Re: [dpdk-dev] [PATCH 1/2] timer: use rte_mp_msg to pass TSC hz to secondary procs

2019-08-26 Thread Bruce Richardson
On Thu, Aug 15, 2019 at 04:27:35AM -0700, Jim Harris wrote: > rte_eal_init() is much faster in secondary processes since > hugepages don't need to be zeroed. But there's still > non-trivial delays in the timer subsystem initialization > due to the 100ms sleep used to calculate TSC hz. So use > th

Re: [dpdk-dev] [PATCH 1/2] timer: use rte_mp_msg to pass TSC hz to secondary procs

2019-08-16 Thread Harris, James R
On 8/16/19, 12:56 AM, "Burakov, Anatoly" wrote: > @@ -89,6 +96,65 @@ set_tsc_freq(void) > eal_tsc_resolution_hz = freq; > } > > +static void > +set_tsc_freq_secondary(void) > +{ > + struct rte_mp_msg mp_req; > + struct rte_mp_re

Re: [dpdk-dev] [PATCH 1/2] timer: use rte_mp_msg to pass TSC hz to secondary procs

2019-08-16 Thread Burakov, Anatoly
On 15-Aug-19 12:27 PM, Jim Harris wrote: rte_eal_init() is much faster in secondary processes since hugepages don't need to be zeroed. But there's still non-trivial delays in the timer subsystem initialization due to the 100ms sleep used to calculate TSC hz. So use the rte_mp_msg framework to a

[dpdk-dev] [PATCH 1/2] timer: use rte_mp_msg to pass TSC hz to secondary procs

2019-08-15 Thread Jim Harris
rte_eal_init() is much faster in secondary processes since hugepages don't need to be zeroed. But there's still non-trivial delays in the timer subsystem initialization due to the 100ms sleep used to calculate TSC hz. So use the rte_mp_msg framework to allow secondary processes to get the TSC hz