I've got no idea what's wrong, or when, and I'm not super interested personally, as I don't have any hardware. But I'd like to enable the lpc43xx directory for building. Can't/won't do that until the examples build at least :)

Further, they all produce warnings about some ram_local2 memory section?

I built and flashed the stm32f429i-discovery/sdram demo.

I get a mass storage device showing up and it auto-mounts as vfat on Fedora 23. But if I try to do anything like cat or head on the datafile the I see on the drive I get an IO error.

Hello,
i have tested the cdc example for the stellaris-ek-lm4f120xl board under Windows. When i plug USB the device gets enumerated (with a patched inf file) and all seems good. But when i try to open the device with a terminal program i am getting an error.

The Problem seems to be a USB_CDC_REQ_SET_LINE_CODING with databits set to 0. As this is invalid the function cdcacm_control_request() returns 0 and the transfer will be stalled.

So i am unsure how to handle the bad USB_CDC_REQ_SET_LINE_CODING request. Maybe ignore the bad databits setting return with no error.

If desired i can make a pull request.

I have added spi_init_slave.

The change looks straightforward and should work. But it's untested, will move on to testing this shortly. Hmmm.... We could remove the baud rate parameter in the _init_slave call. ok, we SHOULD remove.... Done.

I renamed the "br" parameter to "baud" because "br" confused me for a split second.

cat ../../spi_init_slave.patch
diff --git a/include/libopencm3/stm32/common/spi_common_all.h b/include/libopencm3/stm32/common/spi_common_all.h
index fa4affb..a77fcbf 100644
--- a/include/libopencm3/stm32/common/spi_common_all.h
+++ b/include/libopencm3/stm32/common/spi_common_all.h
@@ -386,7 +386,9 @@ specific memorymap.h header before including this header file.*/
 BEGIN_DECLS

 void spi_reset(uint32_t spi_peripheral);
-int spi_init_master(uint32_t spi, uint32_t br, uint32_t cpol, uint32_t cpha,
+int spi_init_master(uint32_t spi, uint32_t baud, uint32_t cpol, uint32_t cpha,
+                   uint32_t dff, uint32_t lsbfirst);
+int spi_init_slave(uint32_t spi, uint32_t cpol, uint32_t cpha,
                    uint32_t dff, uint32_t lsbfirst);
 void spi_enable(uint32_t spi);
 void spi_disable(uint32_t spi);
diff --git a/lib/stm32/common/spi_common_f03.c b/lib/stm32/common/spi_common_f03.c
index 5713653..486bd48 100644
--- a/lib/stm32/common/spi_common_f03.c
+++ b/lib/stm32/common/spi_common_f03.c
@@ -89,17 +89,17 @@ spi_lsbfirst.
 @returns int. Error code.
 */

-int spi_init_master(uint32_t spi, uint32_t br, uint32_t cpol, uint32_t cpha,
-                   uint32_t crcl, uint32_t lsbfirst)
+int spi_init_master_slave (uint32_t spi, uint32_t baud, uint32_t cpol, uint32_t cpha,
+                          uint32_t crcl, uint32_t lsbfirst, uint32_t master)
 {
        uint32_t reg32 = SPI_CR1(spi);

        /* Reset all bits omitting SPE, CRCEN and CRCNEXT bits. */
        reg32 &= SPI_CR1_SPE | SPI_CR1_CRCEN | SPI_CR1_CRCNEXT;

-       reg32 |= SPI_CR1_MSTR;  /* Configure SPI as master. */
+       reg32 |= master;        /* Configure SPI as master. */

-       reg32 |= br;            /* Set baud rate bits. */
+       reg32 |= baud;          /* Set baud rate bits. */
        reg32 |= cpol;          /* Set CPOL value. */
        reg32 |= cpha;          /* Set CPHA value. */
        reg32 |= crcl;          /* Set crc length (8 or 16 bits). */
@@ -112,6 +112,20 @@ int spi_init_master(uint32_t spi, uint32_t br, uint32_t cpol, uint32_t cpha,
        return 0; /* TODO */
 }

+int spi_init_master(uint32_t spi, uint32_t baud, uint32_t cpol, uint32_t cpha,
+                    uint32_t dff, uint32_t lsbfirst)
+{
+       return spi_init_master_slave(spi, baud, cpol, cpha, dff, lsbfirst, SPI_CR1_MSTR);
+}
+
+
+int spi_init_slave(uint32_t spi, uint32_t cpol, uint32_t cpha,
+                   uint32_t dff, uint32_t lsbfirst)
+{
+       return spi_init_master_slave(spi, 0, cpol, cpha, dff, lsbfirst, 0); // maybe we should have a define for SLAVE
+}
+
+
 void spi_send8(uint32_t spi, uint8_t data)
 {
        /* Wait for transfer finished. */
diff --git a/lib/stm32/common/spi_common_l1f124.c b/lib/stm32/common/spi_common_l1f124.c
index 6345350..0d05364 100644
--- a/lib/stm32/common/spi_common_l1f124.c
+++ b/lib/stm32/common/spi_common_l1f124.c
@@ -89,17 +89,17 @@ spi_lsbfirst.
 @returns int. Error code.
 */

