Modify driver to process interrupts from upper half ISR then signal a
bottom half ISR thread which performs callback invokation. This ensures
callbacks run in thread context which is necessary for proper operation of
various classes.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
Change CMakeLists.txt to register Zephyr library only if respective
source code is compiled.
Fixes CMake Warning:
No SOURCES given to Zephyr library: drivers__usb__common__buf
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Issue was corrected where transfers waiting on a busy
endpoint were not being serviced. This was causing some transfers
to never complete.
Signed-off-by: Brandon Hurst <brandon.hurst@analog.com>
Move xfer_in_done and xfer_out_done to after xfer_in
and xfer_out. This is because they will need to call
these functions in the next commit to handle deferred
traffic.
Signed-off-by: Brandon Hurst <brandon.hurst@analog.com>
UDC driver was submitting spurious resume events
when receiving DPACT events while the device was not
suspended.
- Change udc_max32_event callback to only respond to
DPACT events when suspended
Signed-off-by: Brandon Hurst <brandon.hurst@analog.com>
MAX32 USB had some functions in ISR context allocating memory &
acquiring mutexes. This commit fixes this by separating the
transfer completion events into ISR-safe / non ISR-safe steps.
Additionally, some apparent swapping of devicetree configuration
for in/out endpoints is resolved.
- Rename UDC_MAX32_EVT_XFER to UDC_MAX32_EVT_START_XFER
- Add MAX32 UDC events for XFER_IN_DONE and XFER_OUT_DONE
- Refactor xfer_in/out callbacks to only submit messages to a
message queue
- Provide xfer_in_Done and xfer_out_done to handle transfer
completions from UDC thread context.
- Add XFER_IN_DONE and XFER_OUT_DONE events to thread handler
- Fix swapping in devicetree macros of num_of_in_eps /
num_of_out_eps and ep_cfg_in / ep_cfg_out
Signed-off-by: Brandon Hurst <brandon.hurst@analog.com>
Rename it to USB buffers so we can use it in host support without
confusion. Keep the UDC_* macros for now, we can deprecate and remove
them later.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Move udc_stm32_clock_enable() and udc_stm32_clock_disable() above their
callers, grouped with other helper functions. This allows removing an ugly
forward declaration at the top of the driver.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
The (driver-internal!) function udc_stm32_ep_flush() was very short and
only called from one function. Inline its contents inside the caller.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
Move udc_stm32_lock(), udc_stm32_unlock() and udc_stm32_init() above all
other functions which implement callbacks of the UDC API (and are already
in the same order as the initializer for udc_stm32_api).
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
Move HAL_PCDEx_SetConnectionState with the other HAL callbacks
(except SETUP/DATA IN/DATA OUT) to make driver easier to browse.
Also move a relevant comment to a more appropriate place and update it.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
Move HAL callbacks for SETUP/DATA IN/DATA OUT events next to (above) the
lower half/worker thread handler for these events. This makes it easier to
follow the event handling logic.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
The macro `hpcd2data` is defined as a wrapper around CONTAINER_OF(), but
was also adding a `;` at the end which is not right.
Remove the trailing `;` in the macro's definition.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
Add a comment documenting why ordinals are used to create the per-instance
symbol names, as this is a very uncommon practice among STM32 drivers.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
Added register bitmask description with low-level abstraction
Signed-off-by: Roman Leonov <jam_roma@yahoo.com>
Co-authored-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
If the clock device (i.e., RCC) failed to initialize, we have bigger
problems than trying to call clock_control_{off,on,configure} on it.
Don't bother checking to save some footprint.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
Modify the STM32 UDC driver to support multiple instances.
While at it, align the device initialization callback name with other
UDC drivers: `udc_stm32_driver_init0` becomes `udc_stm32_driver_preinit`.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
For some reason, the driver was not requesting the HAL to enable SOF even
if the feature was enabled at UDC stack level.
Fix by requesting HAL to enable SOF based on stack configuration.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
The udc_submit_ep_event() does not document this requirement, but the
`net_buf` provided to it must have been removed from the driver's
internal queue before submission. This wasn't noticed because the list
node member of `net_buf` was untouched, but is now causing list corruption
due to 5137439a47 when multiple packets are
enqueued.
Fix by always dequeuing buffers before submitting them to the UDC stack.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
Add DCP auto status support and update the control stage update
sequence to adapt with hardware auto response mode
Signed-off-by: The Nguyen <the.nguyen.yf@renesas.com>
Move all PHY configuration code to dedicated drivers called using a custom
API (we can't instantiate proper devices because some of these "PHYs" are
really used to initialize an interface to a device NOT owned by the SoC,
such as when an ULPI PHY is in use).
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
Refactor utility macros in the STM32 UDC driver, and move some of them to
the STM32 common USB header as they can be useful in other places.
