greybus: power_supply: free supplies at release

After freeing each individual power_supply, free the top controller, if
not it will leak memory at each module insert/remove.

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

greybus: power_supply: fix unregister on error path

If setup fail the release for each supply needs to know the status at
unregister time. So, add the field to the structure, update it at setup
time and use it at release.

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

greybus: power_supply: do not release failed supply alloc

If allocation of memory for each supply fail, we should get out of
release any individual supply.

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

greybus: power_supply: fix lock imbalance in init error path

Make sure to release the supplies_lock before returning on errors in
gb_power_supplies_setup().

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

greybus: arche-platform: Add wake/detect support along with handshaking with AP

Add wake_detect support to arche-platform driver which is responsible
for SVC control. This patch also adds code for handshaking between
AP <=> SVC. The sequence is,

1.   AP boots
     To keep compatibility between DB3 and EVT platform, SVC will be
     help in reset and AP driver would release it at appropriate time.
 wake/detect pin (WD8A) = Low
 reset (SVC/APB1/APB2) = Asserted (as per polarity)
2.   AP Driver gets inserted
  2.1. AP will deassert reset to SVC (following power on sequence)
  2.2. SVC allows 360 milliseconds to elapse after switch boots to work
       around bug described in ENG-330.
  2.3. AP asserts wake/detect pin (WD8A = HIGH)
3.   SVC detects assertion of wake/detect pin, and sends "wake out" signal to AP
4.   AP receives "wake out" signal, takes AP Bridges through their power
     on reset sequence as defined in the bridge ASIC reference manuals
5.   AP takes USB3613 through its power on reset sequence
6.   AP should enumerates AP Bridges

Note: ISR has been deliberately removed (not merged) as we are still
not sure how it will be used, in runtime usage context. Driver as such
doesn't do anything for runtime assert-n-deassert of reset to SVC/APB's,
it just simply offloads it to user by exporting required gpio's.
The exported gpio's are required for FW flashing from user space.
When it comes to usersace manipulated control sequence, user has to manage.

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

greybus: arche-apb-ctrl: deassert reset at the end of probe

Now the complete handshaking between AP<=>SVC would be
offloaded to parent driver (arche-platform) who is
responsible for SVC control, so the apb-ctrl driver can
just simply bring APB's out of reset during probe itself.

Along with deasserting reset, this patch renames the local fn
to exclusively use it for reset purpose.

Note that, driver is exporting reset gpio to
user, so user can still flash FW from prompt.

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

greybus: arche-apb-ctrl: Remove wake_detect gpio and related code

With TIME_SYNC functionality, assignment of wake/detect pin
will change,
  WD_8A => to bring APB's out of reset
  WD_8B => TIME_SYNC operation

So in order to support this, we can no longer keep wake_detect
gpio in apb-ctrl driver. So remove it.
In the subsequent patches, wake_detect support will be added
to parent SVC driver (arche-platform) who is responsible for SVC control.

Note that, this patch also removes ISR related code.
The APB state still is maintained, for future use.

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

greybus: arche: Remove unwanted headers and rearrange others

This removes few unwanted headers related to irq, interrupt, regulator,
spinlock, etc.

Also arrange the rest in alphabetical order to make it more readable.

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

greybus: arche-apb: Drop unnecessary checks

wake_detect_gpio and resetn_gpio are guaranteed to be valid in
apb_ctrl_cleanup() and irq-handler, no need to check for their validity.

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

greybus: arche-apb: Do cleanup within apb_ctrl_init_seq() for error cases

Relying on apb_ctrl_cleanup() to do the cleanup for errors that occurred
within apb_ctrl_init_seq() isn't a very clean idea. Handle that
separately within apb_ctrl_init_seq().

This will clean apb_ctrl_cleanup() in later patches.

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

greybus: arche-apb: Don't use gpio after failing to request it

If devm_gpio_request() returns an error, we shouldn't try to set the
direction of the same gpio.

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

greybus: arche-apb: Properly use dev_err/info/warn

Use dev_err for errors after which we need to abort the currently
running routine and dev_warn for resource allocation failure, with which
we can continue to work.

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

greybus: arche-apb: NULL is a valid regulator

Since NULL could in theory be a valid regulator we ought to check for
IS_ERR() rather than for NULL. In practice this is unlikely to be an
issue but it's better for neatness.

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

greybus: arche-apb: devm_regulator_get() doesn't return NULL

And so we don't need to check for it.

