The STM32L4x2 SoCs need to control the isolation of the USB features
from VDDUSB. This is done through the PWR_CR2 bit USV, however the
current code checks for the PWR_CR2_PVME1 bit instead, which is only
available on Cat. 3 devices. This bug is also present int the HAL and
likely copied from there.
Replace the check by PWR_CR2_USV instead.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
This patch fixes some bugs found during testing with testusb from
linux kernel.
An unresolved issue is that the stack is not fast enough
to stall (if necessary) the control endpoint during Setup Stage.
This might require a API change so that the usb device stack
can explicit allow the driver to resume token processing.
Signed-off-by: Johann Fischer <j.fischer@phytec.de>
Add a check to raise an error if SYSCFG is disabled,
before doing the pin remaping for F0 SoCs on QFN28
and TSSOP20 packages.
Signed-off-by: Yannis Damigos <giannis.damigos@gmail.com>
In STM32L0 series, HSI48 requires VREFINT and its buffer
with 48 MHz RC to be enabled. This patch enables
VREFINT reference for HSI48 oscillator.
Signed-off-by: Ilya Tagunov <tagunil@gmail.com>
Signed-off-by: Yannis Damigos <giannis.damigos@gmail.com>
USB LL API provides the EP_TYPE_* defines. STM32Cube does not
provide USB LL API for STM32L0 family. Map EP_TYPE_* defines
to PCD_EP_TYPE_* defines.
Signed-off-by: Yannis Damigos <giannis.damigos@gmail.com>
Add usb disconnect pin support.
Some boards use a GPIO pin to controll a transistor,
which drives the pull-up resistor on USB DP pin.
Signed-off-by: Yannis Damigos <giannis.damigos@gmail.com>
Fix issue with Coverity using if / else instead of switch, though it
does not affect code.
Fixes: #7257
CID: 185399
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Because of incorrectly set MPS we always got 0 returned for WRITE
transfers and ZLP were always generated for every packet.
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
This event is no longer exposed for the USBD peripheral,
and the newest MDK version (and consequently new nrfx HAL)
does not contain its definition and related bitfields.
Signed-off-by: Andrzej Głąbek <andrzej.glabek@nordicsemi.no>
For ISOOUT transaction, check if the buffer is empty before starting
the DMA and for ISOIN case check if the buffer is available to start
the DMA for IN transaction. This handling was swapped. So, handle it
accordingly.
Signed-off-by: Sundar Subramaniyan <sundar.subramaniyan@gmail.com>
Problem:
cdc_acm sample outputs junk characters to the console when any key
is pressed. It is supposed to echo the same character as the input.
Analysis:
nRF52840 USB driver always reads 64 bytes from the USB peripheral's
local buffer irrespective of the data length. Even when 1 byte
is sent by the host, the usb driver receives as 64 bytes.
Since the maximum packet size value had been used when setting up
the EasyDMA's MAXCNT register for OUT transactions, DMA transffered
the entire 64 bytes of it's local buffer into the data RAM
irrespective of the OUT data length, containing garbage at the end
of useful payload. For some applications like hci_usb the extra
bytes were being ignored. But for cdc_acm, the entire 64 bytes
is getting echoed back displaying garbled characters on the terminal.
Fix:
For OUT endpoints, the amount of data received by USBD to its local
buffer should be read using SIZE.EPOUT[x] registers in the case of
BULK and INTERRUPT endpoints and SIZE.ISOOUT register for ISOOUT
endpoint. This value should be set to the EasyDMA maxcnt register
to DMA into RAM buffer. Fix this issue for Bulk, Interrupt and
ISOOUT endpoints.
Signed-off-by: Sundar Subramaniyan <sundar.subramaniyan@gmail.com>
For now tx-fifo sizes are not configured (cf usb_dw_set_fifo).
Default fifo size is 136 bytes and sending more than 136 bytes
makes the data split into incorrect chunk size.
This patch prevents this by reducing available fifo size to a
multiple of the endpoint max packet size.
Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
* Add USB device controller driver for nRF52840 SoC.
* Supports only Bulk and Interrupt endpoints for now.
* ISO endpoint support is WIP.
What's tested:
* BTUSB - LE Scanning and connection.
Known issues in btusb tests:
* Observed momentary delays during LE scanning. This is possibly
due to hci_usb's way of calling the usb_write().
* Sometimes, when LE scan stop command is issued, response is not
sent and as a result USB stops functioning. This is also related
to the way btusb is handling the write(). For some reason, the
btusb class handler seems to issue another data packet write before
the USB control transaction is complete.
Miscellaneous:
* Lots of TODO items to discuss and fix.
Origin: Original
Signed-off-by: Sundar Subramaniyan <sundar.subramaniyan@gmail.com>
Transfer is an important concept of USB specification.
During a transfer, several packets can be transmitted.
Today the usb API only provides ep_read and ep_write
functions but the transfer concept is missing.
This is typically ok for basic drivers which don't have
to take care of how data is transfered/splitted.
However there are some drivers like CDC EEM, in which
transfer concept is important for packet management.
Moreover, current ep-write and ep_read method have a
different implementation in usb_dc_dw and usb_dc_stm32
device drivers. For example usb_dc_dw supports only
1-data-packet transfers due to its implementation.
