Squashed commit of the following:
Change all calls to usleep() in the OS proper to calls to nxsig_usleep()
sched/signal: Add a new OS internal function nxsig_usleep() that is functionally equivalent to usleep() but does not cause a cancellaption point and does not modify the errno variable.
sched/signal: Add a new OS internal function nxsig_sleep() that is functionally equivalent to sleep() but does not cause a cancellaption point.
This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
README
======
This README discusses the port of NuttX to the STMicro B-L475E-IOT01A
Discovery kit powered by STM32L475VG Cortex-M4. This board targets IoT
nodes with a choice of connectivity options including WiFi, Bluetooth LE,
NFC, and sub-GHZ RF at 868 or 915 MHz, as well as a long list of various
environmental sensors.
Contents
========
o STATUS
o Board Features
o LEDs and Buttons
o Serial Console
o Configurations
STATUS
======
o 2017-06-10: I have no hardware in hand and I am not sure that I will
even pursue this port. This README is really no more than a thought
experiment at the present time.
A few days ago, I did add support for the STM32L4x5 MCU family to
NuttX. But no work has yet been done for this board port other
than writing this README file.
o 2017-06-13: I just learned that development boards will not be
available for another month.
Board Features
==============
B-L475E-IOT01A Discovery kit key features and specifications:
o MCU: STM32L475 Series MCU based on ARM Cortex-M4 core with 1 MB
Flash memory, 128 KB SRAM
o Storage: 64 Mbit (8MB) Quad-SPI Flash memory (Macronix)
o Connectivity:
- Bluetooth 4.1 LE module (SPBTLE-RF)
- Sub-GHz (868 or 915 MHz) low-power-programmable RF module (SPSGRF-868
or SPSGRF-915)
- Wi-Fi module based on Inventek ISM43362-M3G-L44 (802.11 b/g/n
compliant)
- Dynamic NFC tag based on M24SR with its printed NFC antenna
o Sensors:
- 2x digital omni-directional microphones (MP34DT01)
- Capacitive digital sensor for relative humidity and temperature
(HTS221)
- 3-axis magnetometer (LIS3MDL)
- 3D accelerometer and 3D gyroscope (LSM6DSL)
- 260-1260 hPa absolute digital output barometer (LPS22HB)
- Time-of-Flight and gesture-detection sensor (VL53L0X
o USB – 1x micro USB OTG port (Full speed)
o Expansion – Arduino UNO V3 headers, PMOD header
o Debugging – On-board ST-LINK/V2-1 debugger/programmer with USB
re-enumeration capability: mass storage, virtual COM port and debug
port
o Misc – 2 push-buttons (user and reset)
o Power Supply – 5V via ST LINK USB VBUS or external sources
The board supports ARM mbed online compiler, but can also be programmed
using IDEs such as IAR, Keil, and GCC-based IDEs. STMicro also provides
HAL libraries and code samples as part of the STM32Cube Package, as well
as X-CUBE-AWS expansion software to connect to the Amazon Web Services
(AWS) IoT platform.
NOTES:
1. The board usese Wi-Fi® module Inventek ISM43362-M3G-L44 (802.11 b/g/n
compliant), which consists of BCM43362 and STM32F205 host processor
that has a standard SPI or UART interface capability. It means you
will only use AT command to talk with Wi-Fi® module by SPI. All the
tcp/ip stack is built-in STM32F205 in Wi-Fi® module.
This cannot integrate cleanly with the NuttX network stack. A
USERSOCK option was recently added that would permit implementation
of the Inventek support in an applications. But that would then
preclude the 6LoWPAN integration into IPv6.
2. The board uses Bluetooth® V4.1 module (SPBTLE-RF), which has built-in
BLE stack. Similar with wifi, you only use simple AT command to talk
with this BLE module.
3. STMicro provides contiki 6lowpan for mesh.
http://www.st.com/en/embedded-software/osxcontiki6lp.html but mesh
network is not popular in the market, star network is the mainstream
for its simplicity and robustness.
LEDs and Buttons
================
The black button B1 located on top side is the reset of the STM32L475VGT6.