-int spi_init_master(uint32_t spi, uint32_t br, uint32_t cpol, uint32_t cpha,
-                   uint32_t dff, uint32_t lsbfirst)
+int spi_init_master_slave(uint32_t spi, uint32_t baud, uint32_t cpol, uint32_t cpha,
+                         uint32_t dff, uint32_t lsbfirst, uint32_t master)
 {
        uint32_t reg32 = SPI_CR1(spi);

        /* Reset all bits omitting SPE, CRCEN and CRCNEXT bits. */
        reg32 &= SPI_CR1_SPE | SPI_CR1_CRCEN | SPI_CR1_CRCNEXT;

-       reg32 |= SPI_CR1_MSTR;  /* Configure SPI as master. */
+       reg32 |= master;        /* Configure SPI as master (or not). */

-       reg32 |= br;            /* Set baud rate bits. */
+       reg32 |= baud;          /* Set baud rate bits. */
        reg32 |= cpol;          /* Set CPOL value. */
        reg32 |= cpha;          /* Set CPHA value. */
        reg32 |= dff;           /* Set data format (8 or 16 bits). */
@@ -112,6 +112,21 @@ int spi_init_master(uint32_t spi, uint32_t br, uint32_t cpol, uint32_t cpha,
        return 0; /* TODO */
 }

+
+int spi_init_master(uint32_t spi, uint32_t baud, uint32_t cpol, uint32_t cpha,
+                   uint32_t dff, uint32_t lsbfirst)
+{
+       return spi_init_master_slave(spi, baud, cpol, cpha, dff, lsbfirst, SPI_CR1_MSTR);
+}
+
+
+int spi_init_slave(uint32_t spi, uint32_t cpol, uint32_t cpha,
+                  uint32_t dff, uint32_t lsbfirst)
+{
+       return spi_init_master_slave(spi, 0, cpol, cpha, dff, lsbfirst, 0); // maybe we should have a define for SLAVE
+}
+
+
 /*---------------------------------------------------------------------------*/
 /** @brief SPI Set Data Frame Format to 8 bits

There are two ways of dealing with linker scripts (very roughly):

• Old style: static linker script. Example: defining LDSCRIPT=stm32-h103.ld in the Makefile.
• New style: linker dynamically generated. Example: defining device=STM32F103RBT6 and including genlink-config.mk in the Makefile

libopencm3-examples seems to support both on its main Makefile (rules.mk) but most of the examples (all?) use the old style:

• First of all, I guess the new style is the preferred way in libopencm3?
• If so, why are the examples using the old way? Simply because no one has done? Or is it intentional? If the former, I could help porting some examples to the new style.

Thanks!

when i tried compiling the cdc-acm project i get an error stating "stm32f103_usb_driver undeclared " in line no. 261 (cdcacm.c source file) the usbd_init function, as i tried declaring the variable of type usbd_driver which is typedef structure in usb_private.h still get error as storage size unknown.

Hi,

I have an stm32f4-discovery board labeled MB997D which comes with STM32F407VG. I tested every example on the stm32f4-discovery directory and they all work, except for the USB examples. I didn't get my scope hooked up yet, but if I were to guess, Data Pull up isn't being connected.

I suppose the reason for data pullup to not being connected, even though we clear bit 1 in DCTL, is because the USB IP doesn't realize that VBUS is above 4.4V. Going through the datasheet, I found out that PA9 is not, anymore, used as OTG_FS_VBUS. In fact, AF10 is not a valid mux mode for GPIOA_10.

Any hints about what to do here? I actually have this same failure on two different boards (stm32f4-discovery and stm32-h405 from Olimex).

There's references to SCRIPT_DIR and stlink_flash.scr and so on, it's duplicated in the different stm32 directories, and some of the flash erase commands aren't correct for recent/current openocd releases

https://github.com/libopencm3/libopencm3-examples/blob/master/examples/stm32/f3/stm32f3-discovery/usb_cdcacm/cdcacm.c#L261

what is the use?

Description of steps how to reuse libopencm3-examples build system in his own project is not proper. There also should be contained a step of copying ld_script (it's required for stm32f4_discovery at least).
It would be good idea to add basic templates of makefiles for new projects for each platforms.

I've just started diving into opencm3 and I'm pretty impressed. I quickly got a few demos built and flashed. I started with the stm32f429i-discovery which has quite a lot of good stuff.

However, I quickly got stuck because the readme's are so terse.

The top level readme has:

1. usart_console - Program a USART on the board as a console (requires
   clonsing jumpers on your discovery board connecting USART1 to the
   programmer)

Well I guess clonsing is closing but they are already in the closed position to use the STLINK to flash the board. Did this mean open? But that does not work either. :(

The full readme for that demo says:

Simple USART Example

This example sets up a USART port and provides a few simple
character handling functions to that a short interactive program
can be demonstrated. It re-uses the clock setup from systick_blink
and that means you could do times delays but that aspect isn't used.

So what should I expect / do ?

If I get nothing is there a problem? I would not know because there is so little about what this demo is supposed to do.

In fact this whole jumper story is completely opaque, even in the ST user manual . On or off I never get a ACM device showing. There seems to be more to it but some explicit instructions are needed to get the demo to run.

Some of the readme's in the other branches are pretty good but this branch is a little too light to of much help.

https://github.com/libopencm3/libopencm3-examples/blob/master/examples/stm32/f1/stm32-maple/miniblink/miniblink.c

I sed this as a test hello with a Maple Mini clone and got no output. I looked at the code and noticed that it does rcc_periph_clock_enable(RCC_GPIOC); and then wiggles GPIOA5. Changing the enable to RCC_GPIOA made the pin "blink".

There is no led on that pin on Maple, but that may or may not be the intended outcome.

libopencm3-examples / examples / stm32 / f1 / stm32-h103 / usb_cdcacm
is the device name string from libopencm3?
and is there a way to make windows recognise this device.
it knows it is a cdcacm device but can't find a driver?
maybe it's the device name?

because of the mass of #defines (DISCO_) you have to read the sysconfig.h and the main.c at the same time. to understand the use of the external Interrupt it would be better to write the direct name in the librarys functions..

or just some comments on the interrupt functions...

It can be recognized by Linux, and you also can read any sector you want.

But you can't write, if you write, the kernel unloads this device and report I/O error.

Hello!

In lib/usb/usb_f107 (and also in lib/usb/usb_f207) there are 2 while loops, and the "Wait for AHB idle" one never exits if the OTG clock is disabled. A little bit hard to understand (at least for a beginner) why the system hangs. I recommend usiing some timeout method, and returning an error code indicating that the initilazation was unsuccessful.

static usbd_device *stm32f107_usbd_init(void)
{
OTG_FS_GINTSTS = OTG_GINTSTS_MMIS;

OTG_FS_GUSBCFG |= OTG_GUSBCFG_PHYSEL;
/* Enable VBUS sensing in device mode and power down the PHY. */
OTG_FS_GCCFG |= OTG_GCCFG_VBUSBSEN | OTG_GCCFG_PWRDWN;

/* Wait for AHB idle. */
while (!(OTG_FS_GRSTCTL & OTG_GRSTCTL_AHBIDL));
/* Do core soft reset. */
OTG_FS_GRSTCTL |= OTG_GRSTCTL_CSRST;
while (OTG_FS_GRSTCTL & OTG_GRSTCTL_CSRST);

make in top level directory fails with this error:

make[2]: arm-none-eabi-gcc: Command not found

arm-none-eabi-gcc.exe in my system is located at:

C:\Program Files (x86)\GNU Tools ARM Embedded\4.9 2015q1\bin

I've tried running this command:

export PATH="/c//Python27:/c//Program Files (x86)/GNU Tools ARM Embedded/4.9 2015q1/bin:/usr/local/bin:/usr/bin:/bin"

But it doesn't help. What am I doing wrong?

BACKGROUND:
I have a Lisa/M v2 board with the Aspirin v2.2 10DoM imu on it.
I use mac osx 10.8.4, and installed the newest version of paparazzi tools with macports, the gnu chain I have is:

gcc version 4.7.3 20121207 (release) [ARM/embedded-4_7-branch revision 194305](GNU Tools for ARM Embedded Processors)

I use a BMP to upload my code.

NOW TO THE ISSUE:

When I compile examples/stm/f1/lisa-m-2/usart_printf and upload it everything works as it should (I can see the text output on the usart).

When I compiled examples/stm/f1/lisa-m-2/usart_irq_printf and upload it nothing happens, no usart output at least. By adding led signals into the code I can see that the execution stops behaving as it should right after systick_counter_enable(); is called and I believe when the first interrupt occurs.

I tried this issue on two Lisa/M v2 boards, one factory-flashed, one flashed by myself with Luftboot about a month ago.

I also tried on a Ubuntu machine with the newest version of paparazzi and the gcc toolchain that came with it, same result.

I'm the new learner for libopencm3, I use ubuntu14.04 environment.
I found that's really strange that I cannot build the example "miniblink"
my step are here, then there is error message,
but the same step and build tick_blink, it will success,
I don't know why

git clone https://github.com/libopencm3/libopencm3-examples.git
cd ./libopencm3-examples
git submodule init
git submodule update
cd libopencm3/
make
cd ../examples/stm32/f4/stm32f4-discovery/miniblink/
make

the error message like this

arm-none-eabi/bin/ld:generated.STM32F407VG.ld:16: warning: memory region `rom' not declared
arm-none-eabi/6.3.1/../../../../arm-none-eabi/bin/ld: invalid origin for memory region ram
collect2: error: ld returned 1 exit status
make: *** [miniblink.elf] Error 1

