that fit a
greybus payload size and change the operation creation to adjust to that.

Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: build: remove Android makefile
Michael Scott [Thu, 3 Dec 2015 20:32:30 +0000 (12:32 -0800)]
greybus: build: remove Android makefile

There is no standard way of building a kernel from source in
Android.  Each device/SoC can (and do) implement it in their own way.
To that end, let's remove this makefile and let each device define
how they want to build the modules.

Signed-off-by: Michael Scott <michael.scott@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: spi: add master and device config operations
Rui Miguel Silva [Wed, 2 Dec 2015 11:12:28 +0000 (11:12 +0000)]
greybus: spi: add master and device config operations

Add master and device config operations, one is to merge all the master
operations and the device config will allow to fetch and add devices for
each chip select.

Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: spi: add rdwr field to transfer descriptor
Rui Miguel Silva [Wed, 2 Dec 2015 11:12:27 +0000 (11:12 +0000)]
greybus: spi: add rdwr field to transfer descriptor

Add read and/or write field to transfer descriptor to make it possible
to identify the type of transfer.

Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: fix racy hotplug handling
Johan Hovold [Wed, 2 Dec 2015 17:23:31 +0000 (18:23 +0100)]
greybus: svc: fix racy hotplug handling

Fix racy hotplug handling by serialising all processing of hot-plug and
unplug requests using a single-threaded dedicated workqueue.

This fixes a reported crash during enumeration when processing multiple
events.

The current svc implementation does not handle concurrency at all (e.g.
no interface list lock or refcounting) so we need to use the big hammer
for now.

Note that we will eventually want to process events for different
interfaces in parallel, but that we'd still need a workqueue in order
not to starve other svc requests (e.g. for timesync).

Reported-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: defer processing of hot-unplug requests
Johan Hovold [Wed, 2 Dec 2015 17:23:30 +0000 (18:23 +0100)]
greybus: svc: defer processing of hot-unplug requests

Defer processing also of hot-unplug events.

This is a step towards serialising hot-plug and unplug event processing.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: generalise deferred request handling
Johan Hovold [Wed, 2 Dec 2015 17:23:29 +0000 (18:23 +0100)]
greybus: svc: generalise deferred request handling

Clean up and generalise deferred request handling by simply storing a
reference-counted pointer to the operation itself in the work context.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: clean up interface-remove helper
Johan Hovold [Wed, 2 Dec 2015 17:23:28 +0000 (18:23 +0100)]
greybus: svc: clean up interface-remove helper

Pass the svc rather than its connection to the interface remove helper.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: clean up deferred hotplug handler
Johan Hovold [Wed, 2 Dec 2015 17:23:27 +0000 (18:23 +0100)]
greybus: svc: clean up deferred hotplug handler

Rename the hotplug request message, and clarify that the message size
has already been verified by the primary handler.

Also add a debug message that includes the interface id.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: clean up hotplug/unplug request handlers
Johan Hovold [Wed, 2 Dec 2015 17:23:26 +0000 (18:23 +0100)]
greybus: svc: clean up hotplug/unplug request handlers

Clean up hotplug/unplug request handlers somewhat.

Also add a debug message to both handlers that includes the interface id.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: firmware: Fetch es2 VID/PID to distinguish module vendors
Viresh Kumar [Thu, 26 Nov 2015 10:03:46 +0000 (15:33 +0530)]
greybus: firmware: Fetch es2 VID/PID to distinguish module vendors

The es2 chip doesn't have VID/PID programmed into the hardware and we
need to hack that up to distinguish different modules and their firmware
packages.

This fetches VID/PID (over firmware protocol) for es2 chip only, when
VID/PID already sent during hotplug are 0.

Since only the bootrom contains a firmware protocol cport, this only
affects bootrom's working and not nuttx.

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>
8 years agogreybus: Documentation/sysfs: rename interface and bundle attributes
Johan Hovold [Wed, 25 Nov 2015 14:06:41 +0000 (15:06 +0100)]
greybus: Documentation/sysfs: rename interface and bundle attributes

Update the example sysfs-tree layout under Documentation due renamed
interface and bundle attributes.

The interface and bundle "id" attributes have been renamed
"interface_id" and "bundle_id" respectively to make them self
describing. For consistency reasons the bundle class attribute is
renamed as "bundle_class".