The blue button B1 located top side is available to be used as a digital
input or as alternate function Wake-up. When the button is depressed the logic state is "0", otherwise the logic state is "1".
Two green LEDs (LD1 and LD2), located on the top side are available for the user. To light a LED a high logic state "1" should be written in the corresponding GPIO.
Reference Color Name Comment
B2 blue Wake-up Alternate function Wake-up
LD1 green LED1 PA5 (alternate with ARD.D13)
LD2 green LED2 PB14
These LEDs are not used by the board port unless CONFIG_ARCH_LEDS is
selected. In that case, the usage by the board port is defined in
include/board.h and src/lpc31_leds.c. The LEDs are used to encode
OS-related events as follows:
SYMBOL Meaning LED state
LED2 LED1
------------------- ----------------------- -------- --------
LED_STARTED NuttX has been started OFF OFF
LED_HEAPALLOCATE Heap has been allocated OFF OFF
LED_IRQSENABLED Interrupts enabled OFF OFF
LED_STACKCREATED Idle stack created ON OFF
LED_INIRQ In an interrupt N/C N/C
LED_SIGNAL In a signal handler N/C N/C
LED_ASSERTION An assertion failed N/C N/C
LED_PANIC The system has crashed N/C Blinking
LED_IDLE MCU is is sleep mode Not used
Thus if LED2 is statically on, NuttX has successfully booted and is,
apparently, running normmally. If LED1 is flashing at approximately
2Hz, then a fatal error has been detected and the system has halted.
NOTE: That LED2 is not used after completion of booting and may
be used by other board-specific logic.
Of course, if CONFIG_ARCH_LEDS is not selected, then both LEDs are
available for use by other logic.
Serial Console
==============
Arduino Serial Shield
---------------------
An TLL-to-RS232 Converter shield may be used with UART4:
UART4:
-------------- ---------------- ------------------
STM32L475VGTx Board Signal Arduino Connector
-------------- ---------------- ------------------
UART4_RX PA1 ARD.D0-UART4_RX CN3 pin1 RX/D0
UART4_TX PA0 ARD.D1-UART4_TX CN3 pin2 TX/D1
-------------- ---------------- ------------------
Virtual COM Port
----------------
The serial interface USART1 is directly available as a virtual COM port
of the PC connected to the ST-LINK/V2-1 USB connector CN7.
USART1:
-------------- ---------------- --------------
STM32L475VGTx Board Signal STM32F103CBT6
-------------- ---------------- --------------
USART1_TX PB6 ST-LINK-UART1_TX USART2_RX PA3
UAART1_RX PB7 ST-LINK-UART1_RX USART2_TX PA2
-------------- ---------------- --------------
The virtual COM port settings are configured as: 115200 b/s, 8 bits data,
no parity, 1 stop bit, no flow control.
Other Options
-------------
USART2 - Available on CN10 if solder bridges closed.
-------------- ---------------- ---------------------------
STM32L475VGTx Board Signal PMOD / Solder Bridges
-------------- ---------------- ---------------------------
USART2_RX PD4 PMOD-UART2_RX CN10 pin1 or 2 (SB12, SB14)
USART2_TX PD5 PMOD-UART2_TX CN10 pin2 TX/D1 (SB20)
-------------- ---------------- ---------------------------
USART3 - Dedicated to ISM43362-M3G-L44 Serial-to-Wifi Module.
-------------- ---------------- ------------------
STM32L475VGTx Board Signal Arduino Connector
-------------- ---------------- ------------------
USART3_RX PD9 INTERNAL-UART3_RX CN3 pin1 RX/D0
USART3_TX PD8 INTERNAL-UART3_TX CN3 pin2 TX/D1
-------------- ---------------- ------------------
Configurations
==============
Information Common to All Configurations
----------------------------------------
Each B-L475E-IOT01A configuration is maintained in a sub-directory and
can be selected as follow:
tools/configure.sh [-l|c|u|n] /b-l475e-iot01a/<subdir>
Where:
-l selects the Linux (l) host environment. The [-c|u|n] options
select one of the Windows environments. Default: Use host setup
in the defconfig file
[-c|u|n] selects the Windows host and a Windows environment: Cygwin (c),
Ubuntu under Windows 10 (u), or Windows native (n). Default Cygwin
Before building, make sure that:
1. The PATH environment variable include the correct path to the
directory than holds your toolchain binaries.
