Re: ip6stat percpu counters

2017-02-01 Thread Alexander Bluhm
On Tue, Jan 31, 2017 at 02:21:46PM +0100, Jeremie Courreges-Anglas wrote: > > Updated diff that addresses Martin's points: > - kill IPv6 mbuf stats; don't change the ip6stat structure yet - there > are already other fields that are useless > - add ip6stat_add() > > Keep the M_COUNTERS change

Re: ip6stat percpu counters

2017-02-01 Thread Jeremie Courreges-Anglas
Jeremie Courreges-Anglas writes: > Updated diff that addresses Martin's points: > - kill IPv6 mbuf stats; don't change the ip6stat structure yet - there > are already other fields that are useless > - add ip6stat_add() > > Keep the M_COUNTERS change separate for now, I can

Re: ip6stat percpu counters

2017-01-31 Thread Jeremie Courreges-Anglas
Updated diff that addresses Martin's points: - kill IPv6 mbuf stats; don't change the ip6stat structure yet - there are already other fields that are useless - add ip6stat_add() Keep the M_COUNTERS change separate for now, I can send an updated diff later if needed. Additionally, - stop

Re: ip6stat percpu counters

2017-01-31 Thread Martin Pieuchot
On 31/01/17(Tue) 12:16, Jeremie Courreges-Anglas wrote: > Martin Pieuchot writes: > [...] > Yes, but I don't know which threshold warrants introducing a new > function. So far those are the first cases for counters_dec(); that > includes wip rip6stat, icmp6stat, icmpstat and

Re: ip6stat percpu counters

2017-01-31 Thread Jeremie Courreges-Anglas
Martin Pieuchot writes: > On 30/01/17(Mon) 22:14, Jeremie Courreges-Anglas wrote: >> Very lightly tested. I used the existing percpu counters as examples. > > I like it. > >> I don't like the hardcoding of "32" in ip6_input() but I am not sure how >> to solve it nicely; the

Re: ip6stat percpu counters

2017-01-31 Thread Martin Pieuchot
On 30/01/17(Mon) 22:14, Jeremie Courreges-Anglas wrote: > Very lightly tested. I used the existing percpu counters as examples. I like it. > I don't like the hardcoding of "32" in ip6_input() but I am not sure how > to solve it nicely; the easiest way would be to just kill those > ipv6-specific

ip6stat percpu counters

2017-01-30 Thread Jeremie Courreges-Anglas
Very lightly tested. I used the existing percpu counters as examples. I don't like the hardcoding of "32" in ip6_input() but I am not sure how to solve it nicely; the easiest way would be to just kill those ipv6-specific mbuf stats. Would anyone miss them? Mbuf statistics: