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>
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.
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.
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>
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>
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>
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>
Rui Miguel Silva [Thu, 12 Nov 2015 15:36:02 +0000 (15:36 +0000)]
greybus: power_supply: rework and operation changes
This is a major rework and changes to the current implementation of the
battery protocol. The previous implementation lack the support of a more
dynamic handle of power supply properties and updating of status. Also,
reflect the actual state of the greybus specification
So, with this new approach a set of operations to fetch the battery
module configuration and properties is add, new methods to cache and
update the values of properties, new operation to set properties if
declared writable and an event operation that can be triggered by the
module to force an update read on the properties values.
Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Viresh Kumar [Mon, 9 Nov 2015 09:35:03 +0000 (15:05 +0530)]
greybus: es2: Send cport-id in wValue field to usb_control_msg()
wIndex field has a special meaning, as that can be used by the core to
index into the possible USB interfaces. And that specifically broke with
gbsim, as it has a single USB interface.
Other similar requests (REQUEST_LATENCY_TAG_{EN|DIS}) are already using
wValue field for passing cport-id.
Fix cport_reset() by sending the cport-id in wValue field instead of
wIndex.
Johan Hovold [Wed, 4 Nov 2015 17:55:22 +0000 (18:55 +0100)]
greybus: hd: fix host-device life time issues
Fix host-device life time issues by separating host-device allocation
from registration.
This is needed both to make sure that all host-device resources are
available before registering the device and to prevent such resources
from being deallocated while the device is still in use during device
removal.
This specifically fixes the following warnings during es1 and es2
disconnect:
usb 1-1.1: No free CPort OUT urbs, having to dynamically allocate one!
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Johan Hovold [Tue, 3 Nov 2015 11:11:26 +0000 (12:11 +0100)]
greybus: connection: kill gb_hd_connections_exit
Connections are destroyed as part of interface tear down. If we fail to
do that properly it's a bug that should be fixed rather than papered
over by a fall-back clean up function.
Bryan O'Donoghue [Thu, 29 Oct 2015 16:58:29 +0000 (16:58 +0000)]
greybus: audio-pcm: fix use of variable unitialized
Compiling with clang shows that period_elapsed will be used as a branch
conditional unitialized whenever snd_dev->transfer_done <
runtime->period_size. Since stack can grow up/down as we proceed though
the call stack this should be fixed. This patch fixes by explicitly
initalizing period_elapsed to zero.
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> Reviewed-by: Alex Elder <elder@linaro.org> Reviewed-by: Mark Greer <mgreer@animalcreek.com> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Bryan O'Donoghue [Thu, 29 Oct 2015 16:58:24 +0000 (16:58 +0000)]
greybus: manifest: fix the placement of arguments to pr_err
We're one character out here in the placement of the inputs to pr_err().
Later patches show this up when pushing through checkpatch.pl. This patch
fixes by moving the offending variables one character left.
Alex Elder [Wed, 28 Oct 2015 03:18:41 +0000 (22:18 -0500)]
greybus: es2: don't assume just one AP bridge
Previously we had only one AP bridge connected to the AP. We will
now have two. Now that the have moved the logging related symbols
into the AP device, we can stop assuming there's only on of them.
Specifically, rename symbols like "apb1_*" to be just "apb_*",
because the "1" is no longer meaningful.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Alex Elder [Wed, 28 Oct 2015 03:18:40 +0000 (22:18 -0500)]
greybus: es2: move logging dentries into es2 struct
Finish moving the logging data structures into the es2 AP data
structure rather than having it be a single global. This patch
moves the two dentries related to logging.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Alex Elder [Wed, 28 Oct 2015 03:18:38 +0000 (22:18 -0500)]
greybus: es2: move APB log task into the es2_ap_dev struct
If an APB has a logging task it is associated with that APB. Move
the task pointer into the es2_ap_dev structure rather than having it
be a single global.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Alex Elder [Wed, 28 Oct 2015 03:18:37 +0000 (22:18 -0500)]
greybus: es2: rename es2 data structures
I tried this once before and Greg NAK'd it because at that point the
es2 code was nearly identical to the es1 code. This is no longer
the case, and we need to diverge further, so I think it's time to go
down that path.
The ap_dev structure changed significantly for ES2 versus ES1 as of
this commit: 667f8d3 es2.c: create dedicated struct for cport_in and cport_out
Since the structures are no longer the same, they should not have
the same name.
This patch renames three data structures so the "1" is replaced with
a "2", to reflect the Toshiba AP bridge chip revision we are working
with. The structures are:
es1_ap_dev -> es2_ap_dev
es1_cport_in -> es2_cport_in
es1_cport_out -> es2_cport_out
It changes names of symbols having this type as well. To finish the
job, all references "ES1" (in comments and in symbol names) have
been switched to use "ES2" instead.