.
├── greybus1
│   ├── 1-2
│   │   ├── 1-2.1
│   │   │   ├── bundle_class
│   │   │   ├── bundle_id
│   │   │   └── state
│   │   ├── 1-2.2
│   │   │   ├── bundle_class
│   │   │   ├── bundle_id
│   │   │   └── state
│   │   ├── interface_id
│   │   ├── product_id
│   │   ├── unique_id
│   │   └── vendor_id
│   ├── 1-4
│   │   ├── 1-4.2
│   │   │   ├── bundle_class
│   │   │   ├── bundle_id
│   │   │   ├── gpbridge0
│   │   │   │   ├── gpio
│   │   │   │   │   └── gpiochip490
│   │   │   │   └── i2c-4
│   │   │   └── state
│   │   ├── interface_id
│   │   ├── product_id
│   │   ├── unique_id
│   │   └── vendor_id
│   └── 1-svc
│       ├── ap_intf_id
│       ├── eject
│       ├── endo_id
│       └── unique_id
└── greybus2
    ├── 2-3
    │   ├── 2-3.1
    │   │   ├── bundle_class
    │   │   ├── bundle_id
    │   │   └── state
    │   ├── interface_id
    │   ├── product_id
    │   ├── unique_id
    │   └── vendor_id
    └── 2-svc
        ├── ap_intf_id
        ├── eject
        ├── endo_id
        └── unique_id

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: interface: drop the control bundle
Johan Hovold [Wed, 25 Nov 2015 14:59:26 +0000 (15:59 +0100)]
greybus: interface: drop the control bundle

Drop the control bundle and ignore control descriptors when parsing
manifests.

Every interface has a control connection with a well defined remote
CPort 0 and there's no longer any need to create a bundle for it.

As the control connection is setup and enabled before parsing the
manifest, ignore any legacy descriptors for control cports and bundles
in a manifest.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: manifest: refactor cport-descriptor release
Johan Hovold [Wed, 25 Nov 2015 14:59:25 +0000 (15:59 +0100)]
greybus: manifest: refactor cport-descriptor release

Add helper function to release cport-descriptors with a given bundle id.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: unconditionally enable connections
Johan Hovold [Wed, 25 Nov 2015 14:59:24 +0000 (15:59 +0100)]
greybus: connection: unconditionally enable connections

Remove conditional enabling of connections when binding protocols that
served no purpose as a connection either has no bundle or it has an
interface with a valid device id.

Also remove the now unused GB_PROTOCOL_NO_BUNDLE protocol flag.

This is an intermediate step in moving the protocol binding to
connection_init, but is also needed as the control bundle is going away.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: control: do not assume a control bundle
Johan Hovold [Wed, 25 Nov 2015 14:59:23 +0000 (15:59 +0100)]
greybus: control: do not assume a control bundle

The control bundle is going away so update the code.

Also remove defensive WARN_ON which would not just warn if our
implementation is broken, but also leak further memory unnecessarily.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: handle bundle-less connections in svc helpers
Johan Hovold [Wed, 25 Nov 2015 14:59:22 +0000 (15:59 +0100)]
greybus: connection: handle bundle-less connections in svc helpers

The svc connection helper functions should not assume that all dynamic
connections will have a bundle.

This is needed as the control bundle is going away.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: kill the endo
Johan Hovold [Wed, 25 Nov 2015 14:59:21 +0000 (15:59 +0100)]
greybus: kill the endo

Remove the now unused endo and module code.

Note that the never-implemented serial and version attributes of the
endo can be implemented as svc attributes if needed.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: loopback: remove endo reference
Johan Hovold [Wed, 25 Nov 2015 14:59:20 +0000 (15:59 +0100)]
greybus: loopback: remove endo reference

Replace reference to "endo0" and generate the raw-latency filename based
on the host-device bus id instead.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: clean up log messages
Johan Hovold [Wed, 25 Nov 2015 14:59:19 +0000 (15:59 +0100)]
greybus: svc: clean up log messages

Use dev_err and friends with the svc device for messages.
Clean up error messages.
Demote a few warnings to warning level.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: hd: fix svc-connection handling
Johan Hovold [Wed, 25 Nov 2015 14:59:18 +0000 (15:59 +0100)]
greybus: hd: fix svc-connection handling

Create the svc connection when registering the host-device and
remove the current svc connection hacks that "upgraded" the svc
connection once the endo id and ap interface id was known.

