Mbed-os: Ambiguity in allocation SPI peripheral of SPIFBlockDevice

[lefebvresam]

Description of defect

When using SPIFBlockDevice you have to define the pins for SPI connection in the form:

SPIFBlockDevice spif(PB_5, PB_4, PB_3, PB_10); // mosi,miso,clk,cs,freq(opt)

The problem is that pins PB_3/4/5 are shared for SPI1 and 3. (See datasheet STM32F474RE p75).
When I use another SPI device like the display:

// mosi,miso,sclk,csel,reset,sda,scl,irq,name
RA8875 lcd(PA_7, PA_6, PA_5, PB_6, NC, PB_9, PB_8, PC_7, "tft");

PA5/6/7 als also using SPI1.

The result is that accessing the SPI flash memory is not longer possible after configuration of the LCD.
They probably use both the SPI1 hardware and come in conflict.
The purpose should be that the one uses SPI1 and the other SPI3.
Or maybe having an option to select the hardware peripheral to not be confused by a multi slave config.

Target(s) affected by this defect ?

NUCLEO_G474RE

Toolchain(s) (name and version) displaying this defect ?

GCC_ARM

What version of Mbed-os are you using (tag or sha) ?

mbed-os-6.2.0

a2ada74770f043aff3e61e29d164a8e78274fcd4

What version(s) of tools are you using. List all that apply (E.g. mbed-cli)

mbed-cli 1.10.4

How is this defect reproduced ?

Setup a project:

// mosi,miso,sclk,csel,reset,sda,scl,irq,name
RA8875 lcd(PA_7, PA_6, PA_5, PB_6, NC, PB_9, PB_8, PC_7, "tft");
SPIFBlockDevice spif(PB_5, PB_4, PB_3, PB_10); // mosi,miso,clk,cs,freq(opt)

int main()
{
    lcd.Reset();
    printf("spif type: %s\n", spif.get_type());
}

[LMESTM]

the available pin mapping options are defined here
https://github.com/ARMmbed/mbed-os/blob/master/targets/TARGET_STM/TARGET_STM32G4/TARGET_STM32G474xx/TARGET_NUCLEO_G474RE/PeripheralPins.c#L309

For the kind of case you refer to, there is no easy way to guess whether user want to use SPI3.
So by default PB3/4/5 are mapped to SPI1
In your paticular case, you can instead refer to
PB_4_ALT0, PB_5_ALT0, PB_6_ALT0
and they will be mapped to SPI3 as you can see in the table.

SPIFBlockDevice spif(PB_5_ALT0, PB_4_ALT0, PB_3_ALT0, PB_10); // mosi,miso,clk,cs,freq(opt)

should do the job

[lefebvresam]

Thank you. I tried to solve this issue by rewiring the hardware pins to avoid this ambiguity. I'm still working on a development board. But however because it can happen unexpectedly elsewhere it can be a suggestion to make a note of it on the documentation site.

[LMESTM]

Thank you. I tried to solve this issue by rewiring the hardware pins to avoid this ambiguity. I'm still working on a development board. But however because it can happen unexpectedly elsewhere it can be a suggestion to make a note of it on the documentation site.

This method might be specific to ST targets I am not sure .... Where would you recommend to put the information for users to easily find it ?

[lefebvresam]

Thank you. I tried to solve this issue by rewiring the hardware pins to avoid this ambiguity. I'm still working on a development board. But however because it can happen unexpectedly elsewhere it can be a suggestion to make a note of it on the documentation site.

This method might be specific to ST targets I am not sure .... Where would you recommend to put the information for users to easily find it ?

It's maybe an issue that can happen with all peripherals that use SPI. Maybe it can be recommended to make a separate page for it and put a link to from SPIFBlockDevice.

[lefebvresam]

It is info that belongs to the explanation of the constructor. Or maybe a better idea is to put this info on the SPI page and link to it.

[LMESTM]

This method does not apply only to SPI, the same is valid for PWM, ADC, ETH ...
I understand the need, but finding the proper place is not simple (@jeromecoutant maybe this is already documented in ST pages ?)

[lefebvresam]

In my opinion all devices that can be 'hardware switched' must have a small link to this explanation from the constructor info doc, because everybody who use it can encounter this problem and they not always know how to deal with it.

[jeromecoutant]

Hi
I will try to add info in https://github.com/ARMmbed/mbed-os/blob/master/targets/TARGET_STM/README.md

[ciarmcom]

Thank you for raising this detailed GitHub issue. I am now notifying our internal issue triagers.
Internal Jira reference: https://jira.arm.com/browse/MBOTRIAGE-2806

[jeromecoutant]

FYI: #13513

Maybe we could close this issue ?
Thx

[0xc0170]

I'll close as resolved via #13513 (it's readme update that will be merged soon).

@lefebvresam if not yet resolved, reopen with an update what we have missed.

[0xc0170]

Thanks for reporting!

[lefebvresam]

OK, if the documentation is updated, this issue can be closed. My solution was to swap the pins. Thx