Re: [PATCH net-next] tcp: add TCPMemoryPressuresChrono counter

2017-06-07 Thread Eric Dumazet
On Wed, 2017-06-07 at 14:54 -0400, David Miller wrote: > From: Eric Dumazet > Date: Tue, 06 Jun 2017 12:19:27 -0700 > > > @@ -320,18 +320,37 @@ struct tcp_splice_state { > > * All the __sk_mem_schedule() is of this nature: accounting > > * is strict, actions are advisory and have some latency

Re: [PATCH net-next] tcp: add TCPMemoryPressuresChrono counter

2017-06-07 Thread David Miller
From: Eric Dumazet Date: Tue, 06 Jun 2017 12:19:27 -0700 > @@ -320,18 +320,37 @@ struct tcp_splice_state { > * All the __sk_mem_schedule() is of this nature: accounting > * is strict, actions are advisory and have some latency. > */ > -int tcp_memory_pressure __read_mostly; > +unsigned long

[PATCH net-next] tcp: add TCPMemoryPressuresChrono counter

2017-06-06 Thread Eric Dumazet
From: Eric Dumazet DRAM supply shortage and poor memory pressure tracking in TCP stack makes any change in SO_SNDBUF/SO_RCVBUF (or equivalent autotuning limits) and tcp_mem[] quite hazardous. TCPMemoryPressures SNMP counter is an indication of tcp_mem sysctl limits being hit, but only tracking n