Alex Elder [Fri, 22 May 2015 14:52:43 +0000 (09:52 -0500)]
greybus: endo: pass endo_id to gb_endo_create()
We are going to want to defer creating the endo until we receive a
probe operation from the SVC, which will supply the endo id. Change
gb_endo_create() so it passes the endo_id value as an argument.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Alex Elder [Fri, 22 May 2015 14:52:42 +0000 (09:52 -0500)]
greybus: endo: rename gb_svc
Change the name of "struct gb_svc" to be "struct svc_info". The
structure now contains only the SVC's serial number and version (and
are place holders anyway). We will be defining a structure that
represents the SVC for the SVC protocol connection, and I want to
take back that name.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Mark A. Greer [Thu, 21 May 2015 22:57:04 +0000 (15:57 -0700)]
greybus: gb-audio: Set samples per message during init
A recent commit moved the I2S samples per message operation
to the PCM's 'hw_params' callback. However, the 'hw_params'
callback is called numerous times while the samples per
message need only be done once (or seldom). Eliminate the
unnecessary samples per message operations by doing it only
once at Greybus protocol init time.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com> Acked-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Mark A. Greer [Thu, 21 May 2015 22:57:03 +0000 (15:57 -0700)]
greybus: gb-audio: Set I2S Configuration according to ASOC requests
Currently, the audio driver unconditionally sets the I2S
configuration to have a sample rate of 48KHz, two channels,
16 bits per channel, in little endian order. Make this
more flexible by setting the I2S configuration according to
the arguments passed to the PCM 'hw_params' callback.
To accomplish this, query for the supported I2S configurations
at Greybus protocol init time and save them in the 'snd_dev'
structure. When the 'hw_params' callback is called, compare its
arguments to the table of supported configurations. If there is
a match, set the I2S connection accordingly.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com> Acked-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Mark A. Greer [Thu, 21 May 2015 22:57:02 +0000 (15:57 -0700)]
greybus: gb-audio: Activate TX connection Bundle CPort ID
Currently, the audio driver activates & deactivates a predefined
CPort ID but that can vary depending on the manifest data of the
module. Instead, use the TX connection's Bundle CPort ID which
contains the correct CPort ID.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com> Acked-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Mark A. Greer [Thu, 21 May 2015 22:57:01 +0000 (15:57 -0700)]
greybus: gb-audio: Activate TX CPort in PCM workqueue
Currently, the I2S TX CPort is configured and activated during
the Greybus audio initialization. Unfortunately, this prevents
the audio driver from ever changing the I2S configuration.
To allow the I2S configuration to change according to ASOC requests,
move the CPort activation & deactivation to the audio-pcm workqueue.
Now, when audio is running but the CPort is not active, it will be
activated. When audio is not running and the CPort is active, it
will be deactivated.
This has the side-effect of sending the first piece of audio data
immediately after activating the CPort which is really how it should
work anyway.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com> Acked-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Mark A. Greer [Thu, 21 May 2015 22:57:00 +0000 (15:57 -0700)]
greybus: gb-audio: Remove useless comment
Remove comment about adding start delay since it will
be done when support for A/V synchronization is added.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com> Acked-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Mark A. Greer [Thu, 21 May 2015 22:56:59 +0000 (15:56 -0700)]
greybus: gb-audio: Add I2C device for rt5647 codec
Add the I2C device node for the rt5647 codec.
Eventually, this will be done automatically somewhere
else but for now its done in the audio driver.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com> Acked-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Mark A. Greer [Thu, 21 May 2015 22:56:58 +0000 (15:56 -0700)]
greybus: gb-audio: cpu_dai.fmt does not exist in v4.1
The asoc_simple_dai structure does not contain the 'fmt'
member in Linux kernel version v4.1 and later so only
build code that uses it when the kernel version is earlier
than v4.1.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com> Acked-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Mark A. Greer [Thu, 21 May 2015 22:56:57 +0000 (15:56 -0700)]
greybus: gb-audio: Fix DAI formats and master/slave settings
Set the various DAI formats so the bridge on the module
is the master of all clocks and the codec is the slave.
