On 08/26/2015 09:50 PM, Vincent Guittot wrote:
I agree to this could save power but i don't agree with latency. We have to wait for the end of interruption which can be far longer than an exit from wfi
That depends on the different cpus. For general type, the exit from wfi is usually several hundreds ms. do you have data for how long time a
several hundreds of ms for wfi (Wait For Interrupt) ? it should be more in the range of few micro seconds. At the opposite, waking up from a cluster power down state could take some ms but
Which platform have you got in mind that needs several hundreds of ms to leave WFI or another power down state ?
Uh, my mistaken, the usual idle exit_latency unit is us, not ms. but its span is so big, from 10us to 5000us (exynos coupled idle) or 10000us(one of intel cpu).
Uh, seems it need to check if there some latency lost on this patch.