On Tue, Jun 7, 2022 at 2:26 PM fengchengwen <fengcheng...@huawei.com> wrote: > Hi Bruce, > > Could you please test freebsd platform? I think it also have > the same problem, but I hasn't freebsd enviorment.
I can reproduce a crash with FreeBSD 13. Moving rte_trace_save() and co earlier (as is done for Linux in this proposed patch) avoids the crash, so I can't tell the traces file is valid. dpdk@freebsd:~/dpdk $ git diff diff --git a/lib/eal/freebsd/eal.c b/lib/eal/freebsd/eal.c index a6b20960f2..26fbc91b26 100644 --- a/lib/eal/freebsd/eal.c +++ b/lib/eal/freebsd/eal.c @@ -893,11 +893,11 @@ rte_eal_cleanup(void) eal_get_internal_configuration(); rte_service_finalize(); rte_mp_channel_cleanup(); + rte_trace_save(); + eal_trace_fini(); /* after this point, any DPDK pointers will become dangling */ rte_eal_memory_detach(); rte_eal_alarm_cleanup(); - rte_trace_save(); - eal_trace_fini(); eal_cleanup_config(internal_conf); return 0; } -- David Marchand