and generated.STM32F407VG.ld is

EXTERN(vector_table)
ENTRY(reset_handler)
MEMORY
{
 ram (rwx) : ORIGIN = _RAM_OFF, LENGTH = _RAM
}
SECTIONS
{
 .text : {
  *(.vectors)
  *(.text*)
  . = ALIGN(4);
  *(.rodata*)
  . = ALIGN(4);
 } >rom
 .preinit_array : {
  . = ALIGN(4);
  __preinit_array_start = .;
  KEEP (*(.preinit_array))
  __preinit_array_end = .;
 } >rom
 .init_array : {
  . = ALIGN(4);
  __init_array_start = .;
  KEEP (*(SORT(.init_array.*)))
  KEEP (*(.init_array))
  __init_array_end = .;
 } >rom
 .fini_array : {
  . = ALIGN(4);
  __fini_array_start = .;
  KEEP (*(.fini_array))
  KEEP (*(SORT(.fini_array.*)))
  __fini_array_end = .;
 } >rom
 .ARM.extab : {
  *(.ARM.extab*)
 } >rom
 .ARM.exidx : {
  __exidx_start = .;
  *(.ARM.exidx*)
  __exidx_end = .;
 } >rom
 . = ALIGN(4);
 _etext = .;
 .data : {
  _data = .;
  *(.data*)
  . = ALIGN(4);
  _edata = .;
 } >ram AT >rom
 _data_loadaddr = LOADADDR(.data);
 .bss : {
  *(.bss*)
  *(COMMON)
  . = ALIGN(4);
  _ebss = .;
 } >ram
 /DISCARD/ : { *(.eh_frame) }
 . = ALIGN(4);
 end = .;
}
PROVIDE(_stack = ORIGIN(ram) + LENGTH(ram));

then if I change a part of linker script, it will work successfully,

MEMORY
{
	rom (rx) : ORIGIN = 0x08000000, LENGTH = 1024K
	ram (rwx) : ORIGIN = 0x20000000, LENGTH = 128K
}

Plz add more examples!

Hi,

I can't seem to get /examples/stm32/f4/stm32f4-discovery/usb_cdcacm to
work under the stm32f429, it just doesn't seem to appear as a usb device
under lsusb.

Anyone got any ideas?

cheers

Chris

There's virtually identical copies of makefile.include for many families.

The only real differences are:

• the name of the library
• ARCH_FLAGS (for hard/soft float etc)
• the -DSTM32_XX (and similar)

We should maybe just have a makefile.rules and then even simpler per platform makefiles?

Downsides are that it's a little harder to just copy it into your own project?

It's a whole heap of files that are generated out of the same elf that should probably just be generated on demand?

There's a quite noticeable timing difference, this is on my quad core laptop with make -j8.

Possibly tweak the rules so that "make [format]" or something will do the right thing if people want it, and update the README appropriately?

karlp@pojak:~/src/libopencm3-examples (elvish *)$ cat timing.elfonly 
real    0m9.513s
user    0m22.125s
sys 0m3.160s

real    0m9.293s
user    0m22.086s
sys 0m3.104s

real    0m9.556s
user    0m22.036s
sys 0m3.190s

karlp@pojak:~/src/libopencm3-examples (elvish *)$ cat timing.allbins 
real    0m12.788s
user    0m33.013s
sys 0m5.438s

real    0m13.022s
user    0m33.410s
sys 0m5.379s

real    0m12.841s
user    0m33.099s
sys 0m5.488s

Hi,

I've compiled and flashed the i2c example provided on examples/stm32/f3/stm32f3-discovery/i2c but "Hello, we're running" is printed on serial once the micro is reseted and nothing happens.

I've debugged the problem and found out that it gets stuck inside a loop in the i2c_transfer7 function:
libopencm3/lib/stm32/common/i2c_common_v2.c lines 404 - 411:

	bool wait = true;
	while (wait) {
		if (i2c_transmit_int_status(i2c)) {
			wait = false;
		}
		while (i2c_nack(i2c)); /* FIXME Some error */
	}
	i2c_send_data(i2c, *w++);

The i2c_transmit_int_status(i2c) returns always 0, so the while(wait) { loop never exits. The while (i2c_nack(i2c)); line never gets stuck, so it isn't the culprit.

I've compiled with arm-none-eabi-gcc 7.3.0 on Archlinux.

Yesterday I ran into a few problems linking C++ code with libopencm3 and the provided makefiles. One reason was that it used gcc instead of g++ for linking, the other reason was the missing option --specs=nosys.specs as described in libopencm3/libopencm3#741 .

The following patch fixes these problems for me, but a new variable "OPENCM3_LINK_CPLUSPLUS = 1" has to be set in the makefile to enable C++ linking.

Index: examples/rules.mk
===================================================================
--- examples/rules.mk   (revision 250)
+++ examples/rules.mk   (working copy)
@@ -32,7 +32,11 @@
 
 CC             := $(PREFIX)-gcc
 CXX            := $(PREFIX)-g++
+ifeq ($(OPENCM3_LINK_CPLUSPLUS),1)
+LD             := $(PREFIX)-g++
+else
 LD             := $(PREFIX)-gcc
+endif
 AR             := $(PREFIX)-ar
 AS             := $(PREFIX)-as
 OBJCOPY                := $(PREFIX)-objcopy
@@ -123,7 +127,11 @@
 ###############################################################################
 # Linker flags
 
+ifeq ($(OPENCM3_LINK_CPLUSPLUS),1)
+TGT_LDFLAGS            += --static -nostartfiles --specs=nosys.specs
+else
 TGT_LDFLAGS            += --static -nostartfiles
+endif
 TGT_LDFLAGS            += -T$(LDSCRIPT)
 TGT_LDFLAGS            += $(ARCH_FLAGS) $(DEBUG)
 TGT_LDFLAGS            += -Wl,-Map=$(*).map -Wl,--cref
@@ -208,6 +216,10 @@
        @#printf "  CXX     $(*).cpp\n"
        $(Q)$(CXX) $(TGT_CXXFLAGS) $(CXXFLAGS) $(TGT_CPPFLAGS) $(CPPFLAGS) -o $(*).o -c $(*).cpp
 
