The cdc-acm workqueue thread name is set before initializing it right
now, which results in an empty thread name as apparently it gets cleared
by k_work_queue_init. Fix it by moving the thread name setting after the
workqueu is initialized and started.
Signed-off-by: Fabio Baltieri <fabiobaltieri@google.com>
The consequences of a wSequence mismatch are unspecified and are
primarily for debugging purposes. Our checks are a bit too strict, and
if the header check fails, the sequence is not updated, and the header
check subsequently fails. Rework the code to just log a warning on
mismatch and also reset OUT sequence counter the same way as the IN
sequence counter.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Align the OUT handling with the CDC NCM and change the logic to track
when the alternate data interface is set/reset instead of the class
implementation being part of the active configuration.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
When interface alternate is set back to zero, the implementation still
attempts to submit OUT transfers, but the addressed endpoint is actually
disabled. Change the logic a bit to track when the alternate data
interface is set/reset instead of the class implementation being part of
the active configuration.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Do not fail in iface_start/iface_stop if sending notification fails,
just log the error and try to send connected notification when alternate
interface is set and network interface is started.
Do not block in notification send function and allow to submit more than
one notification which means that the last valid one will also be
delivered.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
The Zephyr network interface should not be disabled by the USB device
functions. This sounds confusing, but in the current implementation they
are different ends of the connection.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
We need to handle notifications periodically, for example, when the
interface alternate changes and we also need to track whether the
notification was actually transfered to the host. There is no need to
trigger notification work on iface_start/iface_stop when the USB
interface is not in active configuration. This commit intentianly
changes log level in the notification handling code.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
In the current implementation, the block length in the OUT direciton
should be equal to the transfer length.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
The implementation should not assume that NDP is placed immediately
after NTH. According to the specification (revision 1.0), NDP can be
anywhere in the transfer buffer. There is also always a "terminating
zero datagram pointer entry", which we should also use to terminate the
datagram processing loop.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
With changes introduced in commit 6a3602a306
("net: buf: Clear `user_data` on allocation")
our memset() calls are redundant.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
IAD must be before the interfaces it associates and therefore there is
no need for the class to be in charge of updating the bFirstInterface.
Update IAD in common initialization code and remove the updates from
classes.
This fixes UAC2 instances where the IAD bFirstInterface is not 0, e.g.
when HID was used together with UAC2.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Application is expected to call usbd_uac2_send() on each enabled USB
Streaming Output Terminal (isochronous IN data endpoint) exactly once
every SOF. The class is bookkeeping queued transfers to make it easier
to determine component at fault when things go wrong. However, this
approach only works fine if the underlying USB device controller buffers
the data to be sent on next SOF and reports the transfer completion
immediately after data is buffered (e.g. nRF52 USBD).
While DWC2 otg also requires the SW to arm endpoint with data for the
next SOF, unlike nRF52 USBD the transfer is only considered complete
after either the IN token for isochronous endpoint is received or after
the Periodic Frame Interval elapses without IN token. This design
inevitably requires the application to be able to have at least two
buffers for isochronous IN endpoints.
Support dual buffering on IN data endpoints to facilitate sending
isochronous IN data on every SOF regardless of the underlying USB device
controller design.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Remove the deprecated HCI driver API which was provided by the hci_driver.h
header file. The deprecation happened in Zephyr 3.7, so the API can now be
removed for Zephyr 4.1.
Signed-off-by: Johan Hedberg <johan.hedberg@silabs.com>
In case the host doesn't pull the new data from the endpoint, the work task
would schedule itself again delayed (at the max. priority). When there is
no terminal program or active application reading the endpoint this
results in a constant polling of the endpoint burning up to 5% of the
CPU cycles.
By using a atomic flag for tx busy, the polling is solved and changed into
a postponed execution of the next work task which saves up to 5% of
CPU cycles and allows a better real-time behavior for other tasks.
Secondly, if the TX interrupt is disabled but there is still data in the TX
FIFO (ring buffer), the implementation will continue to trigger subsequent
TX work and attempt to flush the data to the host.
Signed-off-by: Vincent van der Locht <vincent@synchronicit.nl>
Rework NUM_SPATIAL_LOCATIONS() to evaluate to integer literal to allow
using it with LISTIFY() macro. This is necessary because Feature Unit do
not operate on channel clusters but rather on logical channels.
Track back Output Terminal channel cluster to appropriate entity. This
is necessary because Feature Unit does not repeat the channel cluster
information.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
This prevents recursive logging loop when USB CDC ACM is used for shell
with logging.
Change affects both USB stacks.
Signed-off-by: Xudong Zheng <7pkvm5aw@slicealias.com>
Use delayable work to reduce CPU load when there is no transfer flow in
the host direction. This also improves the performance of poll out, as
introduced in commit commit fed6bde788
("usb: device: cdc_acm: send more than 1 byte in poll out")
for the legacy CDC ACM implementation.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
When the controller is connected to a full speed bus, regardless of
whether the controller supports high speed or not, the transfer size for
the bulk OUT endpoint should be equal to the MPS in the current
configuration.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
As it is still accepted practice, allow fifo_fill and poll_out to be
used simultaneously. The lock around fifo_fill was already in place.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Apply changes in commit c152e0980c
("usb: device: cdc_acm: block in uart_poll_out() routine")
to the new CDC ACM UART poll_out implementation.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Align CDC ACM UART with Interrupt-driven UART API behavior description.