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

greybus: arche-apb: Replace gpio_is_valid() with gpio < 0 checks

There can be no invalid values in the DTS. The actual pin numbers are
assigned by gpiolib when the gpio controller is registered.

And so a simple 'gpio < 0' is enough instead of gpio_is_valid() which
also checks for 'gpio < ARCH_NR_GPIOS'.

This will make the usage of of_get_named_gpio() similar with how it is
done in arche-platform driver.

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

greybus: arche-apb: platform data 'apb' is guaranteed to be valid

Platform data 'apb' is guaranteed to be valid in arche_apb_ctrl_remove()
and so no need to check it.

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

greybus: arche-apb: Spelling and whitespace fixes

- s/premits/permits
- Remove blank lines at the end of the file

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

greybus: arche-platform: propagate errors returned by gpiolib

Propagate errors returned by of_get_named_gpio() instead of sending
-ENODEV.

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

greybus: arche-platform: svc_reset_gpio can't be invalid in arche_platform_cleanup()

svc_reset_gpio is guaranteed to be valid in arche_platform_cleanup. Drop
the useless check.

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

greybus: arche-platform: Put APB in reset if of_platform_populate() fails

The current implementation around of_platform_populate() is not so
great. On error, we first print an error message, followed by a success
message and finally we return an error.

And over that we don't undo what we did initially.

This patch puts the APB back into reset and create a separate error
path to make things clear.

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

greybus: arche-platform: Export GPIOs after populating APBs

Populating APBs operation can potentially fail and it would be better if
we export the GPIOs towards then end of the routine, so that we don't
need to unexport them on error cases.

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

greybus: arche-platform: arche_pdata is guaranteed to be valid

arche_pdata is guaranteed to be valid in arche_platform_remove().

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

greybus: firmware/bootrom: debug output from bootrom protocol

The bootrom protocol issues no dynamic debugging messages when it functions
successfully.  Use dev_dbg() to fix that by issuing kernel logs when firmware
download works.

Signed-off-by: Eli Sennesh <esennesh@leaflabs.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: connection: fix version-request error handling

Use the host device and connection name when logging errors as the
version-request helper must not assume that all connection have a
bundle.

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

greybus: core: fix greybus device matching

The bus code should only match bundle devices for now, and must not
assume all greybus devices are bundles.

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

greybus: core: fix greybus driver registration

Add missing bus type to driver structure when registering a greybus
driver.

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

greybus: core: add drvdata accessors

Add greybus driver-data accessors.

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

greybus: camera: Add support for flags to stream_configure

Add support for the flags field of the stream configure request that was
recently added to the camera protocol and update the debugfs arguments
parsing accordingly. The stream configure response layout is also
updated to the latest protocol specification.

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

greybus: svc: Replace link config hack with standard operation

The link config operation was a hack only designed to fulfill the camera
driver's needs. Now that a standard operation is defined for the same
purpose, implement it and remove the hack.

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

greybus: arche-apb-ctrl: Do not bring APB out of reset in probe

With addition of handshaking between AP <=> SVC, driver
brings out APB out of reset only on reception of WAKE_MOD
signal from SVC. So remove the deassertion from probe.

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

greybus: arche-apb-ctrl: Set wake_detect gpio to low initially

This patch enables handshaking of AP and SVC using wake_detect
gpio (WD_8A and WD_8B).
Note that WAKE_DETECT polarity is active-high, so in order to
enable handshaking between AP <=> SVC, we need to set wake_detect
gpio to low initially, so that driver can send WAKE_DET signal
(active-high) to SVC and then SVC can send back WAKE_MOD signal
(active-low).
And on reception of WAKE_MOD signal, driver would bring respective
APB out of reset.

  WD_8A => APB1
  WD_8B => APB2

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

greybus: arche-platform: Export gpio (reset & sys_boot) to user

In order to allow user to flash the firmware to,

SVC:
user need to assert the reset first, set sysboot pin and
deassert reset. And then issue a flashing command.

And APB:
User need to assert the reset first, and then issue flashing
command.

So this patch exports the gpio's to user.

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

greybus: camera: Raise the CSI-2 bandwidth

Use 4 lanes at 960MHz to support camera modules requiring higher
bandwidths until we implement support for dynamic bandwidth calculation.

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

greybus: arche-platform: merge arche-apb-ctrl and arche-platform

No need to have two separate arche platform drivers, that's just crazy,
so merge them both together to be only one kernel module.