Note that the old implementation was partly based on a misunderstanding
as it was the remote interface id, rather than the local AP interface id,
that used to define a connection (but we also needed the endo_id).

The remote interface is no longer needed as static connections, such as
the svc connection, are now simply defined by the host-device and host
cport id.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: protocol: handle static connections
Johan Hovold [Wed, 25 Nov 2015 14:59:17 +0000 (15:59 +0100)]
greybus: protocol: handle static connections

Use host-device device and connection name for log messages, as not all
connections have a bundle.

Note that the "initial" svc connection has never had a bundle.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: operation: handle static connections
Johan Hovold [Wed, 25 Nov 2015 14:59:16 +0000 (15:59 +0100)]
greybus: operation: handle static connections

Use the host-device device and connection name for error messages, as
the operation code can not assume that a connection has a bundle.

Note that the "initial" svc connection has never had a bundle.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: handle static connections
Johan Hovold [Wed, 25 Nov 2015 14:59:15 +0000 (15:59 +0100)]
greybus: connection: handle static connections

Use host-device device and connection name for log messages rather than
assume that all connections have a bundle (e.g. not true for static
connections).

Note that the "initial" svc connection has never had a bundle.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: add name field
Johan Hovold [Wed, 25 Nov 2015 14:59:14 +0000 (15:59 +0100)]
greybus: connection: add name field

Add a name field to connections that can be used in log messages.

A connection always belongs to a host-device (bus) and can be uniquely
identified by its host-device cport id, but include remote interface and
cport id nonetheless on the following format:

<hd_cport_id>/<intf_id>:<cport_id>

The remote interface and cport id will be zero for static connections.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: kill GB_PROTOCOL_SKIP_SVC_CONNECTION
Johan Hovold [Wed, 25 Nov 2015 14:59:13 +0000 (15:59 +0100)]
greybus: connection: kill GB_PROTOCOL_SKIP_SVC_CONNECTION

Add helper to determine whether a connection is static, and remove the
protocol flag GB_PROTOCOL_SKIP_SVC_CONNECTION.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: fix potential null-deref on create
Johan Hovold [Wed, 25 Nov 2015 14:59:12 +0000 (15:59 +0100)]
greybus: connection: fix potential null-deref on create

Make sure that the interface lookup helper can handle static,
bundle-less connections without oopsing when creating further
connections.

Note that the initial svc-connection has always been bundle-less, but
did not trigger an oops as a bundle was created for it before further
connections were created.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: clean up connection-creation interface
Johan Hovold [Wed, 25 Nov 2015 14:59:11 +0000 (15:59 +0100)]
greybus: connection: clean up connection-creation interface

Clean up the connection-creation interface by clearly separating our two
types of connections: static and dynamic.

Add two convenience functions for creating static and dynamic connections.

A static connection is a pre-setup connection that is defined by a host
device and a host-device cport id. Specifically, the remote interface or
cport id need not be known. The SVC connection is a static connection.

A dynamic connection is defined by a host device and a remote interface
and cport id. This is our normal connections where the host-device cport
is (generally) allocated dynamically.

Note that the new generic interface is marked static, but can be
exported later to allow dynamic connections to be created also from
fixed host-device cports (e.g. for CSI).

Also note that a connection of either type is uniquely identified by its
host-device and host-device cport id once created.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: interface: kill gb_create_bundle_connection
Johan Hovold [Wed, 25 Nov 2015 14:59:10 +0000 (15:59 +0100)]
greybus: interface: kill gb_create_bundle_connection

Kill gb_create_bundle_connection, which was only used to create the
control bundle and connection, and replace it with a specialised static
helper.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: move endo id and ap interface id to svc
Johan Hovold [Wed, 25 Nov 2015 14:59:09 +0000 (15:59 +0100)]
greybus: svc: move endo id and ap interface id to svc

Move endo_id and AP interface id to the svc device.

The endo abstraction is about to be removed, and these attributes are
arguable attributes of the svc anyway.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: register svc device at hello
Johan Hovold [Wed, 25 Nov 2015 14:59:08 +0000 (15:59 +0100)]
greybus: svc: register svc device at hello

Register the svc device upon reception of the HELLO request.

