DMA runs "beside" any interrupts. During the execution of a single assembly instruction there are small time slices where the internal busses of the CPU are not used. DMA happens during these times. So your high responsive interrupt is not affected.
Because SPI is relatively slow compared to CPU clock there is no need to use DMA for that. As long as your time critical interrupt has got the highest priority it will not get disturbed by another int happening, but will happen within the handler of a lower prioritized handler.
Thanks that was what I was hoping that would be the case, but couldn't tell for sure from the documents.
I will try just using SPI from an interrupt initially then.
I was just thinking since the app is a pretty busy one that using the DMA might leave more free CPU cycles. But maybe it won't make much difference.