LeafLabs Garden

A place to share, learn, and grow...

1. 

   crenn
   Moderator

   @ventosus, That's for the note, I thought it was similar to the F1 in which we needed to tie PB2 to GND to get into the bootloader.

   @gbulmer, I'll try to use mbolivar's code on github as a lot of the structure changes should have been implemented. But depends on what I can find and how fast I can make those changes. Help testing would be appreciated as I can only test Windows and Linux (most likely Ubuntu or Debian). Later if we desired, we could also see how hard it would be to bring it to the RPi as well.

   Posted 2 years ago #
2. 

   gbulmer
   Moderator

   Crenn - I don't think I was clear.

   I really only wanted to know what you think your starting code base might be.
   My comment about mbolivar's work wasn't intended to create extra wok for you.

   Posted 2 years ago #
3. 

   crenn
   Moderator

   You were clear. The reason why I'm wanting to use the code on github is that I remember someone from leaflabs (maybe even mbolver) tell me there were a few hacks used to compile the code. This is obviously a problem in the current releases of the IDE as they only are for one series of controller, but if I can't get the IDE working from a fresh compile of the github repository, then I'll look at using a previous version. It could even be the case that minimal, if any changes, are required.

   The mention about the RPi was a sort of passing thought, as it has actually been asked a few times on the forums about if we can use the maple IDE with the RPi.

   Posted 2 years ago #
4. 

   gbulmer
   Moderator

   crenn - Folks have the Arduino IDE working on R-Pi, e.g.
   http://www.scribd.com/doc/150537740/Arduino-i-de-on-Raspberry-Pi
   So the instructions there get Java installed.

   AFAIK, the R-Pi's gcc generates code for Cortex-M3/4. So that part of the toolchain is "solved"!-)

   I expect that the DfuSe loader, dfu-util will work. This thread (with a low signal to noise ratio initially) http://www.raspberrypi.org/phpBB3/viewtopic.php?t=49271&p=386444
   points at several loaders, including dfu-util at http://dfu-util.gnumonks.org/

   That thread, and the "embedded-pi" folks reference a serial bootloader.
   It is the one LeafLabs/we use: http://leaflabs.com/docs/bootloader.html
   or this thread shows how to load from R-Pi via Coocox's serial bootloader, which AFAICT is derived from the same Python code:
   http://www.raspberrypi.org/phpBB3/viewtopic.php?f=45&t=49574

   Loading over serial has the benefit that it should work for any STM32F, not just STM32F's with the built-in USB-bootloader.

   This thread shows how to build for an STM32F1 on the R-Pi
   http://www.raspberrypi.org/phpBB3/viewtopic.php?f=45&t=52260

   Posted 2 years ago #
5. 

   Adam
   Member

   Hey folks,

   I built an Orone-Mini-S8H with an STM32F303CCT6 on it.

   The board seems to work, since the ROM DFU bootloader works enough to enumerate as an STM32 device and allow me to download code. I also did a :force:mass-erase and that didn't give any errors.

   I followed ventosus' instructions on building, and was able to get the example main.cpp downloaded to the device. However, when add the :leave to the dfu-util options command, I get an error:

   $ dfu-util -a 0 -d 0483:df11 -s 0x08000000:leave -D ../libmaple/build/F303CC.bin
   dfu-util 0.7
   
   Copyright 2005-2008 Weston Schmidt, Harald Welte and OpenMoko Inc.
   Copyright 2010-2012 Tormod Volden and Stefan Schmidt
   This program is Free Software and has ABSOLUTELY NO WARRANTY
   Please report bugs to dfu-util@lists.gnumonks.org
   
   Filter on vendor = 0x0483 product = 0xdf11
   Opening DFU capable USB device... ID 0483:df11
   Run-time device DFU version 011a
   Found DFU: [0483:df11] devnum=0, cfg=1, intf=0, alt=0, name="@Internal Flash  /0x08000000/128*0002Kg"
   Claiming USB DFU Interface...
   Setting Alternate Setting #0 ...
   Determining device status: state = dfuERROR, status = 10
   dfuERROR, clearing status
   Determining device status: state = dfuIDLE, status = 0
   dfuIDLE, continuing
   DFU mode device DFU version 011a
   Device returned transfer size 2048
   No valid DFU suffix signature
   Warning: File has no DFU suffix
   DfuSe interface name: "Internal Flash  "
   Downloading to address = 0x08000000, size = 18180
   .........
   File downloaded successfully
   Transitioning to dfuMANIFEST state
   Error during download get_status

   The LED doesn't blink, and I don't get the "Hello" output on the serial interfaces.

   Any ideas on what's going on?

   Posted 2 years ago #
6. 

   ventosus
   Member

   Error during download get_status

   I get that, too ,all the time. It has to do with the :leave hook. dfu-util tries to get a status update from the bootloader after it has sent the magic RESET-TO-USER-MODE-SIGNAL, but the latter does not seem to answer (most probably because it jumped already to user code). Nothing to worry about.

   That's something I wanted to report up-stream, but forgot to do so, thanks for bringing it up.

   The LED doesn't blink and I don't get the "Hello" output on the serial interfaces

   According to your log, you have built for the F303CC target. This is a general purpose target for all the F303xx chips. It can be used directly or serve as a template for a specific board type. It has arbitrary pins assigned as BOARD_BUTTON_PIN, BOARD_LED_PIN and BOARD_USB_DISC_*.

   There is a note in wirish/boards/F303xx/include/board/board.h that those defines have to be set to the right values for a specific board layout.

   So, you have no working LED and USB DISC circuitry because of wrong PIN definitions...