The only DAI protocol currently supported is I2S.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com> Acked-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Mark A. Greer [Thu, 21 May 2015 22:56:56 +0000 (15:56 -0700)]
greybus: gb-audio: Clean up codec name generation
Instead of using the fixed suffix, '6-001b', in the
codec name, generate it from the I2S adapter number
and I2C address of the codec.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com> Acked-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Mark A. Greer [Thu, 21 May 2015 22:54:48 +0000 (15:54 -0700)]
greybus: gb-audio: Allocate space for 20 configurations
The greybus code requires that an operation's response buffer be
the exact size of the response; however, the size of the response to
the GB_I2S_MGMT_TYPE_GET_SUPPORTED_CONFIGURATIONS operation is unknown.
To fix this, an extension to the I2S specification is required.
In the meantime, set the number of configurations returned to 20
because that is how many configurations will be returned (using
"insider knowledge" of the firmware).
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com> Tested-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
(Those are the setting that work, at least). So make the Greybus
audio driver configure the GPBridge with those settings.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com> Tested-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Johan Hovold [Thu, 21 May 2015 09:47:24 +0000 (11:47 +0200)]
greybus: gpbridge: fix section mismatches
Fix section mismatches introduced by b27227ce93c0 ("greybus: Use
gb_gpbridge_protocol_init()"), which added __exit annotation to
gpbridge-protocol exit functions that are called in the error path of
gpbridge_init, which lives in the init section.
This triggered the following modpost warning:
WARNING: modpost: Found 8 section mismatch(es).
Fixes: 16b33d100bff ("protocol: Add gb_gpbridge_protocol_driver()") Fixes: b27227ce93c0 ("greybus: Use gb_gpbridge_protocol_init()") Signed-off-by: Johan Hovold <johan@hovoldconsulting.com> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Viresh Kumar [Wed, 20 May 2015 11:01:19 +0000 (16:31 +0530)]
greybus: raw: include uaccess.h to fix warning
This is what I get over mainline:
greybus/raw.c: In function 'gb_raw_send':
greybus/raw.c:153:2: error: implicit declaration of function 'copy_from_user' [-Werror=implicit-function-declaration]
if (copy_from_user(&request->data[0], data, len)) {
^
greybus/raw.c: In function 'raw_read':
greybus/raw.c:305:2: error: implicit declaration of function 'copy_to_user' [-Werror=implicit-function-declaration]
if (copy_to_user(buf, &raw_data->data[0], raw_data->len)) {
^
This adds a driver that implements the greybus Raw protocol as
specified.
It preserves the message boundries by only allowing a read to receive
a "full" message, and any write() call also is passed in a single
greybus request.
Totally untested, given that we have no raw firmware or gbsim code yet.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Rui Miguel Silva <rui.silva@linaro.org>
All we need is a simple ida, so use that interface instead of the more
"complex" idr one. Bonus is we don't need to fix the locking issue I
forgot about when using an idr, as ida has one built-in.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Alex Elder <elder@linaro.org>
Add some symbols to indicate what kind of "function" the
loopback thread is supposed to be performing--the type of
traffic it generates over its connection.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Alex Elder [Tue, 12 May 2015 02:16:34 +0000 (21:16 -0500)]
greybus: loopback: fix the type attribute check
In gb_loopback_check_attr(), the value of gb->type is checked for
validity. The only valid values are 0, 1, and 2. But the check
allows the value 3. Fix that.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
In commit 1efb53a220b7 ("ASoC: simple-card: Remove support for setting
differing DAI formats"), the .fmt field was removed from struct
asoc_simple_dai. Fix this build breakage by not trying to set it.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Acked-by: John Stultz <john.stultz@linaro.org>
Introduce INSTALL_MOD_PATH to allow for building and installing of the
greybus modules from a different location.
This lets you build the greybus modules on a PC and then install the
modules to an SD card in the appropriate location relative to the SD such
as /media/sdcard/lib/modules/version and subsequent running of depmod in
the same location. If INSTALL_MOD_PATH isn't specified the default
behaviour of installing and depmoding to /lib/modules/version is
maintained.
So here's the current greybus audio driver, which I
wanted to send out for more serious review and
consideration for merging.
I've tried to integrate much of the feedback from the
last round and most of the hotplug issues that I've found
have been resolved. I've tested this via gbsim, and the
Android ARA HAL layer seems to work with it.
Mark has also successfully played audio with this driver,
adding a few hacks to get the codec's i2c connection to
probe.
Current issues:
* Hotplug problem - When gbsim is killed, or the module
removed, the greybus driver gets stuck since the android
mediaserver process is holding the audio device open.
Killing the mediaserver allows things to clean up and
allows greybus to accept new gbsim connections. I have
a workaround patch to the soc-core.c logic which converts
the snd_card_free() call to snd_card_free_when_closed()
which allows the greybus connection cleanup to finish.
