| Message ID | 20220524084235.17796-1-hkalra@marvell.com (mailing list archive) |
|---|---|
| State | Accepted |
| Delegated to: | Jerin Jacob |
| Headers | show |
| Series | [v3,01/11] event/octeontx: fix SSO fastpath | expand |
| Context | Check | Description |
|---|---|---|
| ci/checkpatch | success | coding style OK |
On Tue, May 24, 2022 at 2:13 PM Harman Kalra <hkalra@marvell.com> wrote: > > Segmentation fault is observed as soon as any dpdk application > with ethdev event is launched. Handling the event types > appropriately. > > Also fixing sub event type violation as in OCTEONTX event device we > use sub_event_type to store the ethernet port identifier when we receive > work from OCTEONTX ethernet device. This violates the event device spec > as sub_event_type should be 0 in the initial receive stage. > Set sub_event_type to 0 after copying the port id in single workslot > mode. > > Fixes: 8dc6c2f12ecf ("crypto/octeontx: add crypto adapter data path") Updated the git commit as follows and applied to dpdk-next-net-eventdev/for-main. Thanks event/octeontx: fix SSO fastpath Segmentation fault is observed as soon as any application with ethdev event is launched as ev->mbuf was not set properly. Fixing sub event type violation as in OCTEONTX event device sub_event_type is used to store the ethernet port identifier when work is received from OCTEONTX ethernet device. This violates the event device spec as sub_event_type should be 0 in the initial receive stage. Setting sub_event_type to 0 after copying the port id in single workslot mode. Fixes: 44a2cebbd489 ("crypto/octeontx: add crypto adapter data path") Cc: stable@dpdk.org Signed-off-by: Harman Kalra <hkalra@marvell.com> Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com> > > Signed-off-by: Harman Kalra <hkalra@marvell.com> > Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com> > --- > V3: > * fixing sub event type violation > V2: > * Seperated out a generic patch from the series > > drivers/event/octeontx/ssovf_worker.h | 24 +++++++++++++++--------- > 1 file changed, 15 insertions(+), 9 deletions(-) > > diff --git a/drivers/event/octeontx/ssovf_worker.h b/drivers/event/octeontx/ssovf_worker.h > index e6ee292688..57be476394 100644 > --- a/drivers/event/octeontx/ssovf_worker.h > +++ b/drivers/event/octeontx/ssovf_worker.h > @@ -179,16 +179,22 @@ ssows_get_work(struct ssows *ws, struct rte_event *ev, const uint16_t flag) > ev->event = sched_type_queue | (get_work0 & 0xffffffff); > > if (get_work1) { > - if (ev->event_type == RTE_EVENT_TYPE_ETHDEV) > - get_work1 = (uintptr_t)ssovf_octeontx_wqe_to_pkt( > - get_work1, (ev->event >> 20) & 0x7F, flag, > - ws->lookup_mem); > - else if (ev->event_type == RTE_EVENT_TYPE_CRYPTODEV) > + if (ev->event_type == RTE_EVENT_TYPE_ETHDEV) { > + uint16_t port = (ev->event >> 20) & 0x7F; > + > + ev->sub_event_type = 0; > + ev->mbuf = ssovf_octeontx_wqe_to_pkt( > + get_work1, port, flag, ws->lookup_mem); > + } else if (ev->event_type == RTE_EVENT_TYPE_CRYPTODEV) { > get_work1 = otx_crypto_adapter_dequeue(get_work1); > - ev->u64 = get_work1; > - } else if (unlikely((get_work0 & 0xFFFFFFFF) == 0xFFFFFFFF)) { > - ssovf_octeontx_wqe_free(get_work1); > - return 0; > + ev->u64 = get_work1; > + } else { > + if (unlikely((get_work0 & 0xFFFFFFFF) == 0xFFFFFFFF)) { > + ssovf_octeontx_wqe_free(get_work1); > + return 0; > + } > + ev->u64 = get_work1; > + } > } > > return !!get_work1; > -- > 2.18.0 >
diff --git a/drivers/event/octeontx/ssovf_worker.h b/drivers/event/octeontx/ssovf_worker.h index e6ee292688..57be476394 100644 --- a/drivers/event/octeontx/ssovf_worker.h +++ b/drivers/event/octeontx/ssovf_worker.h @@ -179,16 +179,22 @@ ssows_get_work(struct ssows *ws, struct rte_event *ev, const uint16_t flag) ev->event = sched_type_queue | (get_work0 & 0xffffffff); if (get_work1) { - if (ev->event_type == RTE_EVENT_TYPE_ETHDEV) - get_work1 = (uintptr_t)ssovf_octeontx_wqe_to_pkt( - get_work1, (ev->event >> 20) & 0x7F, flag, - ws->lookup_mem); - else if (ev->event_type == RTE_EVENT_TYPE_CRYPTODEV) + if (ev->event_type == RTE_EVENT_TYPE_ETHDEV) { + uint16_t port = (ev->event >> 20) & 0x7F; + + ev->sub_event_type = 0; + ev->mbuf = ssovf_octeontx_wqe_to_pkt( + get_work1, port, flag, ws->lookup_mem); + } else if (ev->event_type == RTE_EVENT_TYPE_CRYPTODEV) { get_work1 = otx_crypto_adapter_dequeue(get_work1); - ev->u64 = get_work1; - } else if (unlikely((get_work0 & 0xFFFFFFFF) == 0xFFFFFFFF)) { - ssovf_octeontx_wqe_free(get_work1); - return 0; + ev->u64 = get_work1; + } else { + if (unlikely((get_work0 & 0xFFFFFFFF) == 0xFFFFFFFF)) { + ssovf_octeontx_wqe_free(get_work1); + return 0; + } + ev->u64 = get_work1; + } } return !!get_work1;
Segmentation fault is observed as soon as any dpdk application with ethdev event is launched. Handling the event types appropriately. Also fixing sub event type violation as in OCTEONTX event device we use sub_event_type to store the ethernet port identifier when we receive work from OCTEONTX ethernet device. This violates the event device spec as sub_event_type should be 0 in the initial receive stage. Set sub_event_type to 0 after copying the port id in single workslot mode. Fixes: 8dc6c2f12ecf ("crypto/octeontx: add crypto adapter data path") Signed-off-by: Harman Kalra <hkalra@marvell.com> Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com> --- V3: * fixing sub event type violation V2: * Seperated out a generic patch from the series drivers/event/octeontx/ssovf_worker.h | 24 +++++++++++++++--------- 1 file changed, 15 insertions(+), 9 deletions(-)