How do I reset an SPI module and associated DMA streams, without resetting the whole microcontroller?

[arnold_w]

Posted on April 04, 2016 at 18:26The original post was too long to process during our migration. Please click on the attachment to read the original post.

[Tesla DeLorean]

Posted on April 04, 2016 at 18:56

Honestly I don't think you could pay me enough to want to wade into this mire...

You've got software written by people who don't use the hardware, you've got hardware designed by people who've not had to implement software drivers, or a practical system using their own design. In the industry we describe this as not eating the dog food. You can reset the peripherals at a bus/unit level. Like an asynchronous reset input of yore. #define     __DMA1_FORCE_RESET()   (RCC->AHBRSTR |= (RCC_AHBRSTR_DMA1RST)) #define     __DMA1_RELEASE_RESET()   (RCC->AHBRSTR &= ~ (RCC_AHBRSTR_DMA1RST)) The SPI+CRC seems like an unmitigated cluster, you have to track bytes, and change modes, and you have Rx/Tx channels that aren't bonded in any sensible way. The SPI really needs a byte/word counter. And a better thought out chip-select generator/handler.

https://en.wikipedia.org/wiki/Eating_your_own_dog_food

[Radosław]

Posted on April 04, 2016 at 19:34 >>#define     __DMA1_RELEASE_RESET()   (RCC->AHBRSTR &= ~(RCC_AHBRSTR_DMA1RST)) This will reset all streams

[Tesla DeLorean]

Posted on April 04, 2016 at 20:16

Indeed, but not the entire micro-controller. So marginally less worse.

A lot of the design here has been poorly thought through.

[Radosław]

Posted on April 04, 2016 at 20:26

Disabling stream will be enough.

[arnold_w]

Posted on April 06, 2016 at 09:40

The recommendation to use __DMA1_FORCE_RESET() and __DMA1_RELEASE_RESET() pointed me in the direction to try the following:

                  __SPI2_FORCE_RESET();                  __SPI2_RELEASE_RESET();This seems to work fine, now my SPI-module recovers just fine after a faulty transmission. Thank you for your help.