The SVC HELLO request contains the endo id and AP interface id, which
will be exposed from the svc device rather than the endo.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: clean up svc initialisation
Johan Hovold [Wed, 25 Nov 2015 14:59:07 +0000 (15:59 +0100)]
greybus: svc: clean up svc initialisation

Make sure to initialise the svc device fully before adding it to the
host device.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: make svc a device
Johan Hovold [Wed, 25 Nov 2015 14:59:06 +0000 (15:59 +0100)]
greybus: svc: make svc a device

Make the svc object a child device of the host-device device and use
driver core to manage its lifetime.

The svc device name is "<bus_id>-svc", where bus_id is the dynamically
assigned id of the host device.

Note that there is exactly one svc-device per host device (bus).

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: bundle: rename bundle devices
Johan Hovold [Wed, 25 Nov 2015 14:59:05 +0000 (15:59 +0100)]
greybus: bundle: rename bundle devices

Rename bundle devices so that the new device names become
"<bus_id>-<intf_id>.<bundle_id>", where bus_id is the dynamically
allocated host-device bus id and intf_id the svc-allocated interface
id.

Using a period (.) rather than a colon (:) makes dev-messages easier to
read as as those already add a colon after the device name, for example:

greybus 1-4.15: failed to connect cport: -22

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: interface: make interfaces children of host devices
Johan Hovold [Wed, 25 Nov 2015 14:59:04 +0000 (15:59 +0100)]
greybus: interface: make interfaces children of host devices

Make interfaces child devices of host devices.

The new interface device name is "<bus_id>-<intf_id>", where bus_id is
the dynamically allocated bus id for the host device and intf_id is the
svc-allocated interface id.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: drop parent parameter from connection create
Johan Hovold [Wed, 25 Nov 2015 14:59:03 +0000 (15:59 +0100)]
greybus: connection: drop parent parameter from connection create

The parent parameter was only used for debug messages and to name the
connection workqueue. Use the host-device device for this instead.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: hd: make host device a device
Johan Hovold [Wed, 25 Nov 2015 14:59:02 +0000 (15:59 +0100)]
greybus: hd: make host device a device

Make the host device a proper device in the kernel device model.

Host devices will be our new greybus-bus root devices.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: bundle: add bundle_id attribute
Johan Hovold [Wed, 25 Nov 2015 14:59:01 +0000 (15:59 +0100)]
greybus: bundle: add bundle_id attribute

Add a bundle_id attribute for the interface-unique id of a bundle that
user space can use for matching.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: bundle: use hexadecimal notation for class attribute
Johan Hovold [Wed, 25 Nov 2015 14:59:00 +0000 (15:59 +0100)]
greybus: bundle: use hexadecimal notation for class attribute

Use hexadecimal notation for class-attribute value.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: bundle: rename class attribute
Johan Hovold [Wed, 25 Nov 2015 14:58:59 +0000 (15:58 +0100)]
greybus: bundle: rename class attribute

Rename the bundle class-attribute "bundle_class" for consistency reasons
and to make it self documenting.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: interface: add interface_id attribute
Johan Hovold [Wed, 25 Nov 2015 14:58:58 +0000 (15:58 +0100)]
greybus: interface: add interface_id attribute

Add interface_id attribute that user space needs to identify an
interface.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: interface: remove device_id attribute
Johan Hovold [Wed, 25 Nov 2015 14:58:57 +0000 (15:58 +0100)]
greybus: interface: remove device_id attribute

The interface device_id attribute is an implementation detail that does
not need to be exported to user space.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: interface: rename vendor and product attributes
Johan Hovold [Wed, 25 Nov 2015 14:58:56 +0000 (15:58 +0100)]
greybus: interface: rename vendor and product attributes

Rename vendor and product attributes vendor_id and product_id.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: Documentation/sysfs-bus-greybus: update the bus ABI documentation
Johan Hovold [Wed, 25 Nov 2015 14:58:55 +0000 (15:58 +0100)]
greybus: Documentation/sysfs-bus-greybus: update the bus ABI documentation

Update the ABI documentation to match the new device model.

