Currently there is no way to distinguish between a caller
explicitly asking for a semaphore with a limit that
happens to be `UINT_MAX` and a semaphore that just
has a limit "as large as possible".
Add `K_SEM_MAX_LIMIT`, currently defined to `UINT_MAX`, and akin
to `K_FOREVER` versus just passing some very large wait time.
In addition, the `k_sem_*` APIs were type-confused, where
the internal data structure was `uint32_t`, but the APIs took
and returned `unsigned int`. This changes the underlying data
structure to also use `unsigned int`, as changing the APIs
would be a (potentially) breaking change.
These changes are backwards-compatible, but it is strongly suggested
to take a quick scan for `k_sem_init` and `K_SEM_DEFINE` calls with
`UINT_MAX` (or `UINT32_MAX`) and replace them with `K_SEM_MAX_LIMIT`
where appropriate.
Signed-off-by: James Harris <james.harris@intel.com>
Convert drivers to DEVICE_DEFINE instead of DEVICE_AND_API_INIT
so we can deprecate DEVICE_AND_API_INIT in the future.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
The USB infrastructure currently uses the system work queue for
offloading transfers, CDC-ACM UART transmission/reception, and device
firmware activities. This causes problems when the system work queue
is also used to initiate some activities (such as UART) that normally
complete without requiring an external thread: in that case the USB
infrastructure is prevented from making progress because the system
work queue is blocked waiting for the USB infrastructure to provide
data.
Break the dependency by allowing the USB infrastructure to use a
dedicated work queue which doesn't depend on availability of the
system work queue.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
The z length modifier must be applied to avoid build errors when the
format is verified by the compiler on systems where size_t is larger
than int.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
Both tx_irq_ena and rx_irq_ena flags shall be updated only
by using UART API. Those flags are used to let the device
start sending or receiving data. Changing those flags without
awareness of the user may lead to errors.
Signed-off-by: Emil Obalski <emil.obalski@nordicsemi.no>
Usage of `pool_wait_sem` was incorrect. The semaphore shall be
dependent on context it was called for. In case of many CDC_ACM
devices one could possibly block others. This fixes an issue by
making semaphore a part of the internal CDC_ACM device structure.
Signed-off-by: Emil Obalski <emil.obalski@nordicsemi.no>
Now that device_api attribute is unmodified at runtime, as well as all
the other attributes, it is possible to switch all device driver
instance to be constant.
A coccinelle rule is used for this:
@r_const_dev_1
disable optional_qualifier
@
@@
-struct device *
+const struct device *
@r_const_dev_2
disable optional_qualifier
@
@@
-struct device * const
+const struct device *
Fixes#27399
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
This commit fixes an issue when more than one USB transfer is
reserved for same OUT endpoint.
There are 2 scenarios when this could happen:
- The Host sends SET_CONFIGURATION(1) request twice.
- The Host triggers Suspend->Resume->Configured
event sequence for the device.
USB tranfers are not canceled on SUSPEND event
when the device was not configred previously.
Because of that USB transfer slot is reserved twice
for the same OUT endpoint and lead to shortage of USB
transfer slots quickly.
Without this patch CDC ACM class reserves duplicated USB
transfer slots for one transfer. The sequence of
Suspend->Resume events is genereted alongside with
Configured and the stack will shortly run out of transfer
slots.
If the Host, for some reason, decides to send
SET_CONFIGURATION request twice the same issue is seen.
This patch prevents from reserving additional USB transfer
slots twice.
Signed-off-by: Emil Obalski <emil.obalski@nordicsemi.no>
This is called when the USB host changes the virtual baud rate.
This is used by Arduino style boards to signal that the board should
reset into the bootloader.
Signed-off-by: Michael Hope <mlhx@google.com>
Transfers are cancelled on suspend event, OUT transfers
needs to be re-trigger on resume.
Rework event handling to filter out spurious resume
events and re-trigger transfer if previous event was
a suspend event.
Signed-off-by: Johann Fischer <j.fischer@phytec.de>
USB is sending data from LSB to MSB. Same happens for each field
of the USB setup packet.
This patch convert USB setup packet to proper form when its read
out from the line.
Signed-off-by: Emil Obalski <emil.obalski@nordicsemi.no>
This commit makes cdc class to omit notifications about
interfaces swap that may happen if cdc is configured together
with audio.
