* [PATCH v2 1/2] net/virtio: fix vDPA device init advertising control queue
[not found] <20240313125932.1878279-1-maxime.coquelin@redhat.com>
@ 2024-03-13 12:59 ` Maxime Coquelin
2024-03-13 14:12 ` David Marchand
0 siblings, 1 reply; 2+ messages in thread
From: Maxime Coquelin @ 2024-03-13 12:59 UTC (permalink / raw)
To: dev, david.marchand, chenbox, schalla; +Cc: Maxime Coquelin, stable
If the vDPA device advertises control queue support, but
the user neither passes "cq=1" as devarg nor requests
multiple queues, the initialization fails because the
driver tries to setup the control queue without negotiating
related feature.
This patch enables the control queue at driver level as
soon as the device claims to support it, and not only when
multiple queue pairs are requested. Also, enable the
control queue event if multiqueue feature has not been
negotiated and device start time, and disable it at device
stop time.
Fixes: b277308e8868 ("net/virtio-user: advertise control VQ support with vDPA")
Cc: stable@dpdk.org
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
---
.../net/virtio/virtio_user/virtio_user_dev.c | 22 ++++++++++++++-----
1 file changed, 16 insertions(+), 6 deletions(-)
diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c b/drivers/net/virtio/virtio_user/virtio_user_dev.c
index d395fc1676..54187fedf5 100644
--- a/drivers/net/virtio/virtio_user/virtio_user_dev.c
+++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c
@@ -216,6 +216,12 @@ virtio_user_start_device(struct virtio_user_dev *dev)
if (ret < 0)
goto error;
+ if (dev->scvq) {
+ ret = dev->ops->cvq_enable(dev, 1);
+ if (ret < 0)
+ goto error;
+ }
+
dev->started = true;
pthread_mutex_unlock(&dev->mutex);
@@ -248,6 +254,12 @@ int virtio_user_stop_device(struct virtio_user_dev *dev)
goto err;
}
+ if (dev->scvq) {
+ ret = dev->ops->cvq_enable(dev, 0);
+ if (ret < 0)
+ goto err;
+ }
+
/* Stop the backend. */
for (i = 0; i < dev->max_queue_pairs * 2; ++i) {
state.index = i;
@@ -752,7 +764,7 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char *path, uint16_t queues,
if (virtio_user_dev_init_max_queue_pairs(dev, queues))
dev->unsupported_features |= (1ull << VIRTIO_NET_F_MQ);
- if (dev->max_queue_pairs > 1)
+ if (dev->max_queue_pairs > 1 || dev->hw_cvq)
cq = 1;
if (!mrg_rxbuf)
@@ -770,8 +782,9 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char *path, uint16_t queues,
dev->unsupported_features |= (1ull << VIRTIO_NET_F_MAC);
if (cq) {
- /* device does not really need to know anything about CQ,
- * so if necessary, we just claim to support CQ
+ /* Except for vDPA, the device does not really need to know
+ * anything about CQ, so if necessary, we just claim to support
+ * control queue.
*/
dev->frontend_features |= (1ull << VIRTIO_NET_F_CTRL_VQ);
} else {
@@ -871,9 +884,6 @@ virtio_user_handle_mq(struct virtio_user_dev *dev, uint16_t q_pairs)
for (i = q_pairs; i < dev->max_queue_pairs; ++i)
ret |= dev->ops->enable_qp(dev, i, 0);
- if (dev->scvq)
- ret |= dev->ops->cvq_enable(dev, 1);
-
dev->queue_pairs = q_pairs;
return ret;
--
2.44.0
^ permalink raw reply [flat|nested] 2+ messages in thread