The driver's endpoint configuration must be per instance.
Assisted-by: Claude:claude-sonnet-5
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Introduces K_MEM_SLAB_DEFINE_TYPE() and K_MEM_SLAB_DEFINE_STATIC_TYPE()
helpers to allow the user to declare slabs for types without having to
manually ensure the alignment is correct.
Manual slab alignment was very error-prone and this change fixes several
instances of misalignment that would be trapped by the undefined
behavior sanitizer when running on 64-bit targets.
Signed-off-by: Egill Sigurdur <egill@egill.xyz>
Function udc_buf_get_all() was intended to be a helper to remove all
requests from endpoint FIFO. While for just freeing all queue the
it may be argued that there may be some doubtful simplicity argument,
merging multiple submitted transfers into one is just enforcing
unnecessary complexity on class implementations.
At general level, every submitted (enqueued) request should get
corresponding completion (request callback) call. UDC drivers were
violating this sensible behavior when dequeuing (cancelling) requests by
merging all submitted requests into one.
Remove udc_buf_get_all() and replace all uses with simple loops. For
most classes (that submit just one request for an endpoint at a time)
this has no functional difference. For classes that implement double
buffering this simplifies completion handling.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Allocating buffer in response to control transfer is part of processing
and therefore should not be done in UDC driver but rather in the stack.
Simplify UDC driver design by moving all control transfer buffer
allocations and processing to USB stack.
New control transfer handling flow is as follows:
1. USB stack allocates and queues buffer for Setup stage. This informs
UDC driver that USB stack is ready to process new transfer.
2. UDC driver completes enqueued Setup buffer. SETUP data may have
been received by device before Setup buffer was enqueued.
If multiple SETUP data was received, UDC driver must complete
request using last received SETUP data.
3. USB stack processes request. USB stack is responsible for:
* Stalling control endpoint when request cannot be processed
* Allocating and enqueueing Data stage buffer if necessary
- Depending on enqueued endpoint (IN/OUT), UDC driver is
expected to complete the transfer. If host sends new SETUP
data for any reason (e.g. timeout), then Data stage buffer
must be completed with -ECONNRESET code.
- UDC driver may postpone processing Data IN until USB stack
enqueues new Setup stage buffer.
* Handling status stage if applicable
- If Data IN was enqueued, stack immediately allocates and
enqueues Status OUT stage buffer.
- If Data OUT was enqueued, stack waits for Data OUT buffer
completion before doing any further processing.
- If control transfer handling fails, stack is expected to
STALL control endpoint.
* Allocating and enqueuing Setup buffer.
- This informs UDC driver that stack has finished processing
control transfer. UDC driver may choose to start processing
Data IN and/or Status OUT only after new Setup buffer is
enqueued.
4. UDC driver must fail (complete with -ECONNRESET) any enqueued and
not completed Data and Status buffers if host sends new SETUP data.
UDC driver must be able to buffer last received SETUP data until
USB stack is ready (enqueues Setup buffer).
5. UDC driver must complete all Data and Status buffers before it
completes Setup buffer.
6. UDC driver may keep ownership of Setup, Data and/or Status buffers
across USB bus resets. USB stack does not attempt to dequeue any
control transfer buffer it enqueued.
This approach implicitly synchronizes UDC driver against USB stack,
which ensures that only one set of Setup/Data/Status buffers is
allocated at a time.
Another advantage of the rework is drawing a clear line on buffer
ownership. The buffer responsibilities are as follows:
* USB stack is the only entity that both allocates and frees the
buffers. Only USB stack is allowed to set "setup", "data" and
"status" fields in struct udc_buf_info.
* UDC takes ownership of buffers handed to it in udc_ep_enqueue().
* UDC releases buffer ownership by calling udc_submit_ep_event().
Because there is just a single place where the buffer ownership changes,
and all buffers must go USB stack (alloc) -> UDC (perform requests on
the bus) -> USB stack (free) route it would be possible in the future to
implement a "tap" similar to Linux usbmon that would ease debugging.