Note that the SVC unique_id and version attributes are not yet
implemented.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: Documentation/sysfs: update the example sysfs tree
Johan Hovold [Sun, 22 Nov 2015 09:24:08 +0000 (10:24 +0100)]
greybus: Documentation/sysfs: update the example sysfs tree

Update the example sysfs-tree layout under Documentation.

The new layout reflects changes to the kernel device-model that are
needed to eventually be able to handle multiple AP-bridges.

The example tree has two AP-bridges, each with its own view of the
UniPro network, the bus.

.
├── greybus1
│   ├── 1-2
│   │   ├── 1-2.1
│   │   │   ├── class
│   │   │   ├── id
│   │   │   └── state
│   │   ├── 1-2.2
│   │   │   ├── class
│   │   │   ├── id
│   │   │   └── state
│   │   ├── id
│   │   ├── product_id
│   │   ├── unique_id
│   │   └── vendor_id
│   ├── 1-4
│   │   ├── 1-4.2
│   │   │   ├── class
│   │   │   ├── gpbridge0
│   │   │   │   ├── gpio
│   │   │   │   │   └── gpiochip490
│   │   │   │   └── i2c-4
│   │   │   ├── id
│   │   │   └── state
│   │   ├── id
│   │   ├── product_id
│   │   ├── unique_id
│   │   └── vendor_id
│   └── 1-svc
│       ├── ap_intf_id
│       ├── eject
│       ├── endo_id
│       └── unique_id
└── greybus2
    ├── 2-3
    │   ├── 2-3.1
    │   │   ├── class
    │   │   ├── id
    │   │   └── state
    │   ├── id
    │   ├── product_id
    │   ├── unique_id
    │   └── vendor_id
    └── 2-svc
        ├── ap_intf_id
        ├── eject
        ├── endo_id
        └── unique_id

Every bus has exactly one svc device (1-svc and 2-svc). For our system,
the svc device of each bus will be a representation of the same
network-unique SVC device (e.g. endo_id and unique_id will be
identical).

The first bus has two registered interfaces (1-2 and 1-4), while the
second bus has a single interface (2-3). Note that the interface ids (2,
4, and 3) are necessarily unique as these are interfaces on the same
network.

Interface 1-2 has two bundles (1-2.1 and 1-2.2) and interface 1-4 has
a single bundle (1-4.2). The bundle ids are interface-unique and reflect
the ids found in each manifest.

In the example, bundle 1-4.2 has a gbbridge-device, which is the parent
device for a gpiochip device and an i2c bus.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: manifest: simplify descriptor address calculation
Sachin Pandhare [Tue, 24 Nov 2015 02:29:10 +0000 (07:59 +0530)]
greybus: manifest: simplify descriptor address calculation

This patch doesn't change any functionality. It just improves the
readability of the code.
Current code to get 'descriptors' pointer looks as if we are forcing the
pointer type change. To simplify the address calculations, use
'descriptors' member directly from greybus_manifest structure.

Signed-off-by: Sachin Pandhare <sachin.pandhare@linaro.org>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: hd: fix cport-count check
Johan Hovold [Sun, 22 Nov 2015 09:12:58 +0000 (10:12 +0100)]
greybus: hd: fix cport-count check

Fix off-by-one error when checking the number of cports a host-device
supports.

The CPORT_ID_MAX is the largest valid cport id so the maximum number of
cports a host-device can use is CPORT_ID_MAX + 1.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: connection: fix cport-id range
Johan Hovold [Sat, 21 Nov 2015 09:51:58 +0000 (10:51 +0100)]
greybus: connection: fix cport-id range

Fix cport-id allocation that failed to include the highest port id in the
available cport-id range.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: Audio: Add skeleton code for GB virtual codec driver
Vaibhav Agarwal [Mon, 23 Nov 2015 10:27:45 +0000 (15:57 +0530)]
greybus: Audio: Add skeleton code for GB virtual codec driver

This patch adds gb-codec driver with static information for
DAPM widgets, controls & dapm_routes.

Including some changes in kernel code(machine driver):
- Able to register codec and glue it with existing sound card successfully.
- Able to view & modify mixer controls:
        (volume/mute[left/right][input/output])
- Able to view DAPM widgets registered via /debug interface.
- Able to establish DAPM path for playback.

Since, FE<->BE path not yet verified with default jetson build,
registering GB DAI as normal DAI link to verify GB virtual codec
specific DAPM path.