2. Check the .config file. Make sure that the configuration is set for
your build platform (e.g., Linux vs. Windows) and that the toolchain
is set for the toolchain type you are using.
The <subdir> that is provided above as an argument to the
tools/configure.sh must be is one of those listed below.
And then build NuttX by simply typing the following. At the conclusion of
the make, the nuttx binary will reside in an ELF file called, simply,
nuttx.
make
NOTES:
1. These configurations use the mconf-based configuration tool. To
change any of these configurations using that tool, you should:
a. Build and install the kconfig-mconf tool. See nuttx/README.txt
see additional README.txt files in the NuttX tools repository.
b. Execute 'make menuconfig' in nuttx/ in order to start the
reconfiguration process.
2. Unless stated otherwise, all configurations generate console
output on USART1 (i.e., for ST-Link Virtual COM port). The
relevant configuration settings are listed below:
CONFIG_STM32_USART1=y
CONFIG_STM32_USART1_SERIALDRIVER=y
CONFIG_STM32_USART=y
CONFIG_USART1_SERIALDRIVER=y
CONFIG_USART1_SERIAL_CONSOLE=y
CONFIG_USART1_RXBUFSIZE=256
CONFIG_USART1_TXBUFSIZE=256
CONFIG_USART1_BAUD=115200
CONFIG_USART1_BITS=8
CONFIG_USART1_PARITY=0
CONFIG_USART1_2STOP=0
3. All of these configurations are set up to build under Windows using the
"GNU Tools for ARM Embedded Processors" that is maintained by ARM
(unless stated otherwise in the description of the configuration).
https://developer.arm.com/open-source/gnu-toolchain/gnu-rm
That toolchain selection can easily be reconfigured using
'make menuconfig'. Here are the relevant current settings:
Build Setup:
CONFIG_HOST_WINDOWS=y : Window environment
CONFIG_WINDOWS_CYGWIN=y : Cywin under Windows
System Type -> Toolchain:
CONFIG_ARMV7M_TOOLCHAIN_GNU_EABIW=y : GNU ARM EABI toolchain
Configuration sub-directories
-----------------------------
nsh:
Configures the NuttShell (nsh) located at examples/nsh. This
configuration is focused on low level, command-line driver testing.
spirit-6lowpan
This is another version of nsh that is similar to the above 'nsh'
configuration but is focused on testing the Spirit1 integration with
the 6LoWPAN network stack. It supports point-to-point, 6LoWPAN
communications between two b-l47e-iot01a boards. Additional differences
from the 'nsh" configuration are summarized below:
NOTES:
1. You must must have two b-l475e-iot01a boards.
2. IPv6 networking is enabled with TCP/IP, UDP, 6LoWPAN, and NSH
Telnet support.
3. Configuration instructions: NSH does not configuration or
bring up the network. Currently that must be done manually.
The configurations steps are:
a) Assign a unique 8-bit node address to the Spirit1 board in the
WPAN:
nsh> ifconfig wpan0 hw 37
Where 37 the address is an example. It should be different for
each radio, but in the the range 1..ed and ef..fe (ee and ff are
the reserved for multicast and broadcast addresses, respectively.
Zero is a valid address but not recommeded).
b) Bring each the network up on each board in the WPAN:
nsh> ifup wpan0
You can entry nsh> ifconfig to see if the node address and
derived IPv4 are set correctly (the IPv6 address will not be
determined until the network is UP).
4. examples/udp is enabled. This will allow two Spirit1 nodes to
exchange UDP packets. Basic instructions:
On the server node:
nsh> ifconfig
nsh> udpserver &
The ifconfig command will show the IP address of the server. Then on
the client node use this IP address to start the client:
nsh> udpclient <server-ip> &
Where <server-ip> is the IP address of the server that you got above.