While at it, add new helper macro USB_STM32_NODE_PHY_IS_EMBEDDED_FS.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
As per Zephyr coding guideline #59, "operands shall not be of an
inappropriate essential type". This makes sure boolean variables are
assigned true/false values, not 1/0.
Signed-off-by: Benjamin Cabé <benjamin@zephyrproject.org>
Fix submit condition for non-control transfer. Do submit when any
of the following conditions is met:
1. Transfer buffer (net_buf) is full
2. Last packet size is less than mps
3. Isochronous transfer
USB mass may request C1 for sector size (512 bytes)-aligned and C2
for CBW (13 bytes).
Signed-off-by: Chun-Chieh Li <ccli8@nuvoton.com>
With USBD and HSUSBD re-organized, SoC must have USBD and HSUSBD at
the same time to pass compile. However, M2L31 just has USBD and meets
compile error with HSUSBD unsupported. To avoid adding too many
conditional (#if condition) code for the error, create one dummy
HSUSBD device definition for M2L31 to pass compile.
Signed-off-by: Chun-Chieh Li <ccli8@nuvoton.com>
For HSUSBD, Control transfer with Data OUT stage will break. This is
caused by FIFO flush which cannot be done timely even though in Setup
token interrupt handler. For this, error recovery with FIFO is not
done in Setup token/packet handler and rely on USB reset handler to do
it as catch-all.
Signed-off-by: Chun-Chieh Li <ccli8@nuvoton.com>
Some SoC series allow USB controller to be disabled temporarily
on unplug, and then enabled back on re-plug. Add support for such
configuration option in DT and UDC driver.
Signed-off-by: Chun-Chieh Li <ccli8@nuvoton.com>
Create infrastructure for shared USB common code on STM32 family, and move
the Power Controller configuration logic to common code. This removes some
midly unrelated code from the UDC driver while enabling reuse by a future
UHC driver implementation for STM32.
While at it, clean up the migrated code.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
numaker_usbd_ep_fifo_copy_to_user() never reports errors and always
returns 0.
The error check at the call site is therefore dead code.
Make the function void and drop the unused error handling.
Signed-off-by: Gaetan Perrot <gaetan.perrot@spacecubics.com>
numaker_usbd_ep_fifo_copy_from_user() never reports errors and
always returns 0.
The error check at the call site is therefore dead code.
Make the function void and drop the unused error handling.
Signed-off-by: Gaetan Perrot <gaetan.perrot@spacecubics.com>
Upcoming MDK will make nrf52 errata check functions exist only for
nRF52 and therefore the code will no longer compile for nRF5340. Replace
nrf52 errata checks with new NRF_ERRATA_DYNAMIC_CHECK().
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Replace some manually-defined DT_COMPAT_<> Kconfig macro variables with
their automatically generated counterparts. In most cases, this is
straightforward as the manually defined macro is named identically to the
one generated by the build system.
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
Remove duplicated #include directives within the same
preprocessor scope across the Zephyr tree.
Duplicates inside different #ifdef branches are preserved
as they may be intentional.
Signed-off-by: Sylvio Alves <sylvio.alves@espressif.com>
Add support for Nuvoton NuMaker M55M1X high-speed USB 2.0 device
controller. Compared to M46X HSUSBD, M55M1X HSUSBD introduces some
differences:
- DMA must handle cache coherency because net_buf can be cache-able
- USB suspend interrupt becomes continuous. Aavoid being locked by
this interrupt and forward this message just once
- New register bit HSUSBD_OPER_HISHSEN_Msk, which controls to enable
USB handshake
Signed-off-by: Chun-Chieh Li <ccli8@nuvoton.com>
Add support for Nuvoton NuMaker M46X high-speed USB 2.0 device
controller.
The code is re-organized to implement both usbd and hsusbd in single
source file. Multiple instances of either usbd or hsusbd are supported,
but usbd and hsusbd cannot support simultaneously. This limitation is
for easy implementation with just single source file, assuming that real
application just needs one usb device type.
Signed-off-by: Chun-Chieh Li <ccli8@nuvoton.com>
Previous control transfer can be incomplete and then causes not only
net_buf leak but also logic error. This recycles dangling net_buf for
new clean control transfer.
Signed-off-by: Chun-Chieh Li <ccli8@nuvoton.com>
On Control Out failure, this changes to report error message instead of
assert failure because USB bus error is allowed to happen.
Signed-off-by: Chun-Chieh Li <ccli8@nuvoton.com>
Refine code on ATTR undefined bits:
- Add USBD_ATTR_PWRDN_Msk if it is not defined
- Following BSP USBD driver, add note on BIT(6) for hidden
Signed-off-by: Chun-Chieh Li <ccli8@nuvoton.com>
