lkml.org 
[lkml]   [2015]   [Jun]   [2]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [patch 0/7] timers: Footprint diet and NOHZ overhead mitigation
On Wed, 27 May 2015, Eric Dumazet wrote:
> I finally got back two hosts with 72 cpus where I can try your patches.

Waiting for your results :)

> I have one comment about socket TCP timers : They are re-armed very
> often, but they are never canceled (unless the socket is dismantled)
>
> I am too lazy to search for the commit adding this stuff (before git),
> but I guess it helped a lot RPC sessions : No need to cancel a timer if
> we're likely to rearm it in the following 10 usec ;)

Right, but that's a 'cancel + rearm' in the timer core code, unless
the expiry time does not change.

Thanks,

tglx



\
 
 \ /
  Last update: 2015-06-02 16:21    [W:0.202 / U:0.052 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site