Add missing error handling when registering the vibrator protocol during
module init.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

Include function, protocol name and id when printing the version
response debug message.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

Issue a warning if we fail to look up a protocol when dropping a
reference.

This should never happen as we hold a reference to the protocol module
and protocols should only be deregistered at module unload.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

A protocol should be deregistered exactly once when the protocol module
is being unloaded. This means that protocol deregister will never be
called with active users as we take a module reference when looking up a
protocol.

Remove comment suggesting that we could one day forcefully stop a user
of a protocol, and issue a big warning if a protocol is deregistered
more than once or at some other time than during module unload (e.g.
with active users).
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

Clean up and improve error messages.

Demote a warning message to warning level.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

Drop FIXME about sending responses in OOM situations.

If we fail to allocate an operation for an incoming request, we have
bigger problems than to worry about sending a response.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

Replace a couple of pr_err with more informative dev_err and clean up
the messages somewhat.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

Replace the remaining pr_err with dev_err, and drop redundant function
prefixes.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

gb_connection_bind_protocol() returns proper error codes now and we
should destroy the connection on failures.

This change also fixes a NULL deref on hotplug when the control connection fails
to initialize.
Signed-off-by: Fabien Parent <fparent@baylibre.com>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

Toshiba UniPro IP requires to reset the CPort that has been used in a previous
connection. This commit implement a new control request in order to
reset CPorts on an APBridgeA.
Signed-off-by: Fabien Parent <fparent@baylibre.com>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

When translating capabilities from greybus to mmc values add some
parentheses to clarify operation precedence and avoid static analyses
warnings.

[sdio.c:81]: (style) Clarify calculation precedence for '&' and '?'
Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

It was missing the translation between the ocr vdd values of greybus to
mmc_core values. This would make the detection of range voltage fail.

mmc: host doesn't support card's voltages
Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

In kernel versions bellow 3.15, the mmc_card_is_removable helper
function has an extra check used for a suspend/resume hack. This made
the gd_sdio_process_event to behave badly handling the module card
insert event in that versions.

So, just test bit the flag that we need, instead of using the helper
function. This way will work in all kernel versions.
Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

Frequency maximum and minimum are needed to complete the configuration
of the controller. Add them to get_caps response operation.
Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

We need to skip setting E2EFC and other flags to the SVC connection
create request, for all cports, on an interface that need to boot over
unipro, i.e. interfaces required to download firmware.

This also adds a FIXME as we need to do it differently for ES3.
Tested-by: Eli Sennesh <esennesh@leaflabs.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off by: Eli Sennesh <esennesh@leaflabs.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

Currently, the greybus module .ko files are quite large and the
following error was observed during Android build for each greybus module:
strip: Unable to recognise the format of the input file `<module>`

Fix the strip build step by replacing the undefined KERNEL_TOOLCHAIN_PATH
variable with the GREYBUS_CC_PREFIX variable.  Also used as the
CROSS_COMPILER value for the module make.
Signed-off-by: Michael Scott <michael.scott@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

When using greybus build in a different Android setup, it was noted that
several portions of this makefile rely on Linaro specific build items.
Replace these with more generic build steps.

- ANDROID_64 is only defined by Linaro build tasks, use TARGET_ARCH to
  establish ARCH= parameter for greybus build
- KERNEL_TOOLS_PREFIX is only defined by Linaro build tasks.  AOSP has
  a near equivalent variable: TARGET_TOOLS_PREFIX
- build-greybus was dependant on subtask: android_kernel a task defined
  only by Linaro build tasks.  Replace with a generic dependancy to
  the kernel binary located in $OUT (INSTALLED_KERNEL_TARGET).
  End result is the same: kernel must be built before greybus modules
Signed-off-by: Michael Scott <michael.scott@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

The generic_handle_irq_desc api only have changed in 4.3.0, so check
against the correct version, if not will break builds for 4.2.x.

