If you have problems with the dependency build (for example, if you are not
building on C:), then you may need to modify tools/mkdeps.sh
The CodeSourcery Toolchain (2009q1)
-----------------------------------
The CodeSourcery toolchain (2009q1) does not work with default optimization
level of -Os (See Make.defs). It will work with -O0, -O1, or -O2, but not with
-Os.
The Atollic "Pro" and "Lite" Toolchain
--------------------------------------
One problem that I had with the Atollic toolchains is that the provide a gcc.exe
and g++.exe in the same bin/ file as their ARM binaries. If the Atollic bin/ path
appears in your PATH variable before /usr/bin, then you will get the wrong gcc
when you try to build host executables. This will cause to strange, uninterpretable
errors build some host binaries in tools/ when you first make.
The Atollic "Lite" Toolchain
----------------------------
The free, "Lite" version of the Atollic toolchain does not support C++ nor
does it support ar, nm, objdump, or objdcopy. If you use the Atollic "Lite"
toolchain, you will have to set:
CONFIG_HAVE_CXX=n
In order to compile successfully. Otherwise, you will get errors like:
"C++ Compiler only available in TrueSTUDIO Professional"
The make may then fail in some of the post link processing because of some of
the other missing tools. The Make.defs file replaces the ar and nm with
the default system x86 tool versions and these seem to work okay. Disable all
of the following to avoid using objcopy:
CONFIG_RRLOAD_BINARY=n
CONFIG_INTELHEX_BINARY=n
CONFIG_MOTOROLA_SREC=n
CONFIG_RAW_BINARY=n
devkitARM
---------
The devkitARM toolchain includes a version of MSYS make. Make sure that the
the paths to Cygwin's /bin and /usr/bin directories appear BEFORE the devkitARM
path or will get the wrong version of make.
IDEs
====
NuttX is built using command-line make. It can be used with an IDE, but some
effort will be required to create the project (There is a simple RIDE project
in the RIDE subdirectory).
Makefile Build
--------------
Under Eclipse, it is pretty easy to set up an "empty makefile project" and
simply use the NuttX makefile to build the system. That is almost for free
under Linux. Under Windows, you will need to set up the "Cygwin GCC" empty
makefile project in order to work with Windows (Google for "Eclipse Cygwin" -
there is a lot of help on the internet).
Native Build
------------
Here are a few tips before you start that effort:
1) Select the toolchain that you will be using in your .config file
2) Start the NuttX build at least one time from the Cygwin command line
before trying to create your project. This is necessary to create
certain auto-generated files and directories that will be needed.
3) Set up include pathes: You will need include/, arch/arm/src/stm32,
arch/arm/src/common, arch/arm/src/armv7-m, and sched/.
4) All assembly files need to have the definition option -D __ASSEMBLY__
on the command line.
Startup files will probably cause you some headaches. The NuttX startup file
is arch/arm/src/stm32/stm32_vectors.S. With RIDE, I have to build NuttX
one time from the Cygwin command line in order to obtain the pre-built
startup object needed by RIDE.
NuttX buildroot Toolchain
=========================
A GNU GCC-based toolchain is assumed. The files */setenv.sh should
be modified to point to the correct path to the Cortex-M3 GCC toolchain (if
different from the default in your PATH variable).
If you have no Cortex-M3 toolchain, one can be downloaded from the NuttX
SourceForge download site (https://sourceforge.net/project/showfiles.php?group_id=189573).
This GNU toolchain builds and executes in the Linux or Cygwin environment.
1. You must have already configured Nuttx in <some-dir>/nuttx.
cd tools
./configure.sh fire-stm32v2/<sub-dir>
2. Download the latest buildroot package into <some-dir>
3. unpack the buildroot tarball. The resulting directory may
have versioning information on it like buildroot-x.y.z. If so,
rename <some-dir>/buildroot-x.y.z to <some-dir>/buildroot.
4. cd <some-dir>/buildroot
5. cp configs/cortexm3-defconfig-4.3.3 .config
6. make oldconfig
7. make
8. Edit setenv.h, if necessary, so that the PATH variable includes
the path to the newly built binaries.
See the file configs/README.txt in the buildroot source tree. That has more
detailed PLUS some special instructions that you will need to follow if you are
building a Cortex-M3 toolchain for Cygwin under Windows.
