USBserial don’t works – what’s wrong?

MapleMini clone, bootloader works fine – but when control switches to application (even Bootloader Updater) host gets error:
bootloader’s log
[4409331.689199] usb 4-1.6.2: new full-speed USB device number 97 using ehci-pci
[4409331.775147] usb 4-1.6.2: New USB device found, idVendor=1eaf, idProduct=0003
[4409331.775151] usb 4-1.6.2: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[4409331.775152] usb 4-1.6.2: Product: Maple 003
[4409331.775154] usb 4-1.6.2: Manufacturer: LeafLabs
[4409331.775155] usb 4-1.6.2: SerialNumber: LLM 003
[4409334.581412] usb 4-1.6.2: USB disconnect, device number 97


Did you check the value of the pull up resistor in the bluepill board?
Should be 1k5, or as close as possible, other values may cause re-enumeration issues.
Other than that, may help to report what OS you are using, and perhaps testing with an older version of the core in case something has changed recently.

I assume you are on a Linux machine.
I think your Arduino IDE version is a bit old (1.6.2.).
Please try to use at least 1.6.9.

victor_pv wrote:Did you check the value of the pull up resistor in the bluepill board?

vde wrote:MapleMini clone, bootloader works fine – but when control switches to application (even Bootloader Updater) host gets error:
…
ps. it was compiled with -O0 or -Os without any difference. Library downloaded yesterday from Github

vde wrote:

You are wrong, this is Arduino 1.8.1

mrburnette wrote:Therefore, you must select the board Maple Mini and the correct bootloader version from the ArduinoIDE dropdown.

Can you post the compiler verbose output by building the blink sketch? Maybe we can find out some issues from there.
The serial USB should be defined and compiled when the “STM32duino bootloader” is selected in IDE as upload option, otherwise is not sure.

my logs are similar to logs viewtopic.php?t=1032 – but now “volatile” present and moreover I build with -O0 to exclude optimozer’s issues

stevestrong wrote:Can you post the compiler verbose output by building the blink sketch? Maybe we can find out some issues from there.

I didn’t mind the kernel’s log, but the build log output from the IDE.

PS. I have ST-link and Eclipse so can debug STM32 – but USB issues are very hard to debug
Don’t even try to step through the code with a debugger while using USB as Serial. 99% time you will mess with dead everything

Pito wrote:PS. I have ST-link and Eclipse so can debug STM32 – but USB issues are very hard to debug
Don’t even try to step through the code with a debugger while using USB as Serial. 99% time you will mess with dead everything

stevestrong wrote:I didn’t mind the kernel’s log, but the build log output from the IDE.

a similar error also versed there – http://stackoverflow.com/questions/2728 … 4-error-18

And one else – https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64953

As you already figured out, the problem may be caused by the new compiler version 4.9.x.
As temporary solution, try a clean install to downgrade the IDE to 1.6.12, which uses gcc 4.8.3-q4 (I think). Anyway, try to use the gcc 4.8.3.

stevestrong wrote:As you already figured out, the problem may be caused by the new compiler version 4.9.x.
As temporary solution, try a clean install to downgrade the IDE to 1.6.12, which uses gcc 4.8.3-q4 (I think). Anyway, try to use the gcc 4.8.3.

vde wrote:… I can’t use Arduino’s compiler because this is not the only project on ARM

In Win7
C:\Users\USER\AppData\Local\Arduino15\packages\arduino\tools\arm-none-eabi-gcc\4.8.3-2014q1

vde wrote:stevestrong wrote:As you already figured out, the problem may be caused by the new compiler version 4.9.x.
As temporary solution, try a clean install to downgrade the IDE to 1.6.12, which uses gcc 4.8.3-q4 (I think). Anyway, try to use the gcc 4.8.3.

@vde
I am aware that USB can be buggy. I just wanted to ask you to try with 4.8.3 to see whether is working or not.

If it is working, than it is for sure the USB SW + compiler issue, which we could then check.

Of course, you could also help by posting the disassembled version of the questioned USB module(s) for both compiler versions, just to easy our collective work, as we are doing all this in our free time…

stevestrong wrote:I just wanted to ask you to try with 4.8.3 to see whether is working or not.

victor_pv wrote:Could be a bug in the compiler too

vde wrote:victor_pv wrote:Could be a bug in the compiler too

USB COM disappear, do not know where is it, meybe i upload mini boot loader and com port disappear?
And SoftwareSerial stop work

Double-check that the allocated pins for software serial are not disturbing the USB pins (PA11/PA12 on blue pill).
Also, I wold strongly recommend to use the PXY notation for the port pins instead of raw numbers.
For example, use “PA7” instead of “4” (on blue pill).

sorry i am stupid, choose generic – but have maple mini.

