On Fri, Aug 16, 2013 at 11:53:48AM +0100, Srinivas KANDAGATLA wrote:
> Thanks Sean for the comments.
> On 16/08/13 09:38, Sean Young wrote:
> > On Wed, Aug 14, 2013 at 06:27:01PM +0100, Srinivas KANDAGATLA wrote:
> [...]
> >> + /* discard the entire collection in case of errors! */
> >> + dev_info(dev->dev, "IR RX overrun\n");
> >> + writel(IRB_RX_OVERRUN_INT,
> >> + dev->rx_base + IRB_RX_INT_CLEAR);
> >> + continue;
> >> + }
> >> +
> >> + symbol = readl(dev->rx_base + IRB_RX_SYS);
> >> + mark = readl(dev->rx_base + IRB_RX_ON);
> >> +
> >> + if (symbol == IRB_TIMEOUT)
> >> + last_symbol = 1;
> >> +
> >> + /* Ignore any noise */
> >> + if ((mark > 2) && (symbol > 1)) {
> >> + symbol -= mark;
> >> + if (dev->overclocking) { /* adjustments to timings */
> >> + symbol *= dev->sample_mult;
> >> + symbol /= dev->sample_div;
> >> + mark *= dev->sample_mult;
> >> + mark /= dev->sample_div;
> >> + }
> >> +
> >> + ev.duration = US_TO_NS(mark);
> >> + ev.pulse = true;
> >> + ir_raw_event_store(dev->rdev, &ev);
> >> +
> >> + if (!last_symbol) {
> >> + ev.duration = US_TO_NS(symbol);
> >> + ev.pulse = false;
> >> + ir_raw_event_store(dev->rdev, &ev);
> >
> > Make sure you call ir_raw_event_handle() once a while (maybe every time
> > the interrupt handler is called?) to prevent the ir kfifo from
> > overflowing in case of very long IR. ir_raw_event_store() just adds
> > new edges to the kfifo() but does not flush them to the decoders or
> > lirc.
> I agree, but Am not sure it will really help in this case because, we
> are going to stay in this interrupt handler till we get a
> last_symbol(full key press/release event).. So calling
> ir_raw_event_store mulitple times might not help because the
> ir_raw_event kthread(which is clearing kfifo) which is only scheduled
> after returning from this interrupt.
If I read it correctly, then this is only true if the fifo contains an
IRB_TIMEOUT symbol. If not yet, then the interrupt handlers is not
waiting around for those symbols to arrive.
Thanks
Sean
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
