Re: [RFC] Dynamic Tick and Deferrable Timer Support

[Jon Hunter]

Pallipadi, Venkatesh wrote:

> max_delta would depend on the timer in the platform. With HPET this
> should be much larger than 2.15 secs.

So I agree that the HPET hardware in newer devices themselves would 
allow longer sleep periods. However, this is not the problem I was raising.

The problem is that the dynamic tick uses a 32-bit variable, 
max_delta_ns, to define that max sleep time of a device in nanoseconds. 
The maximum value that this variable can be assigned is LONG_MAX or 
0x7fffffff nanoseconds (see function clockevent_delta2ns). The value 
0x7fffffff nanoseconds equates to ~2.15 seconds. Hence, without 
increasing the dynamic range of max_delta_ns (ie. make this a 64-bit 
integer) or change the base of this variable from nanoseconds to 
milliseconds, I don't see how the device will ever sleep for longer than 
~2.15 seconds.

I have spent several weeks trying to suppress kernel timers using the 
deferred timers and lengthen the sleep time. I am now able to get the 
device to sleep for minutes but I found that max_delta_ns is a limiting 
factor. I will be surprised if you can sleep for longer than ~2.15 
seconds with the current implementation.

Let me know if this makes sense.

> Ok. ïThinking about it a bit more, I think we can push this 
> patch along.
> Thomas/Andrew, can one of you pick up this patch..

Great thanks.

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