crypto/scheduler: fix session retrieval for ops
PATCH 1/1 crypto/scheduler: fix session retrieval for ops
In cases where some ops failed to enqueue, the op session was never being reset. This resulted in a segmentation fault when processing ops the next time. To fix this, only set the op session after the failure condition is checked.
Also, the incorrect ops index was being used for session retrieval when dequeueing for the secondary worker.
--- drivers/crypto/scheduler/scheduler_failover.c | 8 ++++- .../scheduler/scheduler_pkt_size_distr.c | 30 +++++++++---------- 2 files changed, 22 insertions(+), 16 deletions(-)
drivers/crypto/scheduler/scheduler_failover.c +7 -1
@@ -50,12 +50,18 @@ schedule_enqueue(void *qp, struct rte_crypto_op **ops, uint16_t nb_ops) enqueued_ops =failover_worker_enqueue (&qp_ctx->primary_worker, ops, nb_ops,PRIMARY_WORKER_IDX ); -if (enqueued_ops < nb_ops) +if (enqueued_ops < nb_ops) { +scheduler_retrieve_session (&ops[enqueued_ops], + nb_ops - enqueued_ops); enqueued_ops +=failover_worker_enqueue ( &qp_ctx->secondary_worker, &ops[enqueued_ops], nb_ops - enqueued_ops,SECONDARY_WORKER_IDX ); +if (enqueued_ops < nb_ops) +scheduler_retrieve_session (&ops[enqueued_ops], + nb_ops - enqueued_ops); + }return enqueued_ops; }
drivers/crypto/scheduler/scheduler_pkt_size_distr.c +15 -15
@@ -89,9 +89,6 @@ schedule_enqueue(void *qp, struct rte_crypto_op **ops, uint16_t nb_ops) ops[i]->sym->auth.data.length;/* decide the target op based on the job length */ target[0 ] = !(job_len[0 ] & psd_qp_ctx->threshold); -if (ops[i]->sess_type ==RTE_CRYPTO_OP_WITH_SESSION ) - ops[i]->sym->session = - sess_ctx[0 ]->worker_sess[target[0 ]]; p_enq_op = &enq_ops[target[0 ]];/* stop schedule cops before the queue is full, this shall
@@ -103,6 +100,9 @@ schedule_enqueue(void *qp, struct rte_crypto_op **ops, uint16_t nb_ops)break ; } +if (ops[i]->sess_type ==RTE_CRYPTO_OP_WITH_SESSION ) + ops[i]->sym->session = + sess_ctx[0 ]->worker_sess[target[0 ]]; sched_ops[p_enq_op->worker_idx][p_enq_op->pos] = ops[i]; p_enq_op->pos++;
@@ -110,9 +110,6 @@ schedule_enqueue(void *qp, struct rte_crypto_op **ops, uint16_t nb_ops) job_len[1 ] += (ops[i +1 ]->sym->cipher.data.length ==0 ) * ops[i+1 ]->sym->auth.data.length; target[1 ] = !(job_len[1 ] & psd_qp_ctx->threshold); -if (ops[i +1 ]->sess_type ==RTE_CRYPTO_OP_WITH_SESSION ) - ops[i +1 ]->sym->session = - sess_ctx[1 ]->worker_sess[target[1 ]]; p_enq_op = &enq_ops[target[1 ]]; if (p_enq_op->pos + in_flight_ops[p_enq_op->worker_idx] ==
@@ -121,6 +118,9 @@ schedule_enqueue(void *qp, struct rte_crypto_op **ops, uint16_t nb_ops)break ; } +if (ops[i +1 ]->sess_type ==RTE_CRYPTO_OP_WITH_SESSION ) + ops[i +1 ]->sym->session = + sess_ctx[1 ]->worker_sess[target[1 ]]; sched_ops[p_enq_op->worker_idx][p_enq_op->pos] = ops[i+1 ]; p_enq_op->pos++;