NOTE: There is no way to stop the UDP test once it has been started
other than by resetting the board.
5. examples/nettest is enabled. This will allow two Spirit1 nodes to
exchange TCP packets. Basic instructions:
On the server node:
nsh> ifconfig
nsh> tcpserver &
The ifconfig command will show the IP address of the server. Then on
the client node use this IP address to start the client:
nsh> tcpclient <server-ip> &
Where <server-ip> is the IP address of the server that you got above.
NOTE: Unlike the UDP test, there the TCP test will terminate
automatically when the packet exchange is complete.
6. The NSH Telnet deamon (server) is enabled. However, it cannot be
started automatically. Rather, it must be started AFTER the network
has been brought up using the NSH 'telnetd' command. You would want
to start the Telent daemon only if you want the node to serve Telent
connections to an NSH shell on the node.
nsh> ifconfig
nsh> telnetd
Note the 'ifconfig' is executed to get the IP address of the node.
This address derives from the 8-bit node address that was assigned
when the node was configured.
7. This configuration also includes the Telnet client program. This
will allow you to execute a NSH one a node from the command line on
a different node. Like:
nsh> telnet <server-ip>
Where <server-ip> is the IP address of the server that you got for
the ifconfig commna on the remote node. Once the telnet session
has been started, you can end the session with:
nsh> exit
STATUS:
2017-08-01: Testing began. The Spirit1 no configurations with no
errors, but there are no tests yet in place to exercise it.
2017-08-02: The nettest, udp, telnet test programs were added.
2017-08-03: Successfully exchanging packets, but there there are
issues with address filtering, CRC calculation, and data integrity
(like bad UDP checksums). Lot's more to be done!
2017-08-04: Fixed some of the address filtering issues: In Basic
packets, need to force the Spirit to send the destination address.
This fixes address filtering. But...
Converted to STack vs Basic packets. We need to do this because
the Basic packets do not provide the source node address. Now
correctly gets the source node address and uncompresses the source
IP address.
In addition, to avoid packet loss due to data overrun, I enabled
the AutoAck, TX retries, the RX timeout options.
With these changes (along with other, significant bugfixes), both
the UDP test is now fully functional. CRC filtering still must be
disabled.
2017-08-05: Add the Telnet client problem. Verified HC06 tests with
no debug output; verified Telnet seessions between two spirit nodes.
At this point everything seems functional, but somewhat reliable.
Sometimes things seem to initialize in a bad state.
2017-08-06: Reducing the FIFO to 94 bytes fixed the problem with the
2 byte CRC.
Test Matrix:
The following configurations have been tested successfully (with
CRC disabled):
TEST DATE
COMPRESSION UDP TCP Telnet
----------- ----- ----- ------
hc06 08/04 08/04 08/05 (CRC disabled)
hc1 --- --- ---
ipv6 --- --- ---
----------- ----- ----- ------
Other configuration options have not been specifically addressed
(such non-compressable ports, non-MAC based IPv6 addresses, etc.)
spirit-starhub and spirit-starpoint
These two configurations implement hub and and star endpoint in a
star topology. Both configurations derive from the spirit-6lowpan
configuration and most of the notes there apply here as well.
1. You must must have three b-l475e-iot01a boards in order to run
these tests: One that serves as the star hub and at least two
star endpoints.
2. The star point configuration differs from the primarily in the
spirit-6lowpan in following is also set:
CONFIG_NET_STAR=y
CONFIG_NET_STARPOINT=y
The CONFIG_NET_STARPOINT selection informs the endpoint that it
must send all frames to the hub of the star, rather than directly
to the recipient.
The star hub configuration, on the other hand, differs from the
spirit-6lowpan in these fundamental ways:
CONFIG_NET_STAR=y
CONFIG_NET_STARHUB=y
CONFIG_NET_IPFORWARD=y
The CONFIG_NET_IPFORWARD selection informs the hub that if it
receives any packets that are not destined for the hub, it should
forward those packets appropriately.