Remaining todos:
* Probably need to break apart the mgmt_setup function
to integrate better with the constraint logic. I took
a really basic stab at this, but more is probably
needed.
* Figure out how to properly find and tie in the
codec's I2C bus-id to the driver.
This code requires that the kernel support the following
config options, which I've enabled in a separate kernel
patch:
CONFIG_SND_SIMPLE_CARD
CONFIG_SND_SOC_SPDIF
CONFIG_SND_SOC_RT5645
I really can't calim to be the sole author of this, since
many many fixes and tweaks that have been folded in have
come from Mark Greer. His analsysis and debugging is really
what has made this dummy-framework driver evolve into an
actual audio driver. So much credit and thanks to Mark!
Signed-off-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Mark A. Greer <mgreer@animalcreek.com> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Alex Elder [Thu, 7 May 2015 18:03:52 +0000 (13:03 -0500)]
greybus: eliminate extra response flag definitions
All protocols use the same value to distinguish between request and
response message types. This is a requirement.
Use GB_MESSAGE_TYPE_RESPONSE rather than GB_OPERATION_TYPE_RESPONSE
for the name of the flag used to distiguish between request and
response messages.
Get rid of the redundant response flag definitions that are
associated with specific protocols.
Describe the symbolic values as "operation types" rather than
"message types" where they are defined. The message type for a
request is the same as the operation type; the message type for a
response is the operation type OR'd with GB_MESSAGE_TYPE_RESPONSE.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Alex Elder [Thu, 7 May 2015 18:00:21 +0000 (13:00 -0500)]
greybus: battery: use feature tag rather than kernel version
Conditionally define a new symbol DRIVER_OWNS_PSY_STRUCT, which is
set in "kernel_ver.h" based on on the kernel version. Use it to
distinguish code used for kernels that differ in whether a power
supply structure is owned by the driver, or by the power supply
core.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Alex Elder [Thu, 7 May 2015 18:00:20 +0000 (13:00 -0500)]
greybus: battery: free struct on error in caller
When a battery connection is initialized, a gb_battery structure for
it is allocated in gb_battery_connection_init(). Currently that
function ends by calling init_and_register(); in the event an error
occurs, init_and_register() is responsible for freeing the allocated
gb_battery structure.
Make the code a bit better balanced by having the function that
allocates the structure be responsible for freeing it in case of
error.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
greybus: battery: update for 4.1 power supply api changes
The 4.1-rc1 kernel changed the power supply apis such that the
structures are now owned by the power supply core, and not the
individual drivers. This broke the greybus battery driver, so update it
to support both the old and the new apis.
The API changes were such that I can't "hide" them in kernel_ver.h, but
rather the driver itself needs to have ugly #ifdefs in it. I tried to
keep it to a minimum, making a sub-function for initializing the power
supply device that is implemented differently for different kernel
versions.
When this is submitted upstream, or if we ever move our AP development
to 4.1 or greater, the support for older kernels can be removed.
Alex Elder [Fri, 17 Apr 2015 19:41:47 +0000 (14:41 -0500)]
greybus: bundle: use kstrdup() for state file
The kernfs code guarantees we'll get a NUL-terminated buffer.
Use kstrdup() rather than kzalloc() + memcpy() in state_store()
making it slightly clearer what we're doing. This has the added
benefit of guaranteeing that the stored string has no NUL character
inside it.
Signed-off-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
greybus: manifest: Warn if descriptor size > expected size
A descriptor passed to AP can be bigger than what AP expects, if
manifest's minor version is higher with same major number as the AP. As
it can have some extra data in descriptor.
But, if AP and manifest versions are identical, or if the AP's minor
version is greater than the manifest version, we should at least warn
(if not fail).
Doing this would require some changes to record the manifest version
somewhere reachable by identify_descriptor().
For now, just warn if descriptor is bigger than expected.
There can be three Endo types: mini, medium and large. And that's what
Endo 'type' should refer to.
But we have named the 16 bit number that uniquely represents a valid
endo, as its type. 'id' seems to be a more suitable name to that instead
of 'type'. Lets rename it.
A bundle has a state file, that is managed by the endo userspace
process. This file can be written to and any process that is polling on
the file will be woken up and can read the new value. It's a "cheap"
IPC for programs that are not allowed to do anything other than
read/write to kernel sysfs files.