Use the same flags in uart_irq_rx_ready(), uart_irq_tx_ready(), and
uart_irq_is_pending(), which are updated after each uart_irq_update()
call. Allow TX FIFO to be filled if there is space.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Update the interrupt driver UART drivers that use `struct ring_buf`
internally to report the number of bytes that can be pushed in
`uart_fifo_fill` without fragmentation.
Signed-off-by: Jordan Yates <jordan@embeint.com>
Since UDC buffers are allocated with `UDC_BUF_GRANULARITY` granularity,
the `net_buf_tailroom` may no longer be equal to the HID report size.
Use `setup->wLength` instead to ensure that proper HID report size is
passed to the application's callback.
Signed-off-by: Marek Pieta <Marek.Pieta@nordicsemi.no>
Add callbacks for setting and getting the sample rate. The callbacks are
optional if all Clock Source entities support only one sample rate.
This commit results in working High-Speed operation with Windows UAC2
driver when the Clock Source is host-programmable. Windows UAC2 driver
won't work if setting sample rate fails even if Clock Source supports
only one sample rate.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Queue explicit feedback write once the previous one finishes in order to
make it possible to cope with 8x faster High-Speed pace. This matches
the originally intended behavior described in comment.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Generate Full-Speed and High-Speed descriptors based on full-speed and
high-speed instance properties. Use appropriate descriptor set based on
active speed.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Generate full and high speed descriptors based on UAC2 instance
properties. Fail the build if an instance does not have speed properties
or if the instance with given speed property exceeds the bandwidth
available at the selected speed.
Remove obsolete no longer used macros to determine offsets inside
complete UAC2 descriptors blob. Only speed-specific descriptors arrays
are supported.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Move the network buffer header file from zephyr/net/buf.h to
zephyr/net_buf.h as the implementation now lives outside of the networking
subsystem.
Add (deprecated) zephyr/net/buf.h header to maintain compatibility with old
file path.
Signed-off-by: Henrik Brix Andersen <henrik@brixandersen.dk>
Add macros for converting between Max Packet Size and total payload
length. Allow drivers specify whether endpoint supports high-bandwidth
interrupt and high-bandwidth isochronous transfers.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Use COND_CODE_1() instead of UTIL_AND() to make it possible to only
define the HID_MPS_LESS_65_ macros up to value 64.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
wMaxPacketSize in endpoint descriptor is stored in little-endian order,
but the mps parameter passed to functions is in host order.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Convert users of net_buf_put() and net_buf_get() functions to use
non-wrapped putters and getters k_fifo_put() and k_fifo_get().
Special handling of net_bufs in k_fifos is no longer needed after commit
3d306c181f, since these actions are now
atomic regardless of any net_buf fragments.
Signed-off-by: Henrik Brix Andersen <henrik@brixandersen.dk>
Some of the fields currently in hid_device_data are constant.
Move them to a const config struct to save some RAM and drop the rest of
the data static initializers to runtime to save some flash as well.
Signed-off-by: Fabio Baltieri <fabiobaltieri@google.com>
Introduce UDC_BUF_POOL_*_DEFINE macros based on NET_BUF_POOL_*_DEFINE
macros but use our own version of alloc and unref callbacks to get
buffers with specific alignment and granularity. Also, do not use ref
callback because it breaks alignment.
Also introduces helper macros for defining and checking UDC
driver-compliant static buffers.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Change updates log level only if UART log backend is enabled in Kconfig
configuration. The log level update is not needed in case logs are
provided over other backend (e.g. RTT).
Change affects both USB stacks.
Signed-off-by: Marek Pieta <Marek.Pieta@nordicsemi.no>
The polling properties are a period in us but are named as "-rate" right
now, which would imply that that's a frequency. Rename them to
"period-us" to make that unambiguous.
Signed-off-by: Fabio Baltieri <fabiobaltieri@google.com>
After the get_report() callback, we need to determine how many bytes the
HID device wrote to the report buffer. Use the callback return value to
do this, and modify the net_buf data length value if get_report was
successful.
Reported-by: Marek Pieta <Marek.Pieta@nordicsemi.no>
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
The ring buffer API is explicitly not thread safe, with users needing to
implement their own locking. As `poll_out` and `fifo_fill` are operating
on the same ringbuffer, these locks are needed.
Signed-off-by: Jordan Yates <jordan@embeint.com>
If CDC ACM uart side has TX enabled when the configuration gets enabled,
depending on fifo state following has to happen:
* if the fifo is empty - trigger TX ready interrupt
* if the fifo is not empty - queue TX data on IN endpoint
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Add function to get string descriptor index and function
to remove linked descriptor from a device. This abstracts
it a bit so that the user does not need to know how it is
handled internally.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Apply the commit 21975231e2
("usb: device: cdc_acm: Prevent recursive logging loop")
to the new CDC ACM implementation.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Add initial HID device support. Unlike the existing HID implementation,
the new implementation uses a devicetree to instantiate a HID device.
To the user, the HID device appears as a normal Zephyr RTOS device.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Since only usbd_class_node contains the class instance data, rename it
to usbd_class_data.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
