Re: [lng-odp] [API-NEXT PATCHv3 4/6] test/validation/traffic_mngr: Add the new traffic_mngr tests.
On 02/18/16 05:03, Bill Fischofer wrote: +static void flush_leftover_pkts(odp_tm_t odp_tm, odp_pktio_t pktio) +{ + odp_packet_t rcv_pkt; + odp_time_t start_time, current_time, duration; + uint64_t min_timeout_ns, max_timeout_ns, duration_ns; + int rc; + + /* Set the timeout to be at least 10 milliseconds and at most 100 +* milliseconds */ + min_timeout_ns = 1000ULL; 10 * ODP_TIME_MSEC_IN_NS + max_timeout_ns = 1ULL; 100 * ODP_TIME_MSEC_IN_NS + start_time = odp_time_local(); ___ lng-odp mailing list lng-odp@lists.linaro.org https://lists.linaro.org/mailman/listinfo/lng-odp
Re: [lng-odp] [API-NEXT PATCHv3 4/6] test/validation/traffic_mngr: Add the new traffic_mngr tests.
Agreed. Again, this was just a rebase of Barry's earlier patch that predated that routine. I can post a follow-on patch to address this. On Thu, Feb 18, 2016 at 6:52 AM, Maxim Uvarovwrote: > On 02/18/16 05:03, Bill Fischofer wrote: > >> +static void busy_wait(uint64_t nanoseconds) >> +{ >> + odp_time_t start_time, end_time; >> + uint32_t cnt; >> + >> + start_time = odp_time_local(); >> + end_time = odp_time_sum(start_time, >> + odp_time_local_from_ns(nanoseconds)); >> + >> + while (odp_time_cmp(odp_time_local(), end_time) < 0) >> + for (cnt = 1; cnt <= 10; cnt++) >> + busy_wait_counter++; >> +} >> > incrementing cnt looks like hack. I think odp_cpu_pause() can be used here. > > Maxim. > ___ > lng-odp mailing list > lng-odp@lists.linaro.org > https://lists.linaro.org/mailman/listinfo/lng-odp > ___ lng-odp mailing list lng-odp@lists.linaro.org https://lists.linaro.org/mailman/listinfo/lng-odp
Re: [lng-odp] [API-NEXT PATCHv3 4/6] test/validation/traffic_mngr: Add the new traffic_mngr tests.
On 02/18/16 05:03, Bill Fischofer wrote: + odp_pool_param_init(_param); + odp_pktio_param_init(_param); + + pktio_param.in_mode = ODP_PKTIN_MODE_DIRECT; + pool_param.pkt.num = 10 * MAX_PKTS; + pool_param.type = ODP_POOL_PACKET; Just my preference to have init code as match as possible to setting code. Maxim. ___ lng-odp mailing list lng-odp@lists.linaro.org https://lists.linaro.org/mailman/listinfo/lng-odp
Re: [lng-odp] [API-NEXT PATCHv3 4/6] test/validation/traffic_mngr: Add the new traffic_mngr tests.
On 02/18/16 05:03, Bill Fischofer wrote: + +static odp_bool_t approx_eq32(uint32_t val, uint32_t correct) +{ + uint64_t low_bound, val_times_100, high_bound; + + if (val == correct) + return true; + + low_bound = 98 * (uint64_t)correct; + val_times_100 = 100 * (uint64_t)val; + high_bound= 102 * (uint64_t)correct; + + if ((low_bound <= val_times_100) && (val_times_100 <= high_bound)) + return true; + else + return false; +} usually checkpatch.pl warns that else word is not needed here. I just realized that in some places in code we use TRUE and FALSE, in other places it's true and false, and current odp_bool_t is int and api does not say which exactly values it wants to see. Maxim. ___ lng-odp mailing list lng-odp@lists.linaro.org https://lists.linaro.org/mailman/listinfo/lng-odp
Re: [lng-odp] [API-NEXT PATCHv3 4/6] test/validation/traffic_mngr: Add the new traffic_mngr tests.
On 02/18/16 05:03, Bill Fischofer wrote: +static void busy_wait(uint64_t nanoseconds) +{ + odp_time_t start_time, end_time; + uint32_t cnt; + + start_time = odp_time_local(); + end_time = odp_time_sum(start_time, + odp_time_local_from_ns(nanoseconds)); + + while (odp_time_cmp(odp_time_local(), end_time) < 0) + for (cnt = 1; cnt <= 10; cnt++) + busy_wait_counter++; +} incrementing cnt looks like hack. I think odp_cpu_pause() can be used here. Maxim. ___ lng-odp mailing list lng-odp@lists.linaro.org https://lists.linaro.org/mailman/listinfo/lng-odp