The result is a lot of changes, but they amount to a global search
and replace of "es1" with "es2" (and "ES1" with "ES2"), and the
result has been compile tested.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
We now have an officially assigned Google USB device id for the
APBridge, so add in support for it in the es2 driver. The old entry can
be removed once the firmware has caught up and uses the new number.
We should use the version field to determine ES1/ES2/ES3, but that will
come later, for now, grab anything with this device id.
Alex Elder [Thu, 22 Oct 2015 18:03:26 +0000 (13:03 -0500)]
greybus: db3-platform: get rid of redundant gpio tests
In apb_ctrl_get_devtree_data(), the value returned by
of_get_named_gpio() has a redundant test for a negative
return value. GPIO numbers are non-negative, so this
test is redundant--testing gpio_is_valid() is sufficient.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Viresh Kumar [Wed, 21 Oct 2015 09:56:45 +0000 (15:26 +0530)]
greybus: module: fix double freeing of module structure
The module will be released by gb_module_release() once all references
for the module are dropped. And so there is no need to free it in the
error path specially.
Viresh Kumar [Wed, 21 Oct 2015 09:56:44 +0000 (15:26 +0530)]
greybus: interface: fix double freeing of interface structure
The interface will be released by gb_interface_release() once all
references for the interface are dropped. And so there is no need to
free it in the error path specially.
Viresh Kumar [Wed, 21 Oct 2015 09:56:43 +0000 (15:26 +0530)]
greybus: bundle: fix double freeing of bundle structure
The bundle will be released by gb_bundle_release() once all references
for the bundle are dropped. And so there is no need to free it in the
error path specially.
Johan Hovold [Wed, 21 Oct 2015 09:51:43 +0000 (11:51 +0200)]
greybus: endo: fix device-id allocation
During endo registration, a unique device id was allocated but then
never used. Instead the default dev_id 0 (due to kzalloc) was used for
device-generation and later for id-deallocation.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Vaibhav Hiremath [Wed, 21 Oct 2015 11:20:20 +0000 (16:50 +0530)]
greybus: platform: Add platform driver for DB3 AP bridge
With DB3, we now have AP as a master as far as AP bridges are concerned.
SVC will talk only to AP and AP will control bridges; unlike other
module interfaces.
So AP supposed to manage/control bridges in all power states including
power on reset.
During power on reset AP should follow below sequence -
Sequence (treated as a Cold Boot)
Stage-1
=======
AP:
- Power On (Power up from PMIC to AP)
- AP start booting
Since power to AP bridges are controlled through gpio, power is gated
to APB1 & 2
No ref_clk to APB available (under SVC's control)
- AP configures USB hub to enable HSIC interface to APB
- As part of platform driver probe, AP follow below sequence
- Set the pinctrl in default state
- Hold APBs in reset by pulling down reset pin
- Enable power to APB by enabling regulator and switches
- De-assert (set 'low') 'boot_ret' signal
- AP will assert (set 'high') the wake_detect signal, triggering
connect/detect event to the SVC
- AP waits for wake pulse from SVC
SVC:
- Power On (power up from PMIC to SVC)
- SVC starts booting
- SVC will de-assert reset signal to unipro switch
- Switch starts booting
- SVC confirms switch boot status using SPI (or something)
- SVC waits for 300 msec (ES2 known issue)
- SVC waits for detect/connect event from AP
Stage-2
=======
SVC:
- ON connect/detect event, SVC send back wake pulse (cold boot)
to AP over wake_detect pin, if SVC boot is completed.
AP:
- On wake pulse from SVC (for cold boot), AP de-asserts (set high')
reset signal to APB 1 and/or 2
- Bridges starts booting
- Eventually Unipro linkup occurs
Testing:
- Build tested against Helium kernel
- Due to unavailability of MSM and DB3 platform, only minimal testing
has been done.
- Code has been modified for validation on Helium + SDB platform.
Mostly dts changes for gpio numbers
And debug messages to check gpio values
- On Helium + SDB platform, with addition of debug messages validated
the sequence.
TODO list:
- Currently _only_ supports power on sequence (cold boot).
Both warm and cold boot support. Cold and Warm boot is
differentiated based on pulse width of wake_detect signal
>=5 msec = Cold boot else Warm boot
- No support for Power management
So the "power-down", "power-off", "wake_in" and "wake_out"
signals are not explored/implemented.
- Support for Work thread
repetitive wake signal if no response from peer
May required for PM support, as we have delays in the sequences
- pinctrl states, specially to make sure we enable right pullup or pulldown
when we set wake_detect pin to input
- Convert gpio list into an array, and associated xxx-gpio-name property
As the svc code is "odd" in how things are set up at initial connection
time, we can't always "know" that we have a valid bundle to use for
error messages. So just punt and always use pr_*() calls to ensure that
we never incorrectly dereference a pointer.
