Andy Green wrote: > Yeah I remember the things about erratic interrupts, I didn't realize > that we worked around that now by polling in a loop that includes the > whole SDIO thing at 100Hz.
It's not even our own workaround - it's just what the SDIO stack does when a driver doesn't support "real" interrupts. drivers/mmc/core/sdio_irq.c:sdio_irq_thread is actually pretty clever and tries to predict when interrupts are likely to happen, so it polls more often when there's activity, then gently slows down again. - Werner
