rxq_cycle_sort is used to compare rx queues by their measured number of cycles. In the event that they are equal, 0 could be returned. However, it is observed that returning 0 results in a different sort order on Windows/Linux. This is ok in practice but it causes a unit test failure for "1007: PMD - pmd-cpu-mask/distribution of rx queues" when running on different OS's.
In order to have a consistent sort result across multiple OS's, introduce a tiebreaker of port/queue. Fixes: 655856ef39b9 ("dpif-netdev: Change rxq_scheduling to use rxq processing cycles.") Reported-by: Alin Gabriel Serdean <aserd...@ovn.org> Tested-by: Alin Gabriel Serdean <aserd...@ovn.org> Co-authored-by: Ilya Maximets <i.maxim...@samsung.com> Signed-off-by: Ilya Maximets <i.maxim...@samsung.com> Signed-off-by: Kevin Traynor <ktray...@redhat.com> --- V4: Added patch into series after Alin tested it as RFC lib/dpif-netdev.c | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-) diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c index 0a62630..f5cdd92 100644 --- a/lib/dpif-netdev.c +++ b/lib/dpif-netdev.c @@ -3464,8 +3464,17 @@ rxq_cycle_sort(const void *a, const void *b) dp_netdev_rxq_set_cycles(qb, RXQ_CYCLES_PROC_HIST, total_qb); - if (total_qa >= total_qb) { - return -1; + if (total_qa != total_qb) { + return (total_qa < total_qb) ? 1 : -1; + } else { + /* Cycles are the same so tiebreak on port/queue id. + * Tiebreaking (as opposed to return 0) ensures consistent + * sort results across multiple OS's. */ + if (qa->port->port_no != qb->port->port_no) { + return (qa->port->port_no > qb->port->port_no) ? 1 : -1; + } else { + return netdev_rxq_get_queue_id(qa->rx) + - netdev_rxq_get_queue_id(qb->rx); + } } - return 1; } -- 1.8.3.1 _______________________________________________ dev mailing list d...@openvswitch.org https://mail.openvswitch.org/mailman/listinfo/ovs-dev