Fixes: e7895cfc476 ("gpio: handle api change in generic_handle_irq_desc()")
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>

generic_handle_irq_desc changed the api in the 4.2 kernel, so fix up the
gpio driver to handle this properly to keep it working.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

Remove legacy interface to "destroy" operations, which is now just a
wrapper for gb_operation_put.

The old interface name hides the fact that all operations are refcounted
and may live on even after having "destroyed" them.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

Reset the hcpriv field before returning the message to greybus core in
the OUT-URB completion callback.

This fixes a use-after-free bug when sending responses to incoming
requests as the final reference is then dropped when the message is
returned.
Reported-by: Michael Scott <michael.scott@linaro.org>
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

After downloading the firmware for the next boot stage, module's
firmware (for current boot stage) jumps into it and the new firmware and
sends hotplug request to SVC. On hotplug request from the SVC, the AP
first removes the existing interface.

At this time, there is no point sending disconnected event for the
firmware protocol, for the firmware used in previous stage, as the new
firmware wouldn't be aware about it.

Set flags for firmware protocol to skip control-disconnected operations.
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>

The spec says that it doesn't need it, so dropping 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>

The callers are ensuring that another interface doesn't exist with the
same interface id and so there is no need to check that from
gb_interface_create() anymore.
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>

The callers already have a valid interface pointer and there is no need
for gb_interface_remove() to find the interface again.
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>

As per the module's boot sequence diagram, the AP needs to read/clear
T_TstSrcIncrement attribute on hotplug (svc) events.

Implement that.

FIXME: This is module-hardware dependent and needs to be extended for
every type of module we want to support.

[ Based on work by Marti & Eli to clear the attribute with DME set]
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>

There are two cases where the AP may receive hotplug event for an
existing interface, without first getting a hot-unplug request for it.

- bootrom loading the firmware image and booting into that, which
  only generates a hotplug event. i.e. no hot-unplug event, as the
  module never went away.
- Or the firmware on the module crashed and sent hotplug request again
  to the SVC, which got propagated to AP.

Handle such cases by first removing the interface, with a clear print
message shown to the user. And then following the normal hotplug
sequence to add the interface.
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>

We already have a variable to access '&op->connection->dev' directly,
lets reuse it.
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Jean Pihet <jean.pihet@newoldbits.com>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

The datapoint we are using to return metrics across modules and cports
shouldn't have a module identifier in it i.e.

/sys/kernel/debug/gb_loopback/raw_latency_endo0
not
/sys/kernel/debug/gb_loopback/raw_latency_endo0:X

This patch removes the module_id used up to this point. Including module_id
actually ends up making life harder in user-space so dropping it.
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Reviewed-by: Patrick Titiano <ptitiano@baylibre.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

dev_name() will return the endo0 component of the string on it's own,
there's no need to include it in the snprintf() when construting the
debugfs name. This fixes 'endo0' appearing more than once in the debugfs
name - shamefully slipped through testing cb570c93783f
('greybus/loopback: use dev_name to populate sysfsname').
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

If a thread is masked out it should not consume CPU cycles during a test.
We set an arbitrary 100 millisecond sleep time for each masked out thread.
Reasonably blunt instrument to ensure threads with nothing to do don't end
up thrashing the acquisition/release of mutexes.
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Reviewed-by: Patrick Titiano <ptitiano@baylibre.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

Currently we have sysfs entries that are created when the first incoming
connection is created as sub-nodes of the module associated with that
connection e.g. /sys/bus/grebus/devices/endo0:X where X is the module
identifier associated with the new connection. This is conceptually
incorrect since the sysfs entries we create actually aren't bound to a
module. Depending on the order connections are brought up we can also have
a situation where /sys/bus/greybus/devices/endo0:X has high-level control
sysfs data-points but /sys/bus/greybus/devices/endo0:Y does not. Rather
than needlessly replicate data-points across each endo0:X, endo0:Y, endo0:Z
sysfs directories it is more sensible to locate the entries in
/sys/bus/greybus/devices/endo0.
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Reviewed-by: Patrick Titiano <ptitiano@baylibre.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