+%.o: %.C
+       @#printf "  CXX     $(*).C\n"
+       $(Q)$(CXX) $(TGT_CXXFLAGS) $(CXXFLAGS) $(TGT_CPPFLAGS) $(CPPFLAGS) -o $(*).o -c $(*).C
+
 clean:
        @#printf "  CLEAN\n"
        $(Q)$(RM) *.o *.d *.elf *.bin *.hex *.srec *.list *.map generated.* ${OBJS} ${OBJS:%.o:%.d}

Ooh, it also recognizes *.C files as C++ files :)

(edited: i've put the patch into a code block)

adc_power_on(ADC1) hangs on while (!adc_is_power_on(adc));

I use gcc-arm-none-eabi-4_9-2015q3 toolchain from https://launchpad.net/gcc-arm-embedded/4.9/4.9-2015-q3-update/+download/gcc-arm-none-eabi-4_9-2015q3-20150921-linux.tar.bz2

Which toolchain should i use ?

While building the examples, the following new warning is created with the newest libopencm3 master:

libopencm3-examples/examples/stm32/f1/stm32-h103/usb_cdcacm$ make
cdcacm.c: In function 'cdcacm_set_config':
cdcacm.c:232:5: warning: passing argument 4 of 'usbd_register_control_callback' from incompatible pointer type
     cdcacm_control_request);
     ^
In file included from cdcacm.c:23:0:
../../../../../libopencm3//include/libopencm3/usb/usbd.h:130:12: note: expected 'usbd_control_callback' but argument is of type 'enum usbd_request_return_codes (*)(struct usbd_device *, struct usb_setup_data *, uint8_t **, uint16_t *, void (**)(struct usbd_device *, struct usb_setup_data *))'
 extern int usbd_register_control_callback(usbd_device *usbd_dev, uint8_t type,
            ^

Trying to compile in Platformio, and got this:
src/usbcdc.cpp:146:3: error: elements of array 'const usb_interface ifaces []' have incomplete type } };
^
src/usbcdc.cpp:146:3: error: storage size of 'ifaces' isn't known src/usbcdc.cpp: In function 'void cdcacm_set_config(usbd_device*, uint16_t)':

src/usbcdc.cpp:222:27: error: invalid conversion from 'int (*)(usbd_device*, usb_setup_data*, uint8_t**, uint16_t*, void (**)(usbd _device*, usb_setup_data*)) {aka int (*)(_usbd_device*, usb_setup_data*, unsigned char**, short unsigned int*, void (**)(_usbd_dev ice*, usb_setup_data*))}' to 'usbd_control_callback {aka int (*)(_usbd_device*, usb_setup_data*, unsigned char**, short unsigned i nt*, int (**)(_usbd_device*, usb_setup_data*))}' [-fpermissive] cdcacm_control_request);
^
In file included from src/usbcdc.cpp:23:0: /home/vlad/.platformio/packages/framework-libopencm3/include/libopencm3/usb/usbd.h:94:12: error: initializing argument 4 of 'int usbd_register_control_callback(usbd_device*, uint8_t, uint8_t, usbd_control_callback)' [-fpermissive] extern int usbd_register_control_callback(usbd_device *usbd_dev, uint8_t type, ^ Compiling .pioenvs/disco_f407vg/FrameworkLibOpenCM3/lib/stm32/common/adc_common_v1.o

Any help?

The status quo is to use, for example, $(PREFIX)gcc instead of $(PREFIX)-gcc. (i.e. PREFIX=arm-none-eabi- instead of PREFIX=arm-none-eabi)

This makes it incompatible with projects that use the former, as libopencm3 will use arm-none-eabi--gcc instead.

It would be nice to be able to extend targets like 'clean' when using the Makefile.include files. Replacing the target with a double-colon rule 'clean::' would allow this. Would this be an acceptable change?

https://github.com/libopencm3/libopencm3-examples/blob/master/examples/stm32/f1/stm32-h103/exti_rising_falling/exti_rising_falling.c

does not mention the AFIO requirements if you happen to want to change to another port for the EXTI

That's wrong, exti_select_source() sets the SYSCFG or the AFIO registers depending on target. original reported bug was misidagnosed.

examples/stm32/f1/other/usb_hid/usbhid.c and lots of others copied there use HSI for USB on F1, which is really not supported, even though it sometimes works. Should replace them all with hse8_72, and usb div 1.5 to show the "right" way of doing it

Would it make any sense to license examples as something less restrictive than LGPL? (such as BSD or even public domain)

Libopencm3 is really nice to use,and it might enjoy more widespread adaptation if users would be able to do base their creations on the examples without have to worry about LGPL's restrictions.

The standard Makefile.include for the STM32F4 includes hard floats (yay!) but when you are using the ARM toolchain (launchpad.net version) and you try to link with -lm it gets the non-FPU version. To get the FPU version you have to add

-L$(TOOLCHAINDIR)/lib/fpu -L$(TOOLCHAINDIR)/lib so that lib/fpu is searched first. At that point it works fine. Also I found that working with my own additional library (which has the utility routines I wrote in it) if I add 'LDFLAGS += -lmylib to the Makefile where the binary is built the default Makefile.include puts libopencm3_stm32.a ahead of it in LDFLAGS and linking fails. But changing the build rule for .elf to put the -llibopencm3_stm32.a on the build rule instead $(LDFLAGS) comes first, allows it to build correctly.