This will get fixed up soon, but for now, let's just live with a bit
messier error messages in the log.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org> Reviewed-by: Alex Elder <elder@linaro.org> Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
greybus: core: remove uevent handling for gb_connection
As we are going to be removing the struct device from gb_connection,
there is no need to do anything for uevents for them. So just remove
the code. It wasn't doing anything anyway, so no functionality is lost
here at all.
As is_gb_connection() is no longer used, that is also removed in this
patch.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org> Reviewed-by: Alex Elder <elder@linaro.org> Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
greybus: bundle.h: add a private field to struct gb_bundle
The gb_bundle structure needs a private field if we are going to be able
to replace the gb_connection device with this one for all use cases.
Ideally we could use the private pointer within the struct device, but
for now let's just try to mirror how people were using the gb_connection
structure to make the patches simpler, and the logic the same.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org> Reviewed-by: Alex Elder <elder@linaro.org> Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
greybus: protocol: use the bundle struct device instead of the connector
We are removing struct device from the gb_connection structure in the
near future. The gb_bundle structure's struct device should be used as
a replacement.
This patch moves the protocol code to use the bundle pointer instead of
the connection pointer when printing out error messages.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org> Reviewed-by: Alex Elder <elder@linaro.org> Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
greybus: greybus_trace: use the bundle struct device instead of the connector
We are removing struct device from the gb_connection structure in the
near future. The gb_bundle structure's struct device should be used as
a replacement.
This patch moves the greybus trace code to use the bundle pointer
instead of the connection pointer when printing out tracing messages.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org> Reviewed-by: Alex Elder <elder@linaro.org> Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
greybus: loopback: use the bundle struct device instead of the connector
We are removing struct device from the gb_connection structure in the
near future. The gb_bundle structure's struct device should be used as
a replacement.
This patch moves the loopback driver to use the bundle pointer instead
of the connection pointer.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org> Reviewed-by: Alex Elder <elder@linaro.org> Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
greybus: connection: use the bundle struct device instead of the connector
We are removing struct device from the gb_connection structure in the
near future. The gb_bundle structure's struct device should be used as
a replacement.
This patch moves the connection code to use the bundle pointer instead
of the connection pointer for some error messages.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org> Reviewed-by: Alex Elder <elder@linaro.org> Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
greybus: connection: remove 'struct device' from 'struct gb_connection'
We don't want this in the driver core, as nothing will be binding to it,
that's the job of a bundle. So remove the struct device and use a kref
to handle reference counting instead.
Note, I don't think we really need a kref, but it keeps the lifetime the
same as before, and is the simplest change for now. In the future it
might be easier to just attach all connections to the bundle and clean
them up when the bundle is removed.
Also remove the cport sysfs documentation as it's no longer relevant.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org> Reviewed-by: Alex Elder <elder@linaro.org> Reviewed-by: Johan Hovold <johan@hovoldconsulting.com>
Rui Miguel Silva [Thu, 15 Oct 2015 22:56:51 +0000 (23:56 +0100)]
greybus: sdio: send data block details at command request
If SDIO request include data to be transfer send details (data blocks
and block size) in command request, as it seems some controllers need
this info prior to set the registers correctly.
Signed-off-by: Rui Miguel Silva <rui.silva@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Bryan O'Donoghue [Thu, 15 Oct 2015 15:10:45 +0000 (16:10 +0100)]
greybus: loopback: capture and present firmware supplied latencies
In order to provide deep inspection of the greybus/UniPro system
instrumentation of
1. APBridge's view of UniPro latency
2. GPBridge's view of internal firmware-only latency
have both been added and reported to the AP in the transfer loopback
response header. When this data are present we latch and average it over
the number of requested cycles, presenting it to user-space via sysfs.
This patch adds the following sysfs entries for each loopback CPort
Bryan O'Donoghue [Thu, 15 Oct 2015 15:10:44 +0000 (16:10 +0100)]
greybus: loopback: add tracker variables to hold firmware timestamps
This patch adds tracker variables to hold the incoming firmware derived
timestamps where apbridge_latency_ts will contain the APBridge's view of
the UniPro turn-around time and gpbridge_latency_ts will contain the
GPBridge's view of it's own internal latency. Both values are reported
in microseconds.
Bryan O'Donoghue [Thu, 15 Oct 2015 15:10:43 +0000 (16:10 +0100)]
greybus: loopback: send command to APBridge to tag throughput
A USB vendor command has been added to APBridge to allow for tagging of
specific CPort identifiers with internal timing data, specifically geared
towards capturing and understanding latencies in the UniPro fabric. This
patch sends a command to APBridge for each known loopback CPort to tag data
as appropriate. Subsequent patches will present this data to user-space for
ongoing integration analysis.