I was recently hacking on one of my drivers which was reporting that its
hardware experienced a FIFO overflow. To me this was a symptom that
interrupt latency was exceptionally high. Not long afterwards, I found
myself in uart_console_write() inside a spinlock (with interrupts
disabled), with the UART's transmitter empty interrupt disabled, and
POLLING the characters out. Obviously, if there are many characters
pending transmission, this would do atrocious things to the interrupt
latency.
Can someone point me at an explanation of uart_write() vs
uart_console_write()? How/when do I find myself in the latter
function? Can I rewrite this to use interrupt-driven output?
Thanks!
-Miles
--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ
