| Message ID | 20220610150923.583202-9-maciej.fijalkowski@intel.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | BPF |
| Headers | show |
| Series | AF_XDP ZC selftests | expand |
On Fri, Jun 10, 2022 at 5:40 PM Maciej Fijalkowski <maciej.fijalkowski@intel.com> wrote: > > Some of the drivers that implement support for AF_XDP Zero Copy (like > ice) can have lazy approach for cleaning Tx descriptors. For ZC, when > descriptor is cleaned, it is placed onto AF_XDP completion queue. This > means that current implementation of wait_for_tx_completion() in > xdpxceiver can get onto infinite loop, as some of the descriptors can > never reach CQ. > > This function can be changed to rely on pkts_in_flight instead. Acked-by: Magnus Karlsson <magnus.karlsson@intel.com> > Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com> > --- > tools/testing/selftests/bpf/xdpxceiver.c | 3 ++- > tools/testing/selftests/bpf/xdpxceiver.h | 2 +- > 2 files changed, 3 insertions(+), 2 deletions(-) > > diff --git a/tools/testing/selftests/bpf/xdpxceiver.c b/tools/testing/selftests/bpf/xdpxceiver.c > index 316f1dd338fc..c9385690af09 100644 > --- a/tools/testing/selftests/bpf/xdpxceiver.c > +++ b/tools/testing/selftests/bpf/xdpxceiver.c > @@ -965,7 +965,7 @@ static int __send_pkts(struct ifobject *ifobject, u32 *pkt_nb) > > static void wait_for_tx_completion(struct xsk_socket_info *xsk) > { > - while (xsk->outstanding_tx) > + while (pkts_in_flight) > complete_pkts(xsk, BATCH_SIZE); > } > > @@ -1269,6 +1269,7 @@ static void *worker_testapp_validate_rx(void *arg) > pthread_mutex_unlock(&pacing_mutex); > } > > + pkts_in_flight = 0; > pthread_exit(NULL); > } > > diff --git a/tools/testing/selftests/bpf/xdpxceiver.h b/tools/testing/selftests/bpf/xdpxceiver.h > index b7aa6c7cf2be..f364a92675f8 100644 > --- a/tools/testing/selftests/bpf/xdpxceiver.h > +++ b/tools/testing/selftests/bpf/xdpxceiver.h > @@ -170,6 +170,6 @@ pthread_barrier_t barr; > pthread_mutex_t pacing_mutex = PTHREAD_MUTEX_INITIALIZER; > pthread_cond_t pacing_cond = PTHREAD_COND_INITIALIZER; > > -int pkts_in_flight; > +volatile int pkts_in_flight; > > #endif /* XDPXCEIVER_H */ > -- > 2.27.0 >
diff --git a/tools/testing/selftests/bpf/xdpxceiver.c b/tools/testing/selftests/bpf/xdpxceiver.c index 316f1dd338fc..c9385690af09 100644 --- a/tools/testing/selftests/bpf/xdpxceiver.c +++ b/tools/testing/selftests/bpf/xdpxceiver.c @@ -965,7 +965,7 @@ static int __send_pkts(struct ifobject *ifobject, u32 *pkt_nb) static void wait_for_tx_completion(struct xsk_socket_info *xsk) { - while (xsk->outstanding_tx) + while (pkts_in_flight) complete_pkts(xsk, BATCH_SIZE); } @@ -1269,6 +1269,7 @@ static void *worker_testapp_validate_rx(void *arg) pthread_mutex_unlock(&pacing_mutex); } + pkts_in_flight = 0; pthread_exit(NULL); } diff --git a/tools/testing/selftests/bpf/xdpxceiver.h b/tools/testing/selftests/bpf/xdpxceiver.h index b7aa6c7cf2be..f364a92675f8 100644 --- a/tools/testing/selftests/bpf/xdpxceiver.h +++ b/tools/testing/selftests/bpf/xdpxceiver.h @@ -170,6 +170,6 @@ pthread_barrier_t barr; pthread_mutex_t pacing_mutex = PTHREAD_MUTEX_INITIALIZER; pthread_cond_t pacing_cond = PTHREAD_COND_INITIALIZER; -int pkts_in_flight; +volatile int pkts_in_flight; #endif /* XDPXCEIVER_H */
Some of the drivers that implement support for AF_XDP Zero Copy (like ice) can have lazy approach for cleaning Tx descriptors. For ZC, when descriptor is cleaned, it is placed onto AF_XDP completion queue. This means that current implementation of wait_for_tx_completion() in xdpxceiver can get onto infinite loop, as some of the descriptors can never reach CQ. This function can be changed to rely on pkts_in_flight instead. Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com> --- tools/testing/selftests/bpf/xdpxceiver.c | 3 ++- tools/testing/selftests/bpf/xdpxceiver.h | 2 +- 2 files changed, 3 insertions(+), 2 deletions(-)