Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Ronak Doshi <doshir@vmware.com>
To:     <netdev@vger.kernel.org>
CC:     <ntanaka@vmware.com>, Ronak Doshi <doshir@vmware.com>,
        Shrikrishna Khare <skhare@vmware.com>,
        "VMware, Inc." <pv-drivers@vmware.com>,
        open list <linux-kernel@vger.kernel.org>
Subject: [PATCH net 1/2] vmxnet3: avoid xmit reset due to a race in vmxnet3
Date:   Tue, 13 Mar 2018 12:40:08 -0700
Message-ID: <20180313194008.12852-1-doshir@vmware.com>
MIME-Version: 1.0
Content-Type: text/plain
Received-SPF: None (EX13-EDG-OU-001.vmware.com: doshir@vmware.com does not
 designate permitted sender hosts)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

The field txNumDeferred is used by the driver to keep track of the number
of packets it has pushed to the emulation. The driver increments it on
pushing the packet to the emulation and the emulation resets it to 0 at
the end of the transmit.

There is a possibility of a race either when (a) ESX is under heavy load or
(b) workload inside VM is of low packet rate.

This race results in xmit hangs when network coalescing is disabled. This
change creates a local copy of txNumDeferred and uses it to perform ring
arithmetic.

Reported-by: Noriho Tanaka <ntanaka@vmware.com>
Signed-off-by: Ronak Doshi <doshir@vmware.com>
Acked-by: Shrikrishna Khare <skhare@vmware.com>
---
 drivers/net/vmxnet3/vmxnet3_drv.c | 13 ++++++++-----
 drivers/net/vmxnet3/vmxnet3_int.h |  4 ++--
 2 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/drivers/net/vmxnet3/vmxnet3_drv.c b/drivers/net/vmxnet3/vmxnet3_drv.c
index 8b39c160743d..052eef2f729f 100644
--- a/drivers/net/vmxnet3/vmxnet3_drv.c
+++ b/drivers/net/vmxnet3/vmxnet3_drv.c
@@ -977,6 +977,8 @@ vmxnet3_tq_xmit(struct sk_buff *skb, struct vmxnet3_tx_queue *tq,
 {
 	int ret;
 	u32 count;
+	int numPkts;
+	int localTxNumDeferred;
 	unsigned long flags;
 	struct vmxnet3_tx_ctx ctx;
 	union Vmxnet3_GenericDesc *gdesc;
@@ -1075,12 +1077,12 @@ vmxnet3_tq_xmit(struct sk_buff *skb, struct vmxnet3_tx_queue *tq,
 #else
 	gdesc = ctx.sop_txd;
 #endif
+	localTxNumDeferred = le32_to_cpu(tq->shared->txNumDeferred);
 	if (ctx.mss) {
 		gdesc->txd.hlen = ctx.eth_ip_hdr_size + ctx.l4_hdr_size;
 		gdesc->txd.om = VMXNET3_OM_TSO;
 		gdesc->txd.msscof = ctx.mss;
-		le32_add_cpu(&tq->shared->txNumDeferred, (skb->len -
-			     gdesc->txd.hlen + ctx.mss - 1) / ctx.mss);
+		numPkts = (skb->len - gdesc->txd.hlen + ctx.mss - 1) / ctx.mss;
 	} else {
 		if (skb->ip_summed == CHECKSUM_PARTIAL) {
 			gdesc->txd.hlen = ctx.eth_ip_hdr_size;
@@ -1091,8 +1093,10 @@ vmxnet3_tq_xmit(struct sk_buff *skb, struct vmxnet3_tx_queue *tq,
 			gdesc->txd.om = 0;
 			gdesc->txd.msscof = 0;
 		}
-		le32_add_cpu(&tq->shared->txNumDeferred, 1);
+		numPkts = 1;
 	}
+	le32_add_cpu(&tq->shared->txNumDeferred, numPkts);
+	localTxNumDeferred += numPkts;
 
 	if (skb_vlan_tag_present(skb)) {
 		gdesc->txd.ti = 1;
@@ -1118,8 +1122,7 @@ vmxnet3_tq_xmit(struct sk_buff *skb, struct vmxnet3_tx_queue *tq,
 
 	spin_unlock_irqrestore(&tq->tx_lock, flags);
 
-	if (le32_to_cpu(tq->shared->txNumDeferred) >=
-					le32_to_cpu(tq->shared->txThreshold)) {
+	if (localTxNumDeferred >= le32_to_cpu(tq->shared->txThreshold)) {
 		tq->shared->txNumDeferred = 0;
 		VMXNET3_WRITE_BAR0_REG(adapter,
 				       VMXNET3_REG_TXPROD + tq->qid * 8,
diff --git a/drivers/net/vmxnet3/vmxnet3_int.h b/drivers/net/vmxnet3/vmxnet3_int.h
index 5ba222920e80..b94fdfd0b6f1 100644
--- a/drivers/net/vmxnet3/vmxnet3_int.h
+++ b/drivers/net/vmxnet3/vmxnet3_int.h
@@ -69,10 +69,10 @@
 /*
  * Version numbers
  */
-#define VMXNET3_DRIVER_VERSION_STRING   "1.4.11.0-k"
+#define VMXNET3_DRIVER_VERSION_STRING   "1.4.12.0-k"
 
 /* a 32-bit int, each byte encode a verion number in VMXNET3_DRIVER_VERSION */
-#define VMXNET3_DRIVER_VERSION_NUM      0x01040b00
+#define VMXNET3_DRIVER_VERSION_NUM      0x01040c00
 
 #if defined(CONFIG_PCI_MSI)
 	/* RSS only makes sense if MSI-X is supported. */
-- 
2.11.0