This can increase latency and reduce performance.
I think this is something we need to fix/improve by
implementing better transfer management.
This patch introduces usb_dc_ ep_transfer method which
can be used to configure IN/OUT transfers. This allows
to configure and request different transfer sizes and
should prevent usage of the current stm32 temporary
buffer.
This method has asynchronous and synchronous mode.
Synchronous mode waits for transfer completion before
returning. Asynchronouse mode (irq safe) configures
the transfer and returns immediately, the provided
callback will be then called on transfer completion.
This also update ep_write and ep_read stm32 implementation
to use this new method but keep their behavior unchanged
for legacy reasons.
Note that for now this method is local to stm32 device
driver, however the goal would be to expose this function
as a new USB device driver API method so that class
drivers use it. This will request same implementation in
the usb_dc_dw_driver.
Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
usb mass example need usb_dc_ep_read_wait/continue API.
test usb mass storage with RAM DISK on stm32f4 series.
Signed-off-by: qianfan Zhao <qianfanguijin@163.com>
Use "select USE_STM32_LL_USB/USE_STM32_HAL_PCD/USE_STM32_HAL_PCD_EX"
to select the needed STM32 HAL files, instead of editing
ext/hal/st/stm32cube/CMakeLists.txt
Signed-off-by: Erwin Rol <erwin@erwinrol.com>
During large USB transfer it's pretty common to call ep_write whereas
the previous TX transfer is not achieved and so the TX FIFO space is
not available. Sleeping 20ms in this case introduce a relatively high
latency and reduce the throughput.
This can be observed when pinging the board with CDC-ECM net class.
ping reply is split into 2 USB TX transfer, the second one is only
triggered after 20ms, making ping latency > 20ms.
To fix this, just continuously read the FIFO availabilty and fire TX
as soon as possible.
Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
In the same way as dw driver, check that FIFO is empty before
writing any new data. This patch introduces a boolean semaphore
which is requested before any new TX transfer and released on
transfer completion.
This fixes usb-ecm support on 96b_carbon board.
Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
Introducing CMake is an important step in a larger effort to make
Zephyr easy to use for application developers working on different
platforms with different development environment needs.
Simplified, this change retains Kconfig as-is, and replaces all
Makefiles with CMakeLists.txt. The DSL-like Make language that KBuild
offers is replaced by a set of CMake extentions. These extentions have
either provided simple one-to-one translations of KBuild features or
introduced new concepts that replace KBuild concepts.
This is a breaking change for existing test infrastructure and build
scripts that are maintained out-of-tree. But for FW itself, no porting
should be necessary.
For users that just want to continue their work with minimal
disruption the following should suffice:
Install CMake 3.8.2+
Port any out-of-tree Makefiles to CMake.
Learn the absolute minimum about the new command line interface:
$ cd samples/hello_world
$ mkdir build && cd build
$ cmake -DBOARD=nrf52_pca10040 ..
$ cd build
$ make
PR: zephyrproject-rtos#4692
docs: http://docs.zephyrproject.org/getting_started/getting_started.html
Signed-off-by: Sebastian Boe <sebastian.boe@nordicsemi.no>
Adding sleep before TX FIFO flash fixes splitting networking packets
sent over USB endpoints making ECM broken since there is no flow
control other then frame sizes.
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
The situation when FIFO is not empty is not a bug and it is spamming
console when only bugs are enabled.
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
The Designware FIFO is filled in units of 32 bit words, but the buffer
we are passed is not guaranteed to be a multiple of 4 bytes long, nor
aligned on a 4-byte boundary. So in theory we are reading 0-3 bytes
of unused garbage from the end of the array.
That's currently benign on supported platforms with this hardware,
which all support misaligned reads. But not all do. And the incoming
arrival of memory protection opens the possibility that those extra
bytes would cross a protection boundary and cause a crash or security
bug.
Do this right.
(Note that this is fixed to little endian byte order. The Designware
databook is frustratingly silent on the endianness it expects, but
existing hardware I can see is definitely LE and I see a few spots in
the Linux dwc2 driver that likewise assume LE).
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The designware hardware in dedicated FIFO mode (which is all we
support right now for lack of shared-FIFO hardware) has one hardware
FIFO per IN (i.e. transmit) endpoint. But it doesn't assign them on
its own, it's the drivers responsibility to populate the TxFNum field
of the DIEPCTL registers with integer indices corresponding to the
desired FIFO.
We weren't doing that, which meant that all IN endpoints were sharing
the same FIFO zero which is supposed to be dedicated to EP0 control
transfers. The net effect is that sometimes outbound transfers would
be corrupted, showing data from the wrong endpoint. More often that
not this would leak from control transfers over to the
higher-bandwidth bulk endpoints of the application, but occasionally
you'd see a control transfer itself get borked and the USB device
would glitch.
Get this right and set the FIFOs properly.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
There is no particular reason this spot in usb_dw_tx() cannot be
reached by racing threads on the same endpoint, though existing API
usage in the tree is all unithreaded. The FIFO state read at the top
of the function must still be true at the bottom or else the packet
byte count will be corrupt.
Also, as described in an existing comment, the databook has some
scary-sounding warnings about access to the registers during FIFO
operations, even if they "should" be on separate endpoints and
unrelated.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Aurelien Jarno