netdev - [RFC PATCH net-next 1/1] ibmveth: Implement BQL

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: <20221024213828.320219-2-nnac123@linux.ibm.com>
Date:   Mon, 24 Oct 2022 16:38:28 -0500
From:   Nick Child <nnac123@...ux.ibm.com>
To:     netdev@...r.kernel.org
Cc:     nick.child@....com, dave.taht@...il.com,
        Nick Child <nnac123@...ux.ibm.com>
Subject: [RFC PATCH net-next 1/1] ibmveth: Implement BQL

Byte Queue Limits allows for dynamic size management of transmit
queues. By informing the higher-level networking layers about in-flight
transmissions, the number queued bytes will be optimized for better
performance and minimal buffer-bloat.

Since the ibmveth driver does not have an irq to receive packet
completion information from firmware, assume the packet has completed
transmission when the hypervisor call H_SEND_LOGICAL_LAN returns.

Signed-off-by: Nick Child <nnac123@...ux.ibm.com>
---
 drivers/net/ethernet/ibm/ibmveth.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/ibm/ibmveth.c b/drivers/net/ethernet/ibm/ibmveth.c
index 3b14dc93f59d..1290848e2e2b 100644
--- a/drivers/net/ethernet/ibm/ibmveth.c
+++ b/drivers/net/ethernet/ibm/ibmveth.c
@@ -578,6 +578,7 @@ static int ibmveth_open(struct net_device *netdev)
 	}
 
 	for (i = 0; i < netdev->real_num_tx_queues; i++) {
+		netdev_tx_reset_queue(netdev_get_tx_queue(netdev, i));
 		if (ibmveth_allocate_tx_ltb(adapter, i))
 			goto out_free_tx_ltb;
 	}
@@ -1027,8 +1028,10 @@ static int ibmveth_set_channels(struct net_device *netdev,
 			continue;
 
 		rc = ibmveth_allocate_tx_ltb(adapter, i);
-		if (!rc)
+		if (!rc) {
+			netdev_tx_reset_queue(netdev_get_tx_queue(netdev, i));
 			continue;
+		}
 
 		/* if something goes wrong, free everything we just allocated */
 		netdev_err(netdev, "Failed to allocate more tx queues, returning to %d queues\n",
@@ -1125,6 +1128,7 @@ static netdev_tx_t ibmveth_start_xmit(struct sk_buff *skb,
 	union ibmveth_buf_desc desc;
 	int i, queue_num = skb_get_queue_mapping(skb);
 	unsigned long mss = 0;
+	struct netdev_queue *txq = netdev_get_tx_queue(netdev, queue_num);
 
 	if (ibmveth_is_packet_unsupported(skb, netdev))
 		goto out;
@@ -1202,6 +1206,7 @@ static netdev_tx_t ibmveth_start_xmit(struct sk_buff *skb,
 	/* finish writing to long_term_buff before VIOS accessing it */
 	dma_wmb();
 
+	netdev_tx_sent_queue(txq, skb->len);
 	if (ibmveth_send(adapter, desc.desc, mss)) {
 		adapter->tx_send_failed++;
 		netdev->stats.tx_dropped++;
@@ -1209,6 +1214,7 @@ static netdev_tx_t ibmveth_start_xmit(struct sk_buff *skb,
 		netdev->stats.tx_packets++;
 		netdev->stats.tx_bytes += skb->len;
 	}
+	netdev_tx_completed_queue(txq, 1, skb->len);
 
 out:
 	dev_consume_skb_any(skb);
-- 
2.31.1