Hello Andrew, > On 11/26/20 7:43 PM, Gregory Etelson wrote: > > According to RTE flow user guide, PMD will not keep flow rules after > > port stop. Application resources that refer to flow rules become > > obsolete after port stop and must not be used. > > Testpmd maintains linked list of active flows for each port. Entries > > in that list are allocated dynamically and must be explicitly released > > to prevent memory leak. > > The patch releases testpmd port flow_list that holds remaining flows > > before port is stopped. > > > > Cc: sta...@dpdk.org > > > > Signed-off-by: Gregory Etelson <getel...@nvidia.com> > > --- > > app/test-pmd/testpmd.c | 3 +++ > > 1 file changed, 3 insertions(+) > > > > diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c index > > 33fc0fddf5..0bb192b2f5 100644 > > --- a/app/test-pmd/testpmd.c > > +++ b/app/test-pmd/testpmd.c > > @@ -2806,6 +2806,9 @@ stop_port(portid_t pid) > > } > > } > > > > + if (port->flow_list) > > + port_flow_flush(pi); > > + > > if (rte_eth_dev_stop(pi) != 0) > > RTE_LOG(ERR, EAL, "rte_eth_dev_stop failed for > port %u\n", > > pi); > > > > port_flow_flush() does rte_flow_flush() which is not strictly required. > Description sounds like we should cleanup testpmd lists only.
You are right, call to rte_flow_flush() is not required, if testpmd calls port_flow_flush() as part of port stop sequence, because PMD will remove flows in that scenario. port_flow_flush() has a general implementation. It destroys all flows without port state consideration - current or future. In this form, port_flow_flush() can be called from any testpmd scenario that needs flows destruction. Regards, Gregory