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

greybus: arche_platform: Remove child's platform device as part of _remove() fn

It seems we need to delete platform_dev of all childs explicitly,
in _remove() fn callback of parent driver.
There were some discussions about having of_platform_unpopulate(),
but it never made it to mainline.

https://lkml.org/lkml/2013/7/19/615

There are some drivers which are removing platform_dev explicitly,
as done in this patch.

Note that, without this, multiple insmod-rmmod won't work, as I see
driver probe gets invoked twice and leads to failure of probe.

Signed-off-by: Vaibhav Hiremath <hiremath_vaibhav@projectara.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: camera: destroy data connection on link-config errors

Make sure to tear down the data connection also on failure to configure
the link by setting the data_connected flag immediately after creating
the connection.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: camera: fix memory leak in capture-request handler

Fix memory leak in capture-request handler by making sure to release the
operation request buffer after sending the request.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: camera: fix memory leak in configure-streams error path

Fix memory leak in configure-streams error path by making sure to
release the operation buffers before returning.

Signed-off-by: Johan Hovold <johan@hovoldconsulting.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: interface: Fetch and expose version of interface's firmware

The version of the currently running firmware on the module is useful
for userspace as it can be used to find if an update is available or
not. This patch fetches interface's version with a new control operation
and exposes the same in userspace.

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: interface: Receive serial-number on hotplug event

Two exactly same modules can be uniquely identified using module's
serial-number. This patch updates the interface hotplug event to also
receive the serial-number of the module.

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: svc: Set interface's hotplug attributes before using them

gb_svc_read_and_clear_module_boot_status() relies on the values of
ddbl1_manufacturer_id and ddbl1_product_id to distinguish between ES2
and ES3 chips, but those values are set for the interface structure only
after gb_svc_read_and_clear_module_boot_status() is called.

This makes ES2 module to fail with following errors:

greybus 1-2: Module not ready yet
greybus 1-svc: failed to clear boot status of interface 2: -19

Fix this by setting these values before calling
gb_svc_read_and_clear_module_boot_status().

Fixes: 51f1dc421b1f ("firmware and svc: detect the difference between ES2 and ES3 chips")
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: svc: Change GB_SVC_TYPE_LINK_CONFIG to 0x10

The greybus specification reserves SVC type 0x0d-0x0f for timesync.
53124d73 ('svc: Add support for the link config operation') allocated the
next available type 0x0d which conflicts with the specification. Change the
type to 0x10 to ensure kernel and specification agree.

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

greybus: control: Use Macro's instead of direct values for major/minor

We already have macro's defined for this, use them.

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

greybus: arche-apb-ctrl: Assert reset to APB at the end of probe

Until we have proper handshake mechanism implemented with SVC
assert reset to APB at the end of probe.
We are safe here to do that, as SVC always enables clock to APB's
currently.
And also from EVT1 perspective, we should be good, as clock control
signals are now moved to AP.

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

greybus: arche-apb-ctrl: Add APB control driver

It was messy to integrate both SVC, APB (and any other arche
platform specific control) into one single driver. Especially
due to cross-dependency.

AP first needs to bringup SVC, as SVC should enable clock to APB.
APB should come up before HUB, as due to some reason HUB wouldn't
enumerate APB's is APB comes up later.
And on top of that we should have clean picture of hardware description
in DT file.

So this patch introduces APB control driver. Mostly copied from original
arche-platform driver + fixed boot sequence.

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

greybus: arche-platform: Rename db3-platform to arche-platform

With multiple platforms getting rolled into ara,
db3 name is confusing. And this driver is applicable
to all arche platforms, so make sense to rename it to
arche-platform.c. Also rename all internal functions
accordingly.

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

greybus: db3(arche)-platform: Remove all APB control code

Current db3(arche)-platform driver was only handling APB control signals,
which is where it all started. But now with completion of DB3/EVT1 bringup
we know that platform driver is more than APB control.

We have to bring SVC up before APB, as SVC supposed to enable
clock to APB's. Note that, in EVT1, AP will have direct control over
APB's clock.
Then we have dependency between APB and USB HUB, where, APB should be
brought up before USB HUB (note that this needs to rootcaused).

This patch cleanup the db3(arche)-platform driver to remove all APB control
code. The idea here is create another driver for APB control
(arche-apb-ctrl.c), which will deal with APB.
And this driver will have generic/common platform specific support,
currently manages SVC resources.

This patch also takes in all the changes from factory branch, discovered
during bringup.