Johan Hovold [Thu, 16 Apr 2015 07:53:59 +0000 (09:53 +0200)]
greybus: loopback: fix 64-bit divisions
The code uses 64-bit divisions, which should be avoided, and also
prevents the module from loading on 32-bit systems:
gb_loopback: Unknown symbol __aeabi_uldivmod (err 0)
Fix by using the kernel's 64-bit by 32-bit division implementation
do_div.
Compile tested only. I did not look very closely at the code itself.
Perhaps this could be worked around in some other way, but this silences
the linker warning and allows the module to be loaded.
Reviewed-by: Alex Elder <elder@linaro.org> Signed-off-by: Johan Hovold <johan@hovoldconsulting.com> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
greybus: Documentation/sysfs-bus-greybus: update kernel version and date
The kernel is now on the 4.XX numbering scheme, and it's going to be a
while before we merge this code, so pick a date sometime in the future
to be safe.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Alex Elder <elder@linaro.org>
This hooks up the endo, and modules, into the device tree. All modules
for a specific endo are created when the host device is initialized.
When an interface is registered, the correct module for it is found and
that module is used for the sysfs tree. When the interface is removed,
the reference on the module is dropped.
When the host device goes away, the whole endo and modules are removed
at once.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Alex Elder <elder@linaro.org>
This adds the attributes power_control and present to a module. It also
removes the unneeded module_id attribute, as that comes from the name of
the module itself.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com> Reviewed-by: Alex Elder <elder@linaro.org>
Johan Hovold [Tue, 7 Apr 2015 09:27:21 +0000 (11:27 +0200)]
greybus: drop host-driver buffer headroom
Drop the host-driver buffer headroom that was used to transfer the cport
id on ES1 and ES2.
Rather than transferring additional bytes on the wire and having to deal
with buffer-alignment issues (e.g. requiring the headroom to be a
multiple of 8 bytes) simply drop the headroom functionality.
Host drivers are expected set up their transfer descriptors separately
from the data buffers and any intermediate drivers (e.g. for Greybus
over USB) can (ab)use the operation message pad bytes for now.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com> Reviewed-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Johan Hovold [Tue, 7 Apr 2015 09:27:19 +0000 (11:27 +0200)]
greybus: es1: fix transfer-buffer alignment
Fix transfer-buffer alignment of outgoing transfers which are currently
byte aligned.
Some USB host drivers cannot handle byte-aligned buffers and will
allocate temporary buffers, which the data is copied to or from on every
transfer. This affects for example musb (e.g. Beaglebone Black) and
ehci-tegra (e.g. Jetson).
Instead of transferring pad bytes on the wire, let's (ab)use the pad
bytes of the operation message header to transfer the cport id. This
gives us properly aligned buffers and more efficient transfers in both
directions.
By using both pad bytes, we can also remove the arbitrary limitation of
256 cports.
Note that the protocol between the host driver and the UniPro bridge is
not necessarily Greybus. As long as the firmware clears the pad bytes
before forwarding the data, and the host driver does the same before
passing received data up the stack, this should be considered "legal"
use.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com> Reviewed-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Make sure to allocate the message transfer-buffer separately from the
containing message structure to avoid data corruption on systems without
DMA-coherent caches.
The message structure contains state that is updated while the buffer
may be used for DMA, something which could lead to data corruption due
to cache-line sharing on some architectures.
Use the (renamed) message cache for the message structure itself and
allocate the buffer separately.
If the additional allocation is a concern, the message structures
could eventually be allocated as part of the operation structure.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com> Reviewed-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Remove unused and unnecessary buffer-alignment define that host driver
were supposed to use.
We can handle unaligned incoming buffers just fine by accessing the
operation-message header via a copy in the receive path, rather than
requiring host drivers to make sure the alignment is correct.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com> Reviewed-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Johan Hovold [Tue, 7 Apr 2015 09:27:13 +0000 (11:27 +0200)]
greybus: operation: fix unaligned memory accesses in receive path
The buffer received from our current host driver is 1-byte aligned and
will therefore cause unaligned memory accesses if simply cast to an
operation-message header.
Fix this by making a properly aligned copy of the header in
gb_connection_recv_response before accessing its fields.
Note that this does not affect protocol drivers as the whole buffer is
copied when creating the corresponding request or response before being
forwarded.
Signed-off-by: Johan Hovold <johan@hovoldconsulting.com> Reviewed-by: Alex Elder <elder@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
projects / mirror_ubuntu-zesty-kernel.git / log