Various USB examples contain structs like usb_device_descriptor which are passed to usb_copy_to_pm, which will read them 16 bit at a time. If the struct happens not to be aligned, this causes a performance penalty on Cortex-M3 upwards, and even breaks on Cortex-M0 which doesn't support unaligned access at all (I ran into this while porting an USB example to STM32L0). gcc supports __attribute__ ((aligned (2))) for variables, which could be used to ensure proper alignment. I don't know if using this feature is worth it for the existing examples, but we definitely have to keep this in mind when USB examples are ported to controllers with an M0 core.

When invoking make from the example directory build does not proceed and gives following error:
make: *** No rule to make target 'target.elf', needed by 'elf'. Stop.

• Clone the examples and initialize libopencm3 submodule
• cd examples/stm32/f1/stm32-h103/timer
• OPENCM3_DIR=../../../../../libopencm3/ make

I was looking at reusing the makefiles from the examples here and noticed the stlink-flash target doesn't seem to be fully setup. In Makefile.rules there's a definition for the %.stlink-flash target that runs st-flash here: https://github.com/libopencm3/libopencm3-examples/blob/master/examples/Makefile.rules#L197-L199 However I don't see anything in the file that depends on a $(BINARY).stlink-flash target which would trigger that flashing rule.

I think there should be something like stlink-flash: $(BINARY).stlink-flash added to this block that generates these flashing targets: https://github.com/libopencm3/libopencm3-examples/blob/master/examples/Makefile.rules#L134-L135 That way you could run 'make stlink-flash' in an example folder and it would use st-flash to upload the binary. Just wanted to check am I missing something else that adds that rule, or would this be a useful pull to send back to this repo?

Thanks for providing the great library & examples BTW!

I manged to issue a "pull request" for the libopencm3.
But somehow my git is now messed up, so I'm getting push/pull/merge errors when I try to clean up the patch. So... here is the patch:

diff --git a/examples/stm32/f1/lisa-m-1/usb_hid/usbhid.c b/examples/stm32/f1/lisa-m-1/usb_hid/usbhid.c
index dd84f80..17765c5 100644
--- a/examples/stm32/f1/lisa-m-1/usb_hid/usbhid.c
+++ b/examples/stm32/f1/lisa-m-1/usb_hid/usbhid.c
@@ -345,11 +345,11 @@ int main(void)
        /* Force to SPI mode. This should be default after reset! */
        SPI2_I2SCFGR = 0;
        spi_init_master(SPI2,
-                       SPI_CR1_BAUDRATE_FPCLK_DIV_256,
-                       SPI_CR1_CPOL_CLK_TO_1_WHEN_IDLE,
-                       SPI_CR1_CPHA_CLK_TRANSITION_2,
-                       SPI_CR1_DFF_8BIT,
-                       SPI_CR1_MSBFIRST);
+                       SPI_BAUDRATE_CLK_DIV_256,
+                       SPI_CLK_TO_1_WHEN_IDLE,
+                       SPI_LATCH_RISING,
+                       SPI_FORMAT_8BIT,
+                       SPI_MSBFIRST);
        /* Ignore the stupid NSS pin. */
        spi_enable_software_slave_management(SPI2);
        spi_set_nss_high(SPI2);
diff --git a/examples/stm32/f1/lisa-m-2/spi/spi.c b/examples/stm32/f1/lisa-m-2/spi/spi.c
index f94aaeb..b1f0c0a 100644
--- a/examples/stm32/f1/lisa-m-2/spi/spi.c
+++ b/examples/stm32/f1/lisa-m-2/spi/spi.c
@@ -68,8 +68,8 @@ static void spi_setup(void) {
    * Data frame format: 8-bit
    * Frame format: MSB First
    */
-  spi_init_master(SPI1, SPI_CR1_BAUDRATE_FPCLK_DIV_64, SPI_CR1_CPOL_CLK_TO_1_WHEN_IDLE,
-                  SPI_CR1_CPHA_CLK_TRANSITION_2, SPI_CR1_DFF_8BIT, SPI_CR1_MSBFIRST);
+  spi_init_master(SPI1, SPI_BAUDRATE_CLK_DIV_64, SPI_CLK_TO_1_WHEN_IDLE,
+                  SPI_LATCH_RISING, SPI_FORMAT_8BIT, SPI_MSBFIRST);

   /*
    * Set NSS management to software.
diff --git a/examples/stm32/f1/lisa-m-2/spi_dma/spi_dma.c b/examples/stm32/f1/lisa-m-2/spi_dma/spi_dma.c
index bf948ba..b460244 100644
--- a/examples/stm32/f1/lisa-m-2/spi_dma/spi_dma.c
+++ b/examples/stm32/f1/lisa-m-2/spi_dma/spi_dma.c
@@ -97,12 +97,12 @@ static void spi_setup(void) {
         * Frame format: MSB First
         */
 #if USE_16BIT_TRANSFERS
-       spi_init_master(SPI1, SPI_CR1_BAUDRATE_FPCLK_DIV_64, SPI_CR1_CPOL_CLK_TO_1_WHEN_IDLE,
-                       SPI_CR1_CPHA_CLK_TRANSITION_2, SPI_CR1_DFF_16BIT, SPI_CR1_MSBFIRST);
+#define MY_FORMAT SPI_FORMAT_16BIT
 #else
-       spi_init_master(SPI1, SPI_CR1_BAUDRATE_FPCLK_DIV_64, SPI_CR1_CPOL_CLK_TO_1_WHEN_IDLE,
-                       SPI_CR1_CPHA_CLK_TRANSITION_2, SPI_CR1_DFF_8BIT, SPI_CR1_MSBFIRST);
+#define MY_FORMAT SPI_FORMAT_8BIT
 #endif
+       spi_init_master(SPI1, SPI_BAUDRATE_CLK_DIV_64, SPI_CLK_TO_1_WHEN_IDLE,
+                       SPI_LATCH_RISING, MY_FORMAT, SPI_MSBFIRST);

        /*
         * Set NSS management to software.
diff --git a/examples/stm32/f1/lisa-m-2/spi_dma_adv/spi_dma_adv.c b/examples/stm32/f1/lisa-m-2/spi_dma_adv/spi_dma_adv.c
index 2182865..5947fb6 100644
--- a/examples/stm32/f1/lisa-m-2/spi_dma_adv/spi_dma_adv.c
+++ b/examples/stm32/f1/lisa-m-2/spi_dma_adv/spi_dma_adv.c
@@ -107,13 +107,14 @@ static void spi_setup(void) {
         * Frame format: MSB First
         */
 #if USE_16BIT_TRANSFERS
-       spi_init_master(SPI1, SPI_CR1_BAUDRATE_FPCLK_DIV_64, SPI_CR1_CPOL_CLK_TO_1_WHEN_IDLE,
-                       SPI_CR1_CPHA_CLK_TRANSITION_2, SPI_CR1_DFF_16BIT, SPI_CR1_MSBFIRST);
-#else
-       spi_init_master(SPI1, SPI_CR1_BAUDRATE_FPCLK_DIV_64, SPI_CR1_CPOL_CLK_TO_1_WHEN_IDLE,
-                       SPI_CR1_CPHA_CLK_TRANSITION_2, SPI_CR1_DFF_8BIT, SPI_CR1_MSBFIRST);
+#define MY_BIT_FORMAT SPI_FORMAT_16BIT
+#else 
+#define MY_BIT_FORMAT SPI_FORMAT_8BIT
 #endif

+       spi_init_master(SPI1, SPI_BAUDRATE_CLK_DIV_64, SPI_CLK_TO_1_WHEN_IDLE,
+                       SPI_LATCH_RISING, MY_BIT_FORMAT, SPI_MSBFIRST);
+
        /*
         * Set NSS management to software.
         *
diff --git a/examples/stm32/f2/jobygps/spi_test/spi_test.c b/examples/stm32/f2/jobygps/spi_test/spi_test.c
index bb570c2..10d6f65 100644
--- a/examples/stm32/f2/jobygps/spi_test/spi_test.c
+++ b/examples/stm32/f2/jobygps/spi_test/spi_test.c
@@ -45,8 +45,8 @@ static void spi_setup(void)
        gpio_set_af(GPIOB, GPIO_AF5, GPIO13 | GPIO14 | GPIO15);

        /* Setup SPI parameters. */
-       spi_init_master(SPI2, SPI_CR1_BAUDRATE_FPCLK_DIV_256, SPI_CR1_CPOL,
-                       SPI_CR1_CPHA, SPI_CR1_DFF_8BIT, SPI_CR1_MSBFIRST);
+       spi_init_master(SPI2, SPI_BAUDRATE_CLK_DIV_256, SPI_CLK_TO_1_WHEN_IDLE,
+                       SPI_LATCH_RISING, SPI_FORMAT_8BIT, SPI_MSBFIRST);
        spi_enable_ss_output(SPI2); /* Required, see NSS, 25.3.1 section. */

        /* Finally enable the SPI. */

I'm trying to compile libopencm3-examples on Arch Linux (LC_ALL=C make V=1), and I get the following error: when the make process reaches examples/stm32/f0/stm32f0-discovery/usart_stdio:

make[1]: Entering directory '/home/chris/src/stm32/libopencm3-examples/examples/stm32/f0/stm32f0-discovery/usart_stdio'
Using /home/chris/src/stm32/libopencm3-examples/libopencm3 path to library
arm-none-eabi-gcc -Os -g -Wextra -Wshadow -Wimplicit-function-declaration -Wredundant-decls -Wmissing-prototypes -Wstrict-prototypes -fno-common -ffunction-sections -fdata-sections -MD -Wall -Wundef -I/home/chris/src/stm32/libopencm3-examples/libopencm3/include -DSTM32F0 -mthumb -mcpu=cortex-m0 -msoft-float -o usart_stdio.o -c usart_stdio.c
usart_stdio.c: In function 'usart_setup':
usart_stdio.c:67:2: error: unknown type name 'cookie_io_functions_t'
  cookie_io_functions_t stub = { _iord, _iowr, NULL, NULL };
  ^

The same unknown type error occurs when trying to compile examples/stm32/f4/stm32f429i-discovery/lcd-dma.