This patch hooks tracepoints for the handoff point to/from hardware. With
these tracepoints in place we can view the time between gb_message_send and
usb_submit_urb and similarly we can view the time between cport_in_callback
and gb_message_recv_response/gb_message_recv_request

- trace_gb_host_device_send
- trace_gb_host_device_recv

It provides standard tracepoints at

/sys/kernel/debug/tracing/events/greybus/gb_host_device_send
/sys/kernel/debug/tracing/events/greybus/gb_host_device_recv

Giving outputs like

gb_host_device_recv: greybus:2-1 if_id=0000 l=10
gb_host_device_send: greybus:2-1 if_id=0000 l=10
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

This patch adds new tracepoint declarations to greybus_trace.h to allow for
capture of greybus host device tx and rx events. These two tracepoints
allow an observer to see the point where the hardware interface driver
performs the relevant read or write to receive or write the data it's been
given from the higher layer greybus driver.

The following two new tracepoints are declared:
- trace_gb_host_device_send
- trace_gb_host_device_recv
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

This patch drops tracking of internal latencies, it's possible to derive
these times via kernel tracepoints and some user-space scripting. Since
there's no other use of the internal timestamp than the loopback driver we
remove the connection.c, connection.h, es1.c, es2.c and loopback.c
inter-dependency in one go.
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

The Qualcomm kernel builds with -Werror so the existing es2.c driver
breaks the build due to unused static functions.  As we are still
hashing out exactly how to implement this logic at the moment, just
comment out the functions to make the build be clean, no logic changes
happen here at all.
Reported-by: Michael Scott <michael.scott@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

This patch hooks tracepoints for greybus messages

- trace_gb_message_send
- trace_gb_message_recv_request
- trace_gb_message_recv_response
- trace_gb_message_cancel_outgoing
- trace_gb_message_cancel_incoming

It provides standard tracepoints at

/sys/kernel/debug/tracing/events/greybus/gb_message_send
/sys/kernel/debug/tracing/events/greybus/gb_message_recv_response
/sys/kernel/debug/tracing/events/greybus/gb_message_recv_request
/sys/kernel/debug/tracing/events/greybus/gb_message_cancel_outgoing
/sys/kernel/debug/tracing/events/greybus/gb_message_cancel_incoming

Giving outputs like

gb_message_recv_request: greybus:1-1.1:0 op=0001 if_id=0000 hd_id=0000 l=2
gb_message_send: greybus:1-1.1:0 op=0001 if_id=0000 hd_id=0000 l=2

Similarly perf events can be viewed with standard perf tools e.g.

root@beaglebone:~# perf list 'greybus:*'
  greybus:gb_message_send                            [Tracepoint event]
  greybus:gb_message_recv_request                    [Tracepoint event]
  greybus:gb_message_recv_response                   [Tracepoint event]
  greybus:gb_message_cancel_outgoing                 [Tracepoint event]
  greybus:gb_message_cancel_incoming                 [Tracepoint event]
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

This patch adds greybus_trace.h with the following trace definitions

- trace_gb_message_send
- trace_gb_message_recv_request
- trace_gb_message_recv_response
- trace_gb_message_cancel_incoming
- trace_gb_message_cancel_outgoing
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

The file names here weren't in sync with what we have today and the
updates give a better picture of the same.
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

These host-driver callbacks were intended to allow host drivers to
prepare a cport, something which can now be handled by the cport
enable/disable callbacks instead.

The current create/destroy are somewhat confusingly named as they were
not supposed to create or destroy connections. They were however called
from the unrelated helper functions that do create and destroy SVC
connections.

Furthermore, no errors were returned should the create callback fail,
which should have caused the connection initialisation to fail.

Remove these unused callbacks for now, and let us use the cport
enable/disable callbacks that should be able handle all host cport
initialisation (possibly after also adding an interface to provide
information for endpoint-cport mapping).
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>