3. TCP and UDP Tests: The same TCP and UDP tests as described for
the spirit-6lowpan coniguration are supported on the star
endpoints, but NOT on the star hub. Therefore, all network testing
is between endpoints with the hub acting, well, only like a hub.
Each node in the configuration must be manually initialized.
Ideally, this would be automatically initialized with software logic
and configuration data in non-volatilbe memory. The the procedure
is manual in this example. These are the basic initialization
steps with E1 and E2 representing the two star endpoints and C
representing the star hub:
C: nsh> ifup wpan0 <-- Brings up the network on the hub
C: nsh> telnetd <-- Starts the Telnet daemon on the hub
C: nsh> ifconfig <-- To get the IP address of the hub
E1: nsh> ifconfig wpan0 hw 37 <-- Sets E1 endpoint node address
E1: nsh> ifup wpan0 <-- Brings up the network on the E1 node
E1: nsh> telnetd <-- Starts the Telnet daemon on the E1 node
E1: nsh> ifconfig <-- To get the IP address of E1 endpoint
E2: nsh> ifconfig wpan0 hw 38 <-- Sets E2 endpoint node address
E2: nsh> ifup wpan0 <-- Brings up the network on the E2 node
E2: nsh> telnetd <-- Starts the Telnet daemon on the E2 node
E2: nsh> ifconfig <-- To get the IP address of E2 endpoint
It is not necessary to set the hub node address, that will automatically
be set to CONFIG_SPIRIT_HUBNODE when the hub boots. CONFIG_SPIRIT_HUBNODE
is the "well-known" address of the star hub.
The modified usage of the TCP test is then show below:
E1: nsh> tcpserver &
E2: nsh> tcpclient <server-ip> &
Where <server-ip> is the IP address of the E1 endpoint.
Similarly for the UDP test:
E1: nsh> udpserver &
E2: nsh> udpclient <server-ip> &
Telenet sessions may be initiated from the any node to any other node:
XX: nsh> telnet <server-ip> <-- Runs the Telnet client on any node XX
Where <server-ip> is the IP address of either the E1 or E2 endpoints
or of the star hub.
4. Hub UDP Test. The hub of the star does not support the same level of
test as for the endpoint-to-endpoint tests described above. The primary
role of the hub is packet forwarding. The hub does support to test
applications, however: (1) A Telnet client that will permit the hub to
establish remote NSH sesstions with any endpoint, and (2) A special
version of the udpclient program to support testing of Spirit broadcast.
IPv6 does not support "broadcast" in the same since as IPv4. IPv6
supports only multicast. The special multicast address, ff02::1 is
the "all-nodes address" and is functionally equivalent to broadcast.
The spirit radios do support both multicast and broadcast with the
special addresses 0xee and 0xff, respectively. So the Spirit driver
will map the all-nodes IPv6 to the Spirit destination address 0xff and
the packet will be broadcast to all Spirit nodes.
Here are the procedures for using the test
C: nsh> ifup wpan0 <-- Brings up the network on the hub
E1: nsh> ifconfig wpan0 hw 37 <-- Sets E1 endpoint node address
E1: nsh> ifup wpan0 <-- Brings up the network on the E1 node
E1: udpserver & <-- Start the UDP server
E2: nsh> ifconfig wpan0 hw 38 <-- Sets E2 endpoint node address
E2: nsh> ifup wpan0 <-- Brings up the network on the E2 node
E2: udpserver & <-- Start the UDP server
C: udpclient & <-- Starts the UDP client side of the test
The client will broadcast the UDP packets and, as each UDP packet is
sent, it will be received by BOTH endpoints.
STATUS:
2017-08-05: Configurations added. Early testing suggests that there is
a problem when packets are received from multiple sources at high rates:
New incoming packets appear to cause RX FIFO errors and the driver does
not recover well.
2017-08-06: The RX FIFO errors are worse when debug is enabled. This led
me to believe that the cause of the RX FIFO errors was due to too many
interactions by the LP and HP work queue. I restructured the tasking to
reduce the amount of interlocking, but this did not eliminate the RX FIFO
errors.
