patches for DPDK stable branches
 help / color / mirror / Atom feed
From: Yongseok Koh <yskoh@mellanox.com>
To: Yongseok Koh <yskoh@mellanox.com>
Cc: dpdk stable <stable@dpdk.org>
Subject: [dpdk-stable] patch 'net/mlx5: fix queue rollback when starting device' has been queued to LTS release 17.11.4
Date: Thu, 26 Jul 2018 19:33:36 -0700	[thread overview]
Message-ID: <20180727023339.38855-1-yskoh@mellanox.com> (raw)

Hi,

FYI, your patch has been queued to LTS release 17.11.4

Note it hasn't been pushed to http://dpdk.org/browse/dpdk-stable yet.
It will be pushed if I get no objections before 07/28/18. So please
shout if anyone has objections.

Thanks.

Yongseok

---
>From 997e25edbb09bb08e6289ad2b77b6e833c18287c Mon Sep 17 00:00:00 2001
From: Yongseok Koh <yskoh@mellanox.com>
Date: Mon, 23 Jul 2018 13:57:04 -0700
Subject: [PATCH] net/mlx5: fix queue rollback when starting device

[ upstream commit 24f653a7e8bff5f500894a07771dd6718966e9be ]

mlx5_rxq_start() and mlx5_rxq_stop() must be strictly paired because
internal reference counter is increased or decreased inside. Also,
mlx5_rxq_get() must be paired with mlx5_rxq_release().

Fixes: 7d6bf6b866b8 ("net/mlx5: add Multi-Packet Rx support")
Fixes: a1366b1a2be3 ("net/mlx5: add reference counter on DPDK Rx queues")
Fixes: 6e78005a9b30 ("net/mlx5: add reference counter on DPDK Tx queues")

Signed-off-by: Yongseok Koh <yskoh@mellanox.com>
---
 drivers/net/mlx5/mlx5_trigger.c | 27 ++++++++++++++++-----------
 1 file changed, 16 insertions(+), 11 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_trigger.c b/drivers/net/mlx5/mlx5_trigger.c
index 49f1c1331..9a1d6f954 100644
--- a/drivers/net/mlx5/mlx5_trigger.c
+++ b/drivers/net/mlx5/mlx5_trigger.c
@@ -73,7 +73,6 @@ mlx5_txq_start(struct rte_eth_dev *dev)
 	unsigned int i;
 	int ret;
 
-	/* Add memory regions to Tx queues. */
 	for (i = 0; i != priv->txqs_n; ++i) {
 		unsigned int idx = 0;
 		struct mlx5_mr *mr;
@@ -94,12 +93,17 @@ mlx5_txq_start(struct rte_eth_dev *dev)
 		}
 	}
 	ret = mlx5_tx_uar_remap(dev, priv->ctx->cmd_fd);
-	if (ret)
+	if (ret) {
+		/* Adjust index for rollback. */
+		i = priv->txqs_n - 1;
 		goto error;
+	}
 	return 0;
 error:
 	ret = rte_errno; /* Save rte_errno before cleanup. */
-	mlx5_txq_stop(dev);
+	do {
+		mlx5_txq_release(dev, i);
+	} while (i-- != 0);
 	rte_errno = ret; /* Restore rte_errno. */
 	return -rte_errno;
 }
@@ -151,7 +155,9 @@ mlx5_rxq_start(struct rte_eth_dev *dev)
 	return 0;
 error:
 	ret = rte_errno; /* Save rte_errno before cleanup. */
-	mlx5_rxq_stop(dev);
+	do {
+		mlx5_rxq_release(dev, i);
+	} while (i-- != 0);
 	rte_errno = ret; /* Restore rte_errno. */
 	return -rte_errno;
 }
@@ -174,28 +180,28 @@ mlx5_dev_start(struct rte_eth_dev *dev)
 	struct mlx5_mr *mr = NULL;
 	int ret;
 
-	dev->data->dev_started = 1;
+	DRV_LOG(DEBUG, "port %u starting device", dev->data->port_id);
 	ret = mlx5_flow_create_drop_queue(dev);
 	if (ret) {
 		DRV_LOG(ERR, "port %u drop queue allocation failed: %s",
 			dev->data->port_id, strerror(rte_errno));
 		goto error;
 	}
-	DRV_LOG(DEBUG, "port %u allocating and configuring hash Rx queues",
-		dev->data->port_id);
 	rte_mempool_walk(mlx5_mp2mr_iter, priv);
 	ret = mlx5_txq_start(dev);
 	if (ret) {
 		DRV_LOG(ERR, "port %u Tx queue allocation failed: %s",
 			dev->data->port_id, strerror(rte_errno));
-		goto error;
+		return -rte_errno;
 	}
 	ret = mlx5_rxq_start(dev);
 	if (ret) {
 		DRV_LOG(ERR, "port %u Rx queue allocation failed: %s",
 			dev->data->port_id, strerror(rte_errno));
-		goto error;
+		mlx5_txq_stop(dev);
+		return -rte_errno;
 	}
+	dev->data->dev_started = 1;
 	ret = mlx5_rx_intr_vec_enable(dev);
 	if (ret) {
 		DRV_LOG(ERR, "port %u Rx interrupt vector creation failed",
@@ -254,8 +260,7 @@ mlx5_dev_stop(struct rte_eth_dev *dev)
 	dev->tx_pkt_burst = removed_tx_burst;
 	rte_wmb();
 	usleep(1000 * priv->rxqs_n);
-	DRV_LOG(DEBUG, "port %u cleaning up and destroying hash Rx queues",
-		dev->data->port_id);
+	DRV_LOG(DEBUG, "port %u stopping device", dev->data->port_id);
 	mlx5_flow_stop(dev, &priv->flows);
 	mlx5_traffic_disable(dev);
 	mlx5_rx_intr_vec_disable(dev);
-- 
2.11.0

             reply	other threads:[~2018-07-27  2:33 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-07-27  2:33 Yongseok Koh [this message]
2018-07-27  2:33 ` [dpdk-stable] patch 'maintainers: update for Mellanox PMDs' " Yongseok Koh
2018-07-27  2:33 ` [dpdk-stable] patch 'net/mlx5: fix error number handling' " Yongseok Koh
2018-07-27  2:33 ` [dpdk-stable] patch 'net/mlx5: fix compilation for rdma-core v19' " Yongseok Koh

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20180727023339.38855-1-yskoh@mellanox.com \
    --to=yskoh@mellanox.com \
    --cc=stable@dpdk.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).