   You should use the 'orone_mini_F3' build target istead, it has everything set right.

   BOARD=orone_mini_F3

   Posted 2 years ago #
7. 

   Adam
   Member

   ventosus,

   Thank you! I rebuilt using the orone_mini_F3 target and it works!

   gbulmer,

   Thank you for the Orone-mini - great board design!

   I'm going to try some performance tests today and tomorrow. I'll report back.

   Posted 2 years ago #
8. 

   ventosus
   Member

   Error during download get_status

   Up-stream reports that this has been fixed already some months ago in dfu-util git.

   Posted 2 years ago #
9. 

   Adam
   Member

   ventosus,

   Error during download get_status
   Up-stream reports that this has been fixed already some months ago in dfu-util git.

   And so it is. I got the latest source and tried out it - no error message.

   Posted 2 years ago #
10. 

    mlundinse
    Member

    gbulmer - I am not sure that the R Pi gcc toolchain will generate barebones arm-none-eabi program files that is needed by the maple. It is configured to generate code to run under a Linux OS. The same assembler instructions but the default startup and library stuff are different. I have no idea how hard, or easy, it is to convince this toolchain to generate ode for barebones arm.

    Posted 2 years ago #
11. 

    gbulmer
    Moderator

    mlundinse - You're right. I allowed my enthusiasm to overcome good sense. Thinking about it, I'm not sure the R Pi gcc toolchain will generate barebones arm-none-eabi program files either.

    I am slightly less worried about the startup and libraries than I am calling conventions (ABI), linker scripts, static linking, ...

    I vaguely remember reading through an enormous number of options in the documentation for gcc for generating ARM code, and almost anything looking 'possible'. I also very vaguely (more than 5 years) remember a mechanism where the name used to execute the compiler was used by the compiler itself to figure out what it was supposed to do.

    However, I am willing to believe that configuring the right set of options is less likely than winning the National Lottery. I am also willing to believe that it can't be done by any normal gcc for ARM and Linux.

    Thanks for keeping my feet on the ground :-)

    (PS - I am getting into Golang, and maybe Beagle Bone Black, so I'm unlikely to try)

    Posted 2 years ago #
12. 

    ventosus
    Member

    I've worked some of my poor ld-script-fu to make the 8kB core coupled memory (CCM) usable. One can now put code in there which will be run at maximum speed (zero wait state compared to 2 wait states when loading code from FLASH). It can also be used as extended SRAM of course.

    The right place to put interrupts and other critical code to get a considerable performance increase...

    There is a convenience macro __CCM__ which sets the corresponding section attributes.

    
#define __CCM__ __attribute__((section(".CCM")))


    Just append it to your data or function prototypes to be put into core coupled memory.

    
int mydata __CCM__ = 23;


    
void myfunc(void) __CCM__;
void myfunc(void) {
...
}


    Posted 2 years ago #
13. 

    gbulmer
    Moderator

    ventosus - That's very interesting!

    It might be helpful if you would post a link to the ld (load) script?

    Have you done any experiments to measure the effect?
    Is there a slow-down if both code and data reside in core coupled memory (CCM)?

    Posted 2 years ago #
14. 

    ventosus
    Member

    My changes are basen on ST's annotationAN4296:
    http://www.st.com/st-web-ui/static/active/en/resource/technical/document/application_note/DM00083249.pdf

    linker script with its section definitions is at:
    support/ld/common.inc

    initialization code, where code from FLASH is copied to CCM is at:
    wirish/start_c.c

    Here is the git commit with the changes:
    https://github.com/ventosus/libmaple/commit/3a745ec937647f48bc94a58b938ab26f882e752c

    Have you done any experiments to measure the effect?
    No real benchmarking, but I've put interrupts and critical routines into CCM in one of my projects for testing and got an immediate speed-up in the range of 10% to 20%.

    Is there a slow-down if both code and data reside in core coupled memory (CCM)?
    This I don't know. Would you expect this?

    Posted 2 years ago #
15. 

    gbulmer
    Moderator

    ventosus - Thank you for the links.

    My changes are based on ST's annotationAN4296:
    http://www.st.com/st-web-ui/static/active/en/resource/technical/document/application_note/DM00083249.pdf
    I had missed that, so thank you very much.

    "Have you done any experiments to measure the effect?"
    No real benchmarking, but I've put interrupts and critical routines into CCM in one of my projects for testing and got an immediate speed-up in the range of 10% to 20%.

    10%-20% is pretty impressive for a chip which is already pretty quick .

    "Is there a slow-down if both code and data reside in core coupled memory (CCM)?"
    This I don't know. Would you expect this?
    Well, I have looked through the STM32F303 datasheet, and all it says about the CCM is
    "2.3 Embedded SRAM
    The STM32F30xx features up to 48 Kbytes of static SRAM. It can be accessed as bytes, halfwords (16 bits) or full words (32 bits):
    ● ...
    ● 8 Kbytes of CCM RAM. ... This memory can be addressed at maximum system clock frequency without wait state."

    So it looks fine. However there isn't anything more concrete.
    Also, I vaguely remember there being a buffer between the CPU and SRAM, which 'pipelined' the 2nd and subsequent write, but was complex enough that it might slow the first write, or interact with reads. I haven't found this mentioned any where for Cortex-M4 so I may be thinking Cortex-M3.

    Summary: I am happy to believe it does everything perfectly, no matter what mix of instructions and data are used, but I am also willing to believe that there are interactions between reading instructions and writing data.

    However, 10-20% is enough to convince me it's worth it anyway :-)

    Posted 2 years ago #

RSS feed for this topic

Reply »

You must log in to post.