> Of course if you have to check of the DCD changed, that may not be
quoted text
> constant.  A check for whether this device currently has anyone that
> wants the timestamp though should be a constant time.

Not having to take the cost of that for other users is far more important
than a time variance of a jump prediction. CPU cache effects alone will
add way more variance already. It's not a meaningful number to chase
compared to the randomness introduced by the IRQ handling itself.

If you want perfect accuracy do what I²IT did years ago and use a card
which timestamps the physical IRQ bus transition and exposes it to the OS.

Alan
--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/