Signed-off-by: Johan Carlsson <johan.carlsson@teenage.engineering>
Signed-off-by: Emil Obalski <emil.obalski@nordicsemi.no>
Set initialization level of CDC ACM class to POST_KERNEL.
The proposed OpenThread Network-Coprocessor uses ACM. OpenThread is
started at POST_KERNEL level by the 802.15.4 radio drivers.
Signed-off-by: Markus Becker <markus.becker@tridonic.com>
When the length of the transfer is an exact multiple of IN endpoint MPS
in the ring buffer, transfer one byte less to avoid zero-length packet.
Otherwise the application running on the host may conclude that there
is no more data to be received (i.e. the transaction has completed),
hence not triggering another I/O Request Packet (IRP).
Fixes#21713.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
In case the ring buffer is empty, ring_buf_get_claim() returns a zero
length. Exit the function in that case as calling usb_transfer() with a
zero length will send a ZLP.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Currently the cdc_acm implementation pass the data to usb_transfer() by
a chunk of IN endpoint MPS. This has 2 drawbacks:
- at higher throughput, each transfer needs 2 packets due to the need of
an extra ZLP;
- a temporary buffer of size USB MPS is needed.
This patch improves the memory consumption and performances by passing
the ring buffer directly to usb_transfer(). It only has a small
performance degradation when the ring buffer wraps and less than a IN
endpoint MPS has to be sent.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
This commit renames two defines:
CS_INTERFACE -> USB_CS_INTERFACE_DESC
CS_ENDPOINT -> USB_CS_ENDPOINT_DESC
in order to match current naming convention when it comes
to descriptors fields.
All relevant files are updated to match renamed macros.
Signed-off-by: Emil Obalski <emil.obalski@nordicsemi.no>
Several macros were documented as deprecated but lacked the
infrastructure to produce deprecation warnings. Add the deprecation
marker, and fix the in-tree references to the deprecated spellings.
Note that one non-deprecated macro should have been deprecated, and
is, referring to a newly added line control bit.
Signed-off-by: Peter A. Bigot <pab@pabigot.com>
Update state after RESUME event.
Use LOG_INF for important events and log a
warning if the data is discarded.
Fixes#17488
Signed-off-by: Johann Fischer <j.fischer@phytec.de>
move misc/byteorder.h to sys/byteorder.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move ring_buffer.h to sys/ring_buffer.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move uart.h to drivers/uart.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
USBD_CFG_DATA_DEFINE macro has not consider that a class
could have more than one set of usb_cfg_data struct.
If a class has more than one set of usb_cfg_data
then they should be sorted the same way like by
USBD_DEVICE_DESCR_DEFINE macro.
Fixes: #16240
Signed-off-by: Johann Fischer <j.fischer@phytec.de>
Remove duplicated execution path for composite configuration, USB
device stack initialization is done inside stack for both cases.
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
In order to unify the legacy and composite code, move the class and
vendor request handler buffer into the USB device code, just like in
composite mode. The option is renamed from USB_COMPOSITE_BUFFER_SIZE
into USB_REQUEST_BUFFER_SIZE and also replaces the USB_DFU_MAX_XFER_SIZE
and USB_HID_MAX_PAYLOAD_SIZE options.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Interface Association descriptor has to be used with Windows 7.
Add CONFIG_CDC_ACM_IAD option to force its usage, disabled by default.
Signed-off-by: Marcin Szymczyk <Marcin.Szymczyk@nordicsemi.no>
Use new method for defining USB CDC ACM devices using UTIL_LISTIFY and
device count from menuconfig.
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Added bBreak, bRingSignal, bFraming, bParity and bOverRun
handling in cdc_acm_line_ctrl_set.
Reference: Chapter 6.5.4 of Universal Serial Bus Communications Class
Subclass Specification for PSTN Devices rev 1.2
Signed-off-by: Marcin Szymczyk <Marcin.Szymczyk@nordicsemi.no>
Merge cb_usb_status_composite and cb_usb_status and use common
forward_status_cb for both composite and normal devices.
Fixes#14882
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Use ringbuf library for handling RX CDC ACM path. Remove old code
artifacts handling specific hardware.
Fixes#14288
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Add new status callback with usb_cfg_data parameter to be able to
identify instance callback.
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Add interface parameter to interface configuration callback to be able
to configure several instances of the same class.
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