I'm using arm-none-eabi-gcc version 5.3.0 from this package but the same error occurs with version 5.3.1 20160307 from the toolchain launchpad distribution.

I also have the arm-non-eabi-newlib package installed, which places its files under /usr/arm-non-eabi, but he launchpad toolchain comes with its own copy of newlib too, AFAICS.

What am I missing?

We need an additional committer or two on this repo.

So Piotr is busy elsewhere it seems and this repo is getting more and more commits backed up waiting to be integrated. I've created the equivalents for the STM32F429-Discovery (the "Disco" board) and am going to port my LCD example to it. It would be nice to put it here.

To be confirmed, but clearly needs rewording at least..

19:40 < Roklobsta> the instructions on https://github.com/libopencm3/libopencm3-examples README 
                   don't work.  "make: *** No rule to make target `miniblink.elf', needed by 
                   `elf'.  Stop."
19:54 < Roklobsta> ah i did mean only the part from REUSE onwards.
19:54 < Roklobsta> the parts before were fine.
19:54 < Roklobsta> sorry should have been clearer.

Hi,

I've just run a make on the entire lot of examples and came across the following. The first 2 don't bother me, it's the last 2...

I'll look into it in the mean time, but if someone else could too as I rarely have free time..

CC sync.c
../../cm3/sync.c:30:2: warning: #warning "sync not supported on ARMv6-M arch" [-Wcpp]
#warning "sync not supported on ARMv6-M arch"

CC sync.c
../../cm3/sync.c:30:2: warning: #warning "sync not supported on ARMv6-M arch" [-Wcpp]
#warning "sync not supported on ARMv6-M arch"

BUILD examples/stm32/f4/stm32f4-discovery/usb_cdcacm/
cdcacm.c: In function 'main':
cdcacm.c:242:4: warning: passing argument 4 of 'usbd_init' discards 'const' qualifier from pointer target type [enabled by default]
usbd_control_buffer, sizeof(usbd_control_buffer));
^
In file included from cdcacm.c:23:0:
/Users/adriansluijters/Documents/Developer/STM32/libopencm3-examples/libopencm3/include/libopencm3/usb/usbd.h:62:22: note: expected 'const char *' but argument is of type 'const char * const'
extern usbd_device * usbd_init(const usbd_driver *driver,
^

Cheers

A

I would love to see some examples for multithreading. Are there any? Is it possible to achieve this functionality without full-blown RTOS just with pure libopencm3?

usb_setup() includes calls that are "setting up usart2 on gpio a11,a12" but those are a) not used, and b) the usb pins, and c) the wrong AF

I suspect those lines can just be dropped, but I don't have an f3 board to try that out.

Hello,
My board is based on MCU stm32f103c8t6.
I am learning and playing with examples and get some misunderstanding with setting prescaler for timers.
I use TIM4, it is connected to APB1 and has frequency 36Mhz for me.
I want counter to be updated every 1ms.
What prescaler should I use: 36000 or 35999?
I have checked examples and here I see 36000:
https://github.com/libopencm3/libopencm3-examples/blob/master/examples/stm32/f1/stm32-h103/timer/timer.c#L91

But here prescaler is decremented by 1 :
https://github.com/libopencm3/libopencm3-examples/blob/master/examples/stm32/l1/stm32l-discovery/button-irq-printf/main.c#L142

According to reference manual, real prescaler should be decremented:

Bits15:0 PSC[15:0]:Prescalervalue
The counter clock frequency CK_CNT is equal to fCK_PSC / (PSC[15:0] + 1).

Is it error in examples or libopencm3 decrements prescaler before set it to register?

Thank you,
Artem

examples/stm32/l1/stm32l-discovery/button-irq-printf-lowpower/main.c uses it's own __WFI() cmsis style operation, instead of the one in cm3.h... double check it works and replace it

Hi,

Both usart and the adc examples on the stm32f0 and the i2c example of the stm32f3 devices are still using deprecated defines (based on libopencm3 head, commit 3922cc7).

usart and adc examples on stm32f0 (stm32f0-discovery).

This examples use the following:

usart_set_stopbits(USART1, USART_CR2_STOP_1_0BIT);

USART_CR2_STOP_1_0BIT it's only used on this examples, all other usart examples use one of the following defines (usart_cr2_stopbits):
#define USART_STOPBITS_1 USART_CR2_STOPBITS_1 /* 1 stop bit /
#define USART_STOPBITS_0_5 USART_CR2_STOPBITS_0_5 / .5 stop bit /
#define USART_STOPBITS_2 USART_CR2_STOPBITS_2 / 2 stop bits /
#define USART_STOPBITS_1_5 USART_CR2_STOPBITS_1_5 / 1.5 stop bit*/

So the following fix it:
usart_set_stopbits(USART1, USART_STOPBITS_1);

i2c example on stm32f3 (stm32f3-discovery).

I2C_CR1_DNF_DISABLED is used and it was removed in commit cb2eff (October 26, 2017). using 0 is fine in this case.

Error:
i2c_set_digital_filter(I2C1, I2C_CR1_DNF_DISABLED);

Solved with:
i2c_set_digital_filter(I2C1, 0);

Regards,
Carlos