> The issue with the CS being held low so long makes me think there is a > little bit more to it.
If you look at function rspi_setup, you'll see that bit SPCMD_SSLKP is set so the CS will stay low after every transfer until SW comes back in and manually puts it back high. I would guess the delay you are seeing there is a kernel scheduling delay of letting the driver resume running again after the wait_event_timeout call. Chris
