On 02-Mar-18 6:48 PM, Stephen Hemminger wrote:
On Tue, 27 Feb 2018 14:59:29 +0000 Anatoly Burakov <anatoly.bura...@intel.com> wrote:+rte_mp_request_async(struct rte_mp_msg *req, const struct timespec *ts) { + struct sync_request *dummy; + struct async_request_shared_param *param = NULL; + struct rte_mp_reply *reply = NULL; + int dir_fd, ret = 0; + DIR *mp_dir; + struct dirent *ent; + struct timeval now; + struct timespec *end = NULL; + + RTE_LOG(DEBUG, EAL, "request: %s\n", req->name); + + if (check_input(req) == false) + return -1; + if (gettimeofday(&now, NULL) < 0) { + RTE_LOG(ERR, EAL, "Faile to get current time\n"); + rte_errno = errno; + return -1; + }gettimeofday is not a good API to use in DPDK. It gets changed by NTP; if you have to use system time you want monotonic clock
We need current time because pthread_cond_timedwait() accepts current time. So it's either that, or reimplementing pthread_cond_timedwait() in DPDK using monotonic clock :) Unless, of course, there already are alternatives that use monotonic clock and that don't need other DPDK machinery (e.g. rte_malloc) to work (like rte_alarm callbacks).
-- Thanks, Anatoly
