Re: rfc: high-resolution timer framework

[christos%astron.com@localhost (Christos Zoulas)]

In article <20080716074303.GV22423%shisha.spb.ru@localhost>,
Alexander Shishkin  <alexander.shishkin%teleca.com@localhost> wrote:
>On Tue, Jul 15, 2008 at 02:41:30PM +0000, Christos Zoulas wrote:
>> Thanks, I will take a look.
>Now that those changes are applied, I updated the high-resolution timer
>patchset [1]. I could still use a hand in solving the locking trick and
>also let's discuss the remaining issues and/or concerns about the
>patchset.
>
>[1]: http://koowaldah.org/people/ash/netbsd/160708/

a. Shouldn't the members

        pt_rb_node,
        pt_on_rb,
        pt_cpuidx,
        pt_hrtim,

   be inside the union since they are only used in realtime timers?

b. Is pt_on_rb necessary? perhaps a -1 cpuid can indicate this?
c. why does hrtimerfix limit the minimum sleep time to 1000ns [1us]?
d. tc_hrtim_getbest() looks very similar to tc_pick. Perhaps they can
   be merged. Also, can't you cache the result, so that you don't need
   to look each time?
e. Perhaps you want to re-order the members of lapic timer so that the
   64 bit members are quadword aligned?

christos