diff options
| author | Petri Savolainen <petri.savolainen@linaro.org> | 2018-07-04 17:17:46 +0300 |
|---|---|---|
| committer | Maxim Uvarov <maxim.uvarov@linaro.org> | 2018-07-11 16:00:45 +0300 |
| commit | c7bbbb603fc4cf2e6590af962e2e255241d18276 (patch) | |
| tree | 4b026ac38b074881841ef10e7684dc01a9740d4f | |
| parent | c0d530e01e91fbea195993293eed60f1677e6181 (diff) | |
linux-gen: sched: support large burst size from pktin
When directly receiving from pktin (atomic or parallel queues),
output event handles directly to destination array (scheduler
stash or application's array). This avoids extra handle copy
and enables supports for large burst sizes.
Signed-off-by: Petri Savolainen <petri.savolainen@linaro.org>
Reviewed-by: Bill Fischofer <bill.fischofer@linaro.org>
Signed-off-by: Maxim Uvarov <maxim.uvarov@linaro.org>
| -rw-r--r-- | platform/linux-generic/odp_schedule_basic.c | 36 |
1 files changed, 23 insertions, 13 deletions
diff --git a/platform/linux-generic/odp_schedule_basic.c b/platform/linux-generic/odp_schedule_basic.c index 78ab61353..f45bc8f19 100644 --- a/platform/linux-generic/odp_schedule_basic.c +++ b/platform/linux-generic/odp_schedule_basic.c @@ -762,17 +762,29 @@ static inline int queue_is_pktin(uint32_t queue_index) return sched->queue[queue_index].poll_pktin; } -static inline int poll_pktin(uint32_t qi, int stash) +static inline int poll_pktin(uint32_t qi, int direct_recv, + odp_event_t ev_tbl[], int max_num) { - odp_buffer_hdr_t *b_hdr[BURST_SIZE]; - int pktio_index, pktin_index, num, num_pktin, i; + int pktio_index, pktin_index, num, num_pktin; + odp_buffer_hdr_t **hdr_tbl; int ret; void *q_int; + odp_buffer_hdr_t *b_hdr[BURST_SIZE]; + + hdr_tbl = (odp_buffer_hdr_t **)ev_tbl; + + if (!direct_recv) { + hdr_tbl = b_hdr; + + /* Limit burst to max queue enqueue size */ + if (max_num > BURST_SIZE) + max_num = BURST_SIZE; + } pktio_index = sched->queue[qi].pktio_index; pktin_index = sched->queue[qi].pktin_index; - num = sched_cb_pktin_poll(pktio_index, pktin_index, b_hdr, BURST_SIZE); + num = sched_cb_pktin_poll(pktio_index, pktin_index, hdr_tbl, max_num); if (num == 0) return 0; @@ -793,12 +805,8 @@ static inline int poll_pktin(uint32_t qi, int stash) return num; } - if (stash) { - for (i = 0; i < num; i++) - sched_local.stash_ev[i] = event_from_buf_hdr(b_hdr[i]); - + if (direct_recv) return num; - } q_int = qentry_from_index(qi); @@ -908,13 +916,16 @@ static inline int do_schedule_grp(odp_queue_t *out_queue, odp_event_t out_ev[], * priorities. Stop scheduling queue when pktio * has been stopped. */ if (pktin) { - int stash = !ordered; - int num_pkt = poll_pktin(qi, stash); + int direct_recv = !ordered; + int num_pkt; + + num_pkt = poll_pktin(qi, direct_recv, + ev_tbl, max_deq); if (odp_unlikely(num_pkt < 0)) continue; - if (num_pkt == 0 || !stash) { + if (num_pkt == 0 || !direct_recv) { ring_enq(ring, ring_mask, qi); break; } @@ -922,7 +933,6 @@ static inline int do_schedule_grp(odp_queue_t *out_queue, odp_event_t out_ev[], /* Process packets from an atomic or * parallel queue right away. */ num = num_pkt; - stashed = 1; } else { /* Remove empty queue from scheduling. * Continue scheduling the same priority |