Hmmm.. this statement appears in the STMicro driver: "Sometimes Spirit1
seems to NOT deliver (correctly) the 'IRQ_RX_DATA_READY' event for
packets which have a length which is close to a multiple of RX FIFO size.
Furthermore, in these cases also the content delivery seems to be
compromised as well as the generation of RX/TX FIFO errors. This can be
avoided by reducing the maximum packet length to a value which is lower
than the RX FIFO size."
I tried implementing the RX FIFO almost full water mark thinking this
might be a work around... it is not. Still RX FIFO errors. From my
reading, the only known work-around is to reduce the maximum packet
size so that it is smaller than 96. I tried setting the maximum packet
length to 84 and that did NOT eliminate the RX FIFO error.
At the end of the TCP test, the "nsh> ifconfig" command shows that
there were two TX timeouts. Perhaps this is related? I found that
the TX timeout was not being cancelled. It must be canceled on each
TX completed or TX error. This DID eliminate the RX FIFO error, but
now the test hangs and does not complete.
Another Errata: "Using the STack packet format and no CRC field, the
reading from RX FIFO to the last received byte, is not possible. ..."
Workaround: "By configuring the packet handler with at least one byte
of CRC, the problem is solved. If the CRC is not required in the
application, configure one byte of CRC in the receiver only, to read
the payload correctly from RX FIFO."
Reducing the FIFO to 94 bytes fixed the problem with the 2 byte CRC
but did not resolve that occasional RX FIFO error.
2017-08-07: The hang noted yesterday was due to logic that did not
restart the poll timer in the event that Spirit was not ready when the
time expired. Just unconditionally performing the poll fixed this.
Then I noticed several assertions. In a busy radio environment, there
are many race conditions. Most typically, just when the driver is
setting up to perform a transmission, the hardware commits to a
reception. The symptom is that the driver times out out waiting to go
into the TX state (because it is in the RX state). The logic needed to
be beefed up to handle this routinely without asserting and without
leaving the Spirit in a bad state.
The TCP test beats the radio very hard and it is actually heartening
that there are no failures that lead to data loss in this environment.
I would say it is functional but fragile in this usage, but probably
robust in a less busy environment.
2017-08-08: Added broadcast packet transfers using the hub-based
broadcast UDP client. This appears to be a problem the HC06
compression and/or decompression. The decompression logic comes up
with the destination address of ff02::ff00:00fe:3500 (which derives
from the receiving node address of 37) instead of the all-nodes
multicast address of ff02::0001. It is then out of sync with the
IPHC headers and is unable to uncompress the rest of the packet
correctly.
Trying again with HC1 compression, I see other isses. The first
frame is received correctly, but the following frames have an incorrect
packet length and generate RX FIFO errors. Forcing the send size to
12 bytes of payload in apps/examples/udp (vs 96), eliminates this
problem and the broadcast works well.
There is probably another issue related to broadcast TX setup: If
we are sending to the multicast or broadcast address, should we
not also disable ACKs, retries, and RX timeouts? What will happen
if multiple radios ACK? At a minimum it could keep the driver
unnecessarily busy. There is some prototype code to do just this
in the driver, but does not seem to work.
2017-08-26: There was only a single buffer for reassemblying larger
packets. This could be a problem issue for the hub configuration
which really needs the capability concurrently reassemble multiple
incoming streams. The design was extended to support multiple
reassembly buffers.
Initial testing shows the same basic behavior as noted before:
The UDP test works and TCP test (usually) works. There are,
however, are errors in reported by the hub in the TCP test.
Occassionally the test will hang when ther server echoes the data
back to the client. These errors are presumably the result of ACKs
from the receiver colliding with frames from the sender.
Needs more investigation.
2017-09-08: The HC06 all nodes address decode problem mentioned on
2017-08-08 has been corrected. The behavior in the test case has
not yet been reverified. I suspect that there made to some radio
configuration problems that are causing the RX FIFO errors and the
strange broadcast behavior. I recently got an STEVAL-IDS001V5M
sniffer that should tell me what is going on. But I have not yet
had sufficient free time to continue this testing.
936df1bcb5