Thread (21 messages) flat view 21 messages, 3 authors, 2017-12-08

[PATCH v8 4/6] clocksource: stm32: only use 32 bits timers

From: Daniel Lezcano <hidden>
Date: 2017-12-08 07:52:18
Also in: linux-devicetree, lkml

On 07/12/2017 21:36, Benjamin Gaignard wrote:
2017-12-07 17:49 GMT+01:00 Daniel Lezcano [off-list ref]:
quoted
On 07/12/2017 17:33, Benjamin Gaignard wrote:
quoted
2017-12-07 16:27 GMT+01:00 Daniel Lezcano [off-list ref]:
quoted
On 14/11/2017 09:52, Benjamin Gaignard wrote:
quoted
The clock driving counters is at 90MHz so the maximum period
for 16 bis counters is around 750 ms which is a short period
for a clocksource.
Isn't it 728us ?
yes it is: 2^16 / 90.000.000 => 728us
Ok, now I can do the connection with the previous patch.

So, the real issue of all this is the 16bits clocksource is wrapping up
every 728us, hence the clockevent periodically expires every ~728us to
keep the timekeeping consistent. Unfortunately, the kernel has a too
high overhead for this as the system is consistently processing this
timer leading to a CPU time resource starvation.

Is that correct ?
Yes that is correct
Oh man. That was unclear since the beginning, we are not talking about
inaccurate clocksource or whatever but just these 16bits timers can't
work on Linux.


-- 
 <http://www.linaro.org/> Linaro.org ? Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help