This commit significantly changes how USB stack communicates with UDC
drivers. It was decided that supporting both the old and new model
simultaneously would require way too much effort. Therefore all UDC
drivers were reworked. Following people worked on driver rework:
* Tomasz Moń - ambiq, dwc2, kinetis, mcux ip3511, nrf, numaker,
renesas ra, rpi pico, smartbond, virtual
* Mathieu Choplain - stm32
* Mark Wang - mcux ehci
* Johann Fischer - sam0, stm32
* Ren Chen - it82xx2
* Brandon Hurst - max32
* Gerson Fernando Budke - sam udp, sam usbc, sam usbhs
Co-authored-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
Signed-off-by: Mathieu Choplain <mathieu.choplain-ext@st.com>
Co-authored-by: Johann Fischer <johann.fischer@nordicsemi.no>
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Co-authored-by: Mark Wang <yichang.wang@nxp.com>
Signed-off-by: Mark Wang <yichang.wang@nxp.com>
Co-authored-by: Ren Chen <Ren.Chen@ite.com.tw>
Signed-off-by: Ren Chen <Ren.Chen@ite.com.tw>
Co-authored-by: Brandon Hurst <brandon.hurst@analog.com>
Signed-off-by: Brandon Hurst <brandon.hurst@analog.com>
Co-authored-by: Gerson Fernando Budke <nandojve@gmail.com>
Signed-off-by: Gerson Fernando Budke <nandojve@gmail.com>
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
UDC API passes struct udc_ep_config to all functions. Some UDC functions
were using endpoint config structure while some were using device and
endpoint number. This API inconsistency led to completely unnecessary
endpoint structure lookups. Remove unnecessary lookups by using endpoint
config structure pointer where it makes sense.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
USB stack does not check api->lock() and api->unlock() return value and
all UDC drivers block without timeout in its lock() and unlock() api
implementations. There is no realistic way to handle lock() and unlock()
errors without making USB device stack API unnecessarily complex.
Remove the return type from lock() and unlock() to make it clear that
the functions must not fail.
Signed-off-by: Tomasz Moń <tomasz.mon@nordicsemi.no>
Instead, use the helper to get the size field from the MPS value.
MPS value may contain information about additional transaction
per microframe (bits 12..11).
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Modify the signature of the k_mem_slab_free() function with a new one,
replacing the old void **mem with void *mem as a parameter.
The following function:
void k_mem_slab_free(struct k_mem_slab *slab, void **mem);
has the wrong signature. mem is only used as a regular pointer, so there
is no need to use a double-pointer. The correct signature should be:
void k_mem_slab_free(struct k_mem_slab *slab, void *mem);
The issue with the current signature, although functional, is that it is
extremely confusing. I myself, a veteran Zephyr developer, was confused
by this parameter when looking at it recently.
All in-tree uses of the function have been adapted.
Fixes#61888.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Only UVB_EVT_REQUEST type passes the pkt argument.
This was overlooked in the last refactoring and
resulted in a zero pointer dereference.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Since we have a dedicated function for UDC_EVT_EP_REQUEST type
events, we can now simplify udc_submit_event() and remove
buf parameter.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Add helper function to send UDC endpoint event to a higher level
that takes fewer arguments.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Use proper format for size_t type eliminating warnings of type:
...
error: format '%u' expects argument of type 'unsigned int', but
argument 2 has type 'size_t' {aka 'long unsigned int'}
[-Werror=format=]
...
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Commit 00adb2a539 ("drivers: udc: remove no more required pending
state flag") removed state parameter from udc_buf_peek and udc_buf_get.
Adapt udc virtual driver so the code compiles.
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
New capability is to be used by the stack to adjust the corresponding
field in the device descriptor.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
udc_ep_flush() is not implemented by the driver and is not
used by the USB device stack. Remove it for less confusion
when porting drivers.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Add support for virtual USB device controller intended for use
by virtual bus and virtual UHC controllers. This driver is not
an emulation of any real host controller.
The driver has initial support for handling control and bulk
transfers.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