@Vadim, please add to the title [SOLVED], if your problem is solved now.

i use stlink pretty much for all, even ‘baite MM’.
if i use MM then stlink is not available, so i tend to use generic 103c and b variant

why should it make a difference, if so what is/are the difference(s) ?

stephen

victor_pv wrote:vde wrote:victor_pv wrote:Do you mean that because ARM approved it would not have bugs? I have to disagree.

vde wrote:That bugs are very rare, and moreover a lots of programs which works fine after 5.4 now even don’t starts after 6.1. Surprise

For the Arduino Community, changes can be difficult; stuff works OK in an older version but the IDE tosses all manner of errors/warning on just a simple ‘.1’ upgrade.
True. The latest 1.8.1 compiler throws all sorts of warnings that weren’t there on 1.6.x, but my program still works as it always did.

There is a Standards organization that set the “rules” for C++ and these rules are incorporated into new releases of GCC

It’s the philosophy which does not help in answering the question “why the VCP does not work with less ancient compiler”

stevestrong wrote:@vde
you could also help by posting the disassembled version of the questioned USB module(s) for both compiler versions, just to easy our collective work, as we are doing all this in our free time…

vde wrote:There is a Standards organization that set the “rules” for C++ and these rules are incorporated into new releases of GCC

It’s the philosophy which does not help in answering the question “why the VCP does not work with less ancient compiler”

Rick

Judging by the PR for to fix the bootloader to make it work with gcc 5.2, there were differences in when the compiler put stuff onto the stack, which caused that particular issue, so I’m sure there are loads of internal differences between compiler versions.
Hence why code normally indicates which compiler (and version) is required.

e.g the bootloader fix was
https://github.com/rogerclarkmelbourne/ … 86dc865f18

I find it a bit odd that people expect to be able to update the compiler to a version which is not supported, and then complain that things don’t work.

Likewise, people complain that the core doesnt work correctly with Platformuino or VisualStudio etc, despite the Readme explicitly saying that the core works with a specific version of the Arduino IDE (and hence people should presume that it won’t necessarily work with any other IDE or compiler that they fancy using)

RogerClark wrote:Rick

Judging by the PR for to fix the bootloader to make it work with gcc 5.2, there were differences in when the compiler put stuff onto the stack, which caused that particular issue, so I’m sure there are loads of internal differences between compiler versions.
Hence why code normally indicates which compiler (and version) is required.

e.g the bootloader fix was
https://github.com/rogerclarkmelbourne/ … 86dc865f18

I find it a bit odd that people expect to be able to update the compiler to a version which is not supported, and then complain that things don’t work.

Likewise, people complain that the core doesnt work correctly with Platformuino or VisualStudio etc, despite the Readme explicitly saying that the core works with a specific version of the Arduino IDE (and hence people should presume that it won’t necessarily work with any other IDE or compiler that they fancy using)

Victor

This situation happens a lot.

People seem to treat open source software as if it was commercial software they paid for and offered service level guarantees and free support.

I find I get random questions in comments to my blog, and to my youtube videos and also emails direct to me (often via youtube), as if I offer some free support service for anything related to STM32, nRF51 or microcontrollers etc in general.

I’ve given up replying to these, as I have had personal abuse for pointing out that its open source / free / non commercial and its up to them to fix it if it doesn’t happen to work for their person configuration.

victor_pv wrote:
…RogerClark wrote:
…
I find it a bit odd that people expect to be able to update the compiler to a version which is not supported, and then complain that things don’t work.

Likewise, people complain that the core doesnt work correctly with Platformuino or VisualStudio etc, despite the Readme explicitly saying that the core works with a specific version of the Arduino IDE (and hence people should presume that it won’t necessarily work with any other IDE or compiler that they fancy using)

Thanks Ray

It did stress me out for about 5 mins, but now that I just delete any such email, I got some of my life back

I get enough stress with work (being freelance) that I don’t need additional load and stress from people I don’t know, making unreasonable demands of me.

But I consider it their problem not mine.