Re: [PATCH V4] Switch from select() to poll() in xenconsoled's IO loop

[Mats Petersson <mats.petersson@citrix.com>]

On 07/01/13 15:01, Wei Liu wrote:
> On Mon, 2013-01-07 at 14:41 +0000, Mats Petersson wrote:
>
>>>                           return;
>>> @@ -982,11 +1024,7 @@ void handle_io(void)
>>>                   /* Re-calculate any event counter allowances & unblock
>>>                      domains with new allowance */
>>>                   for (d = dom_head; d; d = d->next) {
>>> -                       /* Add 5ms of fuzz since select() often returns
>>> -                          a couple of ms sooner than requested. Without
>>> -                          the fuzz we typically do an extra spin in select()
>>> -                          with a 1/2 ms timeout every other iteration */
>>> -                       if ((now+5) > d->next_period) {
>>> +                       if (now > d->next_period) {
>> Is poll more accurate than select? I would have thought that they were
>> based on the same timing, and thus equally "fuzzy"?
> Is there any actual proof that the fuzz is needed? Specs of both
> select() and poll() don't seem to mention this behaviour at all.
That's a good question. I don't know. The tricky part with this sort of 
thing is that it may well depend on configurations, hardware 
differences, etc, so you may find that it works just fine on your 
test-box, but some big customer with Another-brand Co's servers don't 
work, because there is some subtle difference in hardware. Or it stops 
working if you have more than X number of CPU's. If you are convinced 
it's fine as it is, then by all means. I'm just thinking that it 
probably wasn't put there "by accident".

--
Mats
>
>>>    #include "utils.h"
>>>
>>>    struct xs_handle *xs;
>>> +struct pollfd *xs_pollfd;
>> I don't see this used outside of io.c  - am I missing something?
>>
>> Adding more dependencies between different source files seems
>> unnecessary...
>>
> Fixed.
>
>
> Wei.
>
>