netdev - [PATCH net-next 05/11] s390/qeth: allow configuration of TX queues for IQD devices

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20200318125455.5838-6-jwi@linux.ibm.com>
Date:   Wed, 18 Mar 2020 13:54:49 +0100
From:   Julian Wiedmann <jwi@...ux.ibm.com>
To:     David Miller <davem@...emloft.net>
Cc:     netdev <netdev@...r.kernel.org>,
        linux-s390 <linux-s390@...r.kernel.org>,
        Heiko Carstens <heiko.carstens@...ibm.com>,
        Ursula Braun <ubraun@...ux.ibm.com>,
        Julian Wiedmann <jwi@...ux.ibm.com>
Subject: [PATCH net-next 05/11] s390/qeth: allow configuration of TX queues for IQD devices

Similar to the support for z/VM NICs, but we need to take extra care
about the dedicated mcast queue:

1. netdev_pick_tx() is unaware of this limitation and might select the
   mcast txq. Catch this.
2. require at least _two_ TX queues - one for ucast, one for mcast.
3. when reducing the number of TX queues, there's a potential race
   where netdev_cap_txqueue() over-rules the selected txq index and
   falls back to index 0. This would place ucast traffic on the mcast
   queue, and result in TX errors.
   So for IQD, reject a reduction while the interface is running.

Signed-off-by: Julian Wiedmann <jwi@...ux.ibm.com>
---
 drivers/s390/net/qeth_core_main.c |  6 +++++-
 drivers/s390/net/qeth_ethtool.c   | 19 ++++++++++++++++---
 2 files changed, 21 insertions(+), 4 deletions(-)

diff --git a/drivers/s390/net/qeth_core_main.c b/drivers/s390/net/qeth_core_main.c
index f13495d9209b..aa493edc0082 100644
--- a/drivers/s390/net/qeth_core_main.c
+++ b/drivers/s390/net/qeth_core_main.c
@@ -6644,9 +6644,13 @@ EXPORT_SYMBOL_GPL(qeth_get_stats64);
 u16 qeth_iqd_select_queue(struct net_device *dev, struct sk_buff *skb,
 			  u8 cast_type, struct net_device *sb_dev)
 {
+	u16 txq;
+
 	if (cast_type != RTN_UNICAST)
 		return QETH_IQD_MCAST_TXQ;
-	return QETH_IQD_MIN_UCAST_TXQ;
+
+	txq = netdev_pick_tx(dev, skb, sb_dev);
+	return (txq == QETH_IQD_MCAST_TXQ) ? QETH_IQD_MIN_UCAST_TXQ : txq;
 }
 EXPORT_SYMBOL_GPL(qeth_iqd_select_queue);
 
diff --git a/drivers/s390/net/qeth_ethtool.c b/drivers/s390/net/qeth_ethtool.c
index 19b9c8302d36..715ee0015847 100644
--- a/drivers/s390/net/qeth_ethtool.c
+++ b/drivers/s390/net/qeth_ethtool.c
@@ -180,14 +180,27 @@ static int qeth_set_channels(struct net_device *dev,
 {
 	struct qeth_card *card = dev->ml_priv;
 
-	if (IS_IQD(card) || !IS_VM_NIC(card))
-		return -EOPNOTSUPP;
-
 	if (channels->rx_count == 0 || channels->tx_count == 0)
 		return -EINVAL;
 	if (channels->tx_count > card->qdio.no_out_queues)
 		return -EINVAL;
 
+	if (IS_IQD(card)) {
+		if (channels->tx_count < QETH_IQD_MIN_TXQ)
+			return -EINVAL;
+
+		/* Reject downgrade while running. It could push displaced
+		 * ucast flows onto txq0, which is reserved for mcast.
+		 */
+		if (netif_running(dev) &&
+		    channels->tx_count < dev->real_num_tx_queues)
+			return -EPERM;
+	} else {
+		/* OSA still uses the legacy prio-queue mechanism: */
+		if (!IS_VM_NIC(card))
+			return -EOPNOTSUPP;
+	}
+
 	return netif_set_real_num_tx_queues(dev, channels->tx_count);
 }
 
-- 
2.17.1