net/octeontx: fix failures when available ports > queues

Message ID 1542187282-30800-1-git-send-email-anoob.joseph@caviumnetworks.com (mailing list archive)
State Accepted, archived
Delegated to: Ferruh Yigit
Headers
Series net/octeontx: fix failures when available ports > queues |

Checks

Context Check Description
ci/Intel-compilation success Compilation OK
ci/intel-Performance-Testing success Performance Testing PASS
ci/mellanox-Performance-Testing success Performance Testing PASS

Commit Message

Anoob Joseph Nov. 14, 2018, 9:22 a.m. UTC
  When eth_octeontx is initialized before event_octeontx, eth_octeontx
would initiate setting up of the event device with the max ports and
queues available. If number of ports is more than the number of queues,
some ports would be left unused when the ports and queues are mapped
1:1. But even in that case the ports need to be setup, or otherwise it
would lead to a segmentation fault when event device is started.

Fixes: f7be70e5130e ("net/octeontx: add net device probe and remove")

Signed-off-by: Anoob Joseph <anoob.joseph@caviumnetworks.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
---
 drivers/net/octeontx/octeontx_ethdev.c | 20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)
  

Comments

Jerin Jacob Nov. 14, 2018, 10:40 a.m. UTC | #1
-----Original Message-----
> Date: Wed, 14 Nov 2018 14:52:03 +0530
> From: "Joseph, Anoob" <Anoob.Joseph@cavium.com>
> To: Ferruh Yigit <ferruh.yigit@intel.com>, "Jacob,  Jerin"
>  <Jerin.JacobKollanukkaran@cavium.com>
> CC: "Joseph, Anoob" <Anoob.Joseph@cavium.com>, "Athreya, Narayana Prasad"
>  <NarayanaPrasad.Athreya@cavium.com>, "Bhagavatula, Pavan"
>  <Pavan.Bhagavatula@cavium.com>, "dev@dpdk.org" <dev@dpdk.org>,
>  "stable@dpdk.org" <stable@dpdk.org>
> Subject: [PATCH] net/octeontx: fix failures when available ports > queues
> 
> When eth_octeontx is initialized before event_octeontx, eth_octeontx
> would initiate setting up of the event device with the max ports and
> queues available. If number of ports is more than the number of queues,
> some ports would be left unused when the ports and queues are mapped
> 1:1. But even in that case the ports need to be setup, or otherwise it
> would lead to a segmentation fault when event device is started.
> 
> Fixes: f7be70e5130e ("net/octeontx: add net device probe and remove")
Cc: stable@dpdk.org
> 
> Signed-off-by: Anoob Joseph <anoob.joseph@caviumnetworks.com>
> Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
> ---

Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
  
Ferruh Yigit Nov. 16, 2018, 11:05 a.m. UTC | #2
On 11/14/2018 10:40 AM, Jerin Jacob wrote:
> -----Original Message-----
>> Date: Wed, 14 Nov 2018 14:52:03 +0530
>> From: "Joseph, Anoob" <Anoob.Joseph@cavium.com>
>> To: Ferruh Yigit <ferruh.yigit@intel.com>, "Jacob,  Jerin"
>>  <Jerin.JacobKollanukkaran@cavium.com>
>> CC: "Joseph, Anoob" <Anoob.Joseph@cavium.com>, "Athreya, Narayana Prasad"
>>  <NarayanaPrasad.Athreya@cavium.com>, "Bhagavatula, Pavan"
>>  <Pavan.Bhagavatula@cavium.com>, "dev@dpdk.org" <dev@dpdk.org>,
>>  "stable@dpdk.org" <stable@dpdk.org>
>> Subject: [PATCH] net/octeontx: fix failures when available ports > queues
>>
>> When eth_octeontx is initialized before event_octeontx, eth_octeontx
>> would initiate setting up of the event device with the max ports and
>> queues available. If number of ports is more than the number of queues,
>> some ports would be left unused when the ports and queues are mapped
>> 1:1. But even in that case the ports need to be setup, or otherwise it
>> would lead to a segmentation fault when event device is started.
>>
>> Fixes: f7be70e5130e ("net/octeontx: add net device probe and remove")
> Cc: stable@dpdk.org
>>
>> Signed-off-by: Anoob Joseph <anoob.joseph@caviumnetworks.com>
>> Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
>> ---
> 
> Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
>
Applied to dpdk-next-net/master, thanks.
  

Patch

diff --git a/drivers/net/octeontx/octeontx_ethdev.c b/drivers/net/octeontx/octeontx_ethdev.c
index 0681486..9c421d5 100644
--- a/drivers/net/octeontx/octeontx_ethdev.c
+++ b/drivers/net/octeontx/octeontx_ethdev.c
@@ -1237,15 +1237,8 @@  octeontx_probe(struct rte_vdev_device *dev)
 		res = -EINVAL;
 		goto parse_error;
 	}
-	if (pnum > qnum) {
-		/*
-		 * We don't poll on event ports
-		 * that do not have any queues assigned.
-		 */
-		pnum = qnum;
-		PMD_INIT_LOG(INFO,
-			"reducing number of active event ports to %d", pnum);
-	}
+
+	/* Enable all queues available */
 	for (i = 0; i < qnum; i++) {
 		res = rte_event_queue_setup(evdev, i, NULL);
 		if (res < 0) {
@@ -1255,6 +1248,7 @@  octeontx_probe(struct rte_vdev_device *dev)
 		}
 	}
 
+	/* Enable all ports available */
 	for (i = 0; i < pnum; i++) {
 		res = rte_event_port_setup(evdev, i, NULL);
 		if (res < 0) {
@@ -1263,6 +1257,14 @@  octeontx_probe(struct rte_vdev_device *dev)
 						i, res);
 			goto parse_error;
 		}
+	}
+
+	/*
+	 * Do 1:1 links for ports & queues. All queues would be mapped to
+	 * one port. If there are more ports than queues, then some ports
+	 * won't be linked to any queue.
+	 */
+	for (i = 0; i < qnum; i++) {
 		/* Link one queue to one event port */
 		qlist = i;
 		res = rte_event_port_link(evdev, i, &qlist, NULL, 1);