@@ -128,9 +128,6 @@ schedule_enqueue(void *qp, struct rte_crypto_op **ops, uint16_t nb_ops) job_len[2 ] += (ops[i +2 ]->sym->cipher.data.length ==0 ) * ops[i +2 ]->sym->auth.data.length; target[2 ] = !(job_len[2 ] & psd_qp_ctx->threshold); -if (ops[i +2 ]->sess_type ==RTE_CRYPTO_OP_WITH_SESSION ) - ops[i +2 ]->sym->session = - sess_ctx[2 ]->worker_sess[target[2 ]]; p_enq_op = &enq_ops[target[2 ]]; if (p_enq_op->pos + in_flight_ops[p_enq_op->worker_idx] ==
@@ -139,6 +136,9 @@ schedule_enqueue(void *qp, struct rte_crypto_op **ops, uint16_t nb_ops)break ; } +if (ops[i +2 ]->sess_type ==RTE_CRYPTO_OP_WITH_SESSION ) + ops[i +2 ]->sym->session = + sess_ctx[2 ]->worker_sess[target[2 ]]; sched_ops[p_enq_op->worker_idx][p_enq_op->pos] = ops[i+2 ]; p_enq_op->pos++;
@@ -146,9 +146,6 @@ schedule_enqueue(void *qp, struct rte_crypto_op **ops, uint16_t nb_ops) job_len[3 ] += (ops[i +3 ]->sym->cipher.data.length ==0 ) * ops[i +3 ]->sym->auth.data.length; target[3 ] = !(job_len[3 ] & psd_qp_ctx->threshold); -if (ops[i +3 ]->sess_type ==RTE_CRYPTO_OP_WITH_SESSION ) - ops[i +3 ]->sym->session = - sess_ctx[3 ]->worker_sess[target[3 ]]; p_enq_op = &enq_ops[target[3 ]]; if (p_enq_op->pos + in_flight_ops[p_enq_op->worker_idx] ==
@@ -157,6 +154,9 @@ schedule_enqueue(void *qp, struct rte_crypto_op **ops, uint16_t nb_ops)break ; } +if (ops[i +3 ]->sess_type ==RTE_CRYPTO_OP_WITH_SESSION ) + ops[i +3 ]->sym->session = + sess_ctx[3 ]->worker_sess[target[3 ]]; sched_ops[p_enq_op->worker_idx][p_enq_op->pos] = ops[i+3 ]; p_enq_op->pos++; }
@@ -171,8 +171,6 @@ schedule_enqueue(void *qp, struct rte_crypto_op **ops, uint16_t nb_ops) job_len += (ops[i]->sym->cipher.data.length ==0 ) * ops[i]->sym->auth.data.length; target = !(job_len & psd_qp_ctx->threshold); -if (ops[i]->sess_type ==RTE_CRYPTO_OP_WITH_SESSION ) - ops[i]->sym->session = sess_ctx->worker_sess[target]; p_enq_op = &enq_ops[target]; if (p_enq_op->pos + in_flight_ops[p_enq_op->worker_idx] ==
@@ -181,6 +179,8 @@ schedule_enqueue(void *qp, struct rte_crypto_op **ops, uint16_t nb_ops)break ; } +if (ops[i]->sess_type ==RTE_CRYPTO_OP_WITH_SESSION ) + ops[i]->sym->session = sess_ctx->worker_sess[target]; sched_ops[p_enq_op->worker_idx][p_enq_op->pos] = ops[i]; p_enq_op->pos++; }
@@ -251,7 +251,7 @@ schedule_dequeue(void *qp, struct rte_crypto_op **ops, uint16_t nb_ops) nb_deq_ops_sec =rte_cryptodev_dequeue_burst (worker->dev_id, worker->qp_id, &ops[nb_deq_ops_pri], nb_ops - nb_deq_ops_pri); -scheduler_retrieve_session (ops, nb_deq_ops_sec); +scheduler_retrieve_session (&ops[nb_deq_ops_pri], nb_deq_ops_sec); worker->nb_inflight_cops -= nb_deq_ops_sec;if (!worker->nb_inflight_cops)
Signature · 1 line
2.25.1
Akhil Goyal
AppliedQuoted 5 lines
In cases where some ops failed to enqueue, the op session was never being reset. This resulted in a segmentation fault when processing ops the next time. To fix this, only set the op session after the failure condition is checked.
Quoted 7 lines
Also, the incorrect ops index was being used for session retrieval when dequeueing for the secondary worker.
Fixes 6812b9bf470e ("crypto/scheduler: use unified session")Reported-by Kevin O'Sullivan <[email protected]>Signed-off-by Ciara Power <[email protected]>
Applied to dpdk-next-crypto
Thanks.