Signed-off-by: Vaibhav Agarwal <vaibhav.agarwal@linaro.org>
Reviewed-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: add class matching macro
Johan Hovold [Sat, 21 Nov 2015 09:52:04 +0000 (10:52 +0100)]
greybus: add class matching macro

Add matching macro for bundle class.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: move id-matching back to core
Johan Hovold [Sat, 21 Nov 2015 09:52:03 +0000 (10:52 +0100)]
greybus: move id-matching back to core

Move id-matching back to core and the bus code where it belongs.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: fix vendor and product matching
Johan Hovold [Sat, 21 Nov 2015 09:52:02 +0000 (10:52 +0100)]
greybus: fix vendor and product matching

Fix vendor and product matching by matching on the 32-bit Ara vendor and
product ids.

Remove the "fake" 16-bit vendor and product ids and export the Ara ids
using the "vendor" and "product" interface attributes instead.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: interface: remove unique id
Johan Hovold [Sat, 21 Nov 2015 09:52:01 +0000 (10:52 +0100)]
greybus: interface: remove unique id

Remove the unimplemented interface unique-id.

There will eventually be an interface-serial-number attribute provided,
but let's not export it or commit to a name for this attribute until we
need it.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: remove unique-id matching
Johan Hovold [Sat, 21 Nov 2015 09:52:00 +0000 (10:52 +0100)]
greybus: remove unique-id matching

Remove unique-id matching as it does not make much sense to have a
driver match a specific device serial number.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: fix bundle-id match macros
Johan Hovold [Sat, 21 Nov 2015 09:51:59 +0000 (10:51 +0100)]
greybus: fix bundle-id match macros

The matching flags were renamed over a year ago but the so far unused
id-macros were never updated.

Also rename the GREYBUS_ID_MATCH_DEVICE mask to use the common
GREYBUS_ID_MATCH-prefix.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: svc: fix missing version-request sanity checks
Johan Hovold [Thu, 19 Nov 2015 17:28:01 +0000 (18:28 +0100)]
greybus: svc: fix missing version-request sanity checks

Add missing sanity checks on version-request payload size.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: firmware: break long lines
Johan Hovold [Thu, 19 Nov 2015 17:28:00 +0000 (18:28 +0100)]
greybus: firmware: break long lines

Break lines longer than 80 cols, and clean up an error message while at
it.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: firmware: fix information leak
Johan Hovold [Thu, 19 Nov 2015 17:27:59 +0000 (18:27 +0100)]
greybus: firmware: fix information leak

Add missing sanity checks on get_firmware-request offset and size
parameters to fix potential information leaks.

This prevents remotely controlled information leaks as the requestor
currently controls both the 32-bit firmware-image offset and the amount
of data that is returned (up to host-device MTU).

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: loopback: allocate a response even for a 0-byte request
Bartosz Golaszewski [Thu, 19 Nov 2015 12:46:43 +0000 (13:46 +0100)]
greybus: loopback: allocate a response even for a 0-byte request

If payload length of a transfer packet is 0, no response is allocated.
Send a well-formed response even in that case.

Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: loopback: fix invalid response size
Alexandre Bailon [Thu, 19 Nov 2015 11:44:46 +0000 (12:44 +0100)]
greybus: loopback: fix invalid response size

The size of timestamps is not taken into account, which makes the
loopback driver in the firmware drop invalid packages.

Signed-off-by: Alexandre Bailon <abailon@baylibre.com>
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: manifest: fix bundle descriptor parse
Rui Miguel Silva [Mon, 16 Nov 2015 19:23:25 +0000 (19:23 +0000)]
greybus: manifest: fix bundle descriptor parse

The descriptor list is walked in two points, in the bundle parsing and
cport parsing, this can make the next descriptor pointer in bundle to be
already removed by the cport remove descriptor and become invalid.

So, just get the next bundle until there no more left.

Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
8 years agogreybus: audio: Allocate protocol and class values
Mark Greer [Fri, 13 Nov 2015 21:44:13 +0000 (14:44 -0700)]
greybus: audio: Allocate protocol and class values

Allocate protocol and class values for the Audio Device Class
Protocol.  Two values of each type are allocated: one for Audio
Management Connections and one for Audio Data Connections.

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