Building an SdCard active object on the i.MX RT1062: hardware card detect, the blocking-API problem that forces a worker task, and the coupled CD line that made detection fail unpredictably until I stopped trusting the interrupt.
How I configured FlexSPI for a Winbond W25Q128JV, measured 518 Mbps read (97% of theoretical), and why the write measurement that looked great was actually a bug in my busy-poll loop.
The three interrupt sources follow one NVIC pattern but differ in where the trigger lives and how the flag clears. Plus the audit that matters: which of your ISRs sit above the FreeRTOS syscall barrier, and therefore must never touch a FreeRTOS API.