DFU and JTAG
============
Enbling Support for the DFU Bootloader
--------------------------------------
The linker files in these projects can be configured to indicate that you
will be loading code using STMicro built-in USB Device Firmware Upgrade (DFU)
loader or via some JTAG emulator. You can specify the DFU bootloader by
adding the following line:
CONFIG_STM32_DFU=y
to your .config file. Most of the configurations in this directory are set
up to use the DFU loader.
If CONFIG_STM32_DFU is defined, the code will not be positioned at the beginning
of FLASH (0x08000000) but will be offset to 0x08003000. This offset is needed
to make space for the DFU loader and 0x08003000 is where the DFU loader expects
to find new applications at boot time. If you need to change that origin for some
other bootloader, you will need to edit the file(s) ld.script.dfu for the
configuration.
The DFU SE PC-based software is available from the STMicro website,
http://www.st.com. General usage instructions:
1. Convert the NuttX Intel Hex file (nuttx.hex) into a special DFU
file (nuttx.dfu)... see below for details.
2. Connect the M3 Wildfire board to your computer using a USB
cable.
3. Start the DFU loader on the M3 Wildfire board. You do this by
resetting the board while holding the "Key" button. Windows should
recognize that the DFU loader has been installed.
3. Run the DFU SE program to load nuttx.dfu into FLASH.
What if the DFU loader is not in FLASH? The loader code is available
inside of the Demo dirctory of the USBLib ZIP file that can be downloaded
from the STMicro Website. You can build it using RIDE (or other toolchains);
you will need a JTAG emulator to burn it into FLASH the first time.
In order to use STMicro's built-in DFU loader, you will have to get
the NuttX binary into a special format with a .dfu extension. The
DFU SE PC_based software installation includes a file "DFU File Manager"
conversion program that a file in Intel Hex format to the special DFU
format. When you successfully build NuttX, you will find a file called
nutt.hex in the top-level directory. That is the file that you should
provide to the DFU File Manager. You will end up with a file called
nuttx.dfu that you can use with the STMicro DFU SE program.
Enabling JTAG
-------------
If you are not using the DFU, then you will probably also need to enable
JTAG support. By default, all JTAG support is disabled but there NuttX
configuration options to enable JTAG in various different ways.
These configurations effect the setting of the SWJ_CFG[2:0] bits in the AFIO
MAPR register. These bits are used to configure the SWJ and trace alternate function I/Os. The SWJ (SerialWire JTAG) supports JTAG or SWD access to the
Cortex debug port. The default state in this port is for all JTAG support
to be disable.
CONFIG_STM32_JTAG_FULL_ENABLE - sets SWJ_CFG[2:0] to 000 which enables full
SWJ (JTAG-DP + SW-DP)
CONFIG_STM32_JTAG_NOJNTRST_ENABLE - sets SWJ_CFG[2:0] to 001 which enable
full SWJ (JTAG-DP + SW-DP) but without JNTRST.
CONFIG_STM32_JTAG_SW_ENABLE - sets SWJ_CFG[2:0] to 010 which would set JTAG-DP
disabled and SW-DP enabled
The default setting (none of the above defined) is SWJ_CFG[2:0] set to 100
which disable JTAG-DP and SW-DP.
OpenOCD
=======
I have also used OpenOCD with the M3 Wildfire. In this case, I used
the Olimex USB ARM OCD. See the script in configs/fire-stm32v2/tools/oocd.sh
for more information. Using the script:
1) Start the OpenOCD GDB server
cd <nuttx-build-directory>
configs/fire-stm32v2/tools/oocd.sh $PWD
2) Load Nuttx
cd <nuttx-built-directory>
arm-none-eabi-gdb nuttx
gdb> target remote localhost:3333
gdb> mon reset
gdb> mon halt
gdb> load nuttx
3) Running NuttX
gdb> mon reset
gdb> c
LEDs
====
The M3 Wildfire has 3 LEDs labeled LED1, LED2 and LED3. These LEDs are not
used by the NuttX port unless CONFIG_ARCH_LEDS is defined. In that case, the
usage by the board port is defined in include/board.h and src/up_autoleds.c.
The LEDs are used to encode OS-related events as follows:
