How do I register the interrupt for DMA buffer handling?

[John Griswold]

Posted on September 27, 2017 at 22:55

I have serial data coming in at a high rate, so DMA is the only reasonable way to handle this firehose of data. I believe I have _nearly_ everything set up properly, but for the interrupt that signals DMA completion and the need to reset pointers and restart DMA. 

I know (or perhaps just think) I should ultimately get to DMA2_Channel7_IRQHandler() for the selections I've made, but the interrupt never occurs. So I don't think I've properly asked it to be triggered, and wonder just where I should do that.

All of the underpinnings were generated by STM32CubeMX version 4.22.0. Is this where my troubles begin? There seems to be a lot of anti-CubeMX sentiment in the ranks.

Until I get this interrupt properly set up, the 0x800 bytes of buffer fill up, and the pointers don't change. 

Please?

[John Griswold]

Posted on September 28, 2017 at 15:23

Digging deeper, I find that the interrupt in question is set up in this routine:

static void MX_DMA_Init(void)

{ /* DMA controller clock enable */ __HAL_RCC_DMA1_CLK_ENABLE(); __HAL_RCC_DMA2_CLK_ENABLE();

/* DMA interrupt init */

/* DMA1_Channel4_IRQn interrupt configuration */ HAL_NVIC_SetPriority(DMA1_Channel4_IRQn, 0, 0); HAL_NVIC_EnableIRQ(DMA1_Channel4_IRQn); /* DMA2_Channel7_IRQn interrupt configuration */ HAL_NVIC_SetPriority(DMA2_Channel7_IRQn, 0, 0); HAL_NVIC_EnableIRQ(DMA2_Channel7_IRQn);

}

so one might think that this should work. But the interrupt doesn't fire. I supposed it could be a hardware failure, but I need to exhaust all of my culpability before I jump down that rabbit hole.

Any ideas?

[John Griswold]

Posted on September 28, 2017 at 16:09

So, I jumped down that rabbit hole with a quick prototype, and with the new hardware, that interrupt fires OK. This is going to anger the hardware engineer  Oh, well, some things just can't be helped.

Thanks,John