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

greybus: firmware and svc: detect the difference between ES2 and ES3 chips

The Greybus SVC code needs to read and clear the module boot status upon
hotplug; this requires reading two different attributes depending on whether
we're running on ES2 or ES3.  On Marti Bolivar's (mbolivar@leaflabs.com)
advice, we detect ES2 using the unique ES2_DDBL1_MFR_ID and ES2_DDBL1_PROD_ID
for ES2 hardware, and treat all other chips as ES3 appropriately.  This patch
detects the difference and adds the appropriate definitions for ES3 hardware.

Signed-off-by: Eli Sennesh <esennesh@leaflabs.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: interface: Expose DDBL1 manufacturing and production id in sysfs

These ids are already fetched from the SVC, but were never exposed to
sysfs. Userspace may be interested in using these values and hence these
must be exposed to 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: interface: Prefix DDBL1 attributes with ddbl1_ instead of unipro_

The Device descriptor block Level 1 (DDBL1) attributes are specified by
the MIPI standard and prefixing them with 'unipro_' isn't the best thing
to do. They should be prefixed with DDBL1 instead.

To make it more readable/clear:
- rename macros and variable by prefixing them with ddbl1_.
- write full names for mfg and prod ids as manufacturer and product ids.
- replace mfg (manufacturing) with mfr (manufacturer)

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: es2: clean up csi-config request

Use __packed and __u8 for csi-config request that is going out on the
wire.

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

greybus: svc: drop copy-pasted function header

Drop function header that was copied but never updated or removed when
adding gb_svc_link_config.

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

greybus: interface: Prefix hexadecimal values with '0x'

In order to clearly specify the base of values printed using sysfs
files, prefix hexadecimal values with '0x'. Also force the minimum width
(to be printed) for hexadecimal values to their sizes.

To make it more readable make the second argument to gb_interface_attr()
a proper string.

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: es2: Don't use stack memory as USB request data

USB request data must be DMAble memory, allocate it with kzalloc()
instead of declaring it as a local variable.

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

greybus: es2: Fix endian conversion issues

Convert all USB request fields between CPU and protocol endianness.

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

greybus: camera: Fix remaining endian conversion issues

Convert all Greybus operation fields between CPU and protocol
endianness.

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

greybus: camera: Set link power mode to HS-G2 with 2 lanes

HS-G1 isn't enough for all camera modules. The gear will need to be
computed dynamically, but for now hardcode it to 2 with 2 lanes.

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

greybus: camera: Configure the bridge CSI transmitter

Start or stop the CSI transmitter when configuring and unconfiguring the
streams respectively. The CSI configuration parameters are currently
hardcoded.

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

greybus: es2: Add support for CSI transmitter configuration

Export a function from the es2 driver to configure the CSI transmitter
through the corresponding USB vendor control request.

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

greybus: camera: Configure link speed for the data connection

Hardcode the speed to HS-G1 for now.

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

greybus: svc: Add support for the link config operation

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

greybus: camera: Fix endian conversion issues

Convert all Greybus operation fields between CPU and protocol
endianness.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reported-by: Rui Silva <rui.silva@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: tools: don't always build them.

This breaks the kernel-only build as it can't find any userspace headers
with the cross-compiler, so don't build the tools by "default" unless
you ask for them.

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

greybus: loopback_test: null terminate the dict structure

This lets us test for any number of entries, no need to do an
ARRAY_SIZE-type comparison.  This fixes a build warning of comparing
signed/unsigned values.)

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

greybus: loopback_test: fix warning about signed/unsigned comparison

We read an int, don't treat it as a unsigned value, especially when
comparing it to a signed value.

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

greybus: tools: Add tools directory to greybus repo and add loopback

Move the loopback test to the greybus main repo, as we will be adding
more tests over time and it doesn't need to be burried in the gbsim
repo.

This moves the latest version from gbsim to this repo and fixes up the
Makefile to be a bit more "smart" when building the code.

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

greybus: camera: fix data types of operations

In some operations definitions it was introduce some new fields with
the wrong data types, u8, instead of __u8. And because of this gbsim
build was broken.

Fixes: 3a1d7aa15bf6 ("greybus: Add camera protocol definition")

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

greybus: spi: fix and cleanup spi devices handling

Cleanup and remove the spi_devices from the greybus spi handling as they
are not needed and they were completely misused.

With this the gb_spi_init does not make sense to exist anymore, so just
remove it and handle the master config directly from connection init.

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

greybus: spi: unregister master on device add fail

When registering devices if any of it fail, just cleanup and release spi
master.

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

greybus: spi: validate spi master register

Check for error in registering spi master, even though the current code
will fail a little more ahead when trying to register devices in the
master.

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

greybus: spi: rename label to undo operation

Rename error path label to a more significant name related to the free
operation done.

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

greybus: interface: fix hot-unplug control-connection timeouts

Set an interface disconnected flag when the interface has been
hot-unplugged (e.g. forcibly removed or after a reboot), and use it to
disable the control connection early when deregistering the interface
and its bundles.

This avoids a one-second (default) timeout for every enabled connection
(e.g. one per bundle) at hot-unplug, something which for the default
gpbridge manifest currently amounts to five seconds.

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

greybus: interface: clean up control-connection handling

Clean up control-connection handling by managing it through the control
structure and a higher-level control interface. Also make both the
control structure and connection lifetimes coincide with that of the
interface.

The control connection is now only enabled and disabled when the
interface is initialised and removed.

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

greybus: es2: demote dynamic-urb error message

Demote error message when needing to dynamically allocate an URB to
debug level.

This isn't really an error as much as a reminder of how the current es2
implementation works: It uses a fixed number of pre-allocated URBs, but
allocates URBs dynamically when enough messages are sent in rapid
succession to exhaust the URB pool.

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

greybus: Camera: remove f_dentry usage

On newer kernels f_dentry is gone, so use f_path.dentry instead.

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

greybus: Add driver for the camera class protocol

Integration with the V4L2 camera drivers isn't available yet, a debugfs
interface is exposed instead to call the camera Greybus operations.

The debugfs interface will be kept for module testing purpose in order
to exercise all the protocol operations with various valid and invalid
parameters.

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

greybus: Add camera protocol definition

Define the camera data protocol ID and all the protocol operations data
structures.

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

greybus: loopback: Add asynchronous backoff

A specific request from the firmware people is the ability to back-off from
sending more asynchronous operations once a specific number of operations
are in-flight.

This patch adds that ability - with a new sysfs parameter
'outstanding_operations_max' which controls the maximum number of
operations that can be outstanding/in-flight at any time.

When outstanding_operations_max contains a non-zero value and asynchronous
operations are being used - we will back-off until the completion counter
is < outstanding_operations_max. Tested in both synchronous and
asynchronous mode and with gb_loopback_connection_exit() interrupting
in-flight operations.

Suggested-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: loopback: Wait for all async operations to complete on exit

On gb_loopback_connection_exit() we should ensure every issued asynchronous
operation completes before exiting connection_exit(). This patch introduces
a waitqueue with a counter which represents the number of incomplete
asynchronous operations. When the counter reaches zero connection_exit()
will complete. At the point which we wait for outstanding operations to
complete the connection-specific loopback thread will have ceased to issue
new operations. Tested with both synchronous and asynchronous operations.

Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Suggested-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: loopback: Retrun -ENOMEM if operation allocation fails

If operation allocation fails we should return -ENOMEM in the asynchronous
operation send routine. If we don't return here then the
gb_loopback_async_operation_put() later can dereference a NULL pointer if
the previous gb_operation_create() failed.

Reported-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: loopback: Drop NULL check on container_of pointer

container_of cannot return NULL and the pointer passed to this context uses
reference counter bumped inside a spinlock, so the base pointer will be
valid at this point.

Suggested-by: Johan Hovold <johan@hovoldconstulting.com>
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: loopback: register a struct device.

Instead of having the loopback attributes in the bundle device,
Add a struct device to the gb_loopback struct and register it on
connection_init, deregister it at connection_exit, and move the
loopback attribute group over to the new device.

Use device_create_with_groups to create sysfs attributes
together with device.

Suggested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Suggested-by: Johan Hovold <johan@hovoldconsulting.com>
Signed-off-by: Axel Haslam <ahaslam@baylibre.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: loopback: remove mask attribute

The mask attribute is not used on the driver anymore and can be
removed.

Signed-off-by: Axel Haslam <ahaslam@baylibre.com>
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>

greybus: spi: move chipselect to one byte size

Fixed in the specification, some values for chipselect count and index
were different in size, just fix that for all reference to chipselect
and move all to one byte size and remove byte order operations.

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

greybus: es2: Reserve CPorts 16 and 17

CPorts 16 and 17 are reserved for CDSI0 and CDSI1 by the ES2 APB, make
sure they won't be allocated dynamically.

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

greybus: interface: clean up error messages

Drop redundant __func__ and add missing errno values.

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>

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>

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>

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>

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>

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>

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>

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>

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>

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>

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>

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>

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>

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>

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>