linux-kernel - [PATCH net-next 10/12] net: hns3: add 8 BD limit for tx flow

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: <20190126164921.3872-11-tanhuazhong@huawei.com>
Date:   Sun, 27 Jan 2019 00:49:19 +0800
From:   Huazhong Tan <tanhuazhong@...wei.com>
To:     <davem@...emloft.net>
CC:     <netdev@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        <huangdaode@...ilicon.com>, <yisen.zhuang@...wei.com>,
        <salil.mehta@...wei.com>, <linuxarm@...wei.com>,
        Peng Li <lipeng321@...wei.com>,
        Huazhong Tan <tanhuazhong@...wei.com>
Subject: [PATCH net-next 10/12] net: hns3: add 8 BD limit for tx flow

From: Peng Li <lipeng321@...wei.com>

A single transmit packet can span up to 8 descriptors according
to the HW limit. If a skb has more than 8 frags, driver uses
skb_copy to get a new skb which has less frags.

Signed-off-by: Peng Li <lipeng321@...wei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@...wei.com>
---
 .../net/ethernet/hisilicon/hns3/hns3_enet.c   | 28 ++++++++++++++++++-
 1 file changed, 27 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index 93e3e38ba852..216f3dd25cd6 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -1133,6 +1133,7 @@ static int hns3_nic_maybe_stop_tso(struct sk_buff **out_skb, int *bnum,
 				   struct hns3_enet_ring *ring)
 {
 	struct sk_buff *skb = *out_skb;
+	struct sk_buff *new_skb = NULL;
 	struct skb_frag_struct *frag;
 	int bdnum_for_frag;
 	int frag_num;
@@ -1155,7 +1156,19 @@ static int hns3_nic_maybe_stop_tso(struct sk_buff **out_skb, int *bnum,
 		buf_num += bdnum_for_frag;
 	}
 
-	if (buf_num > ring_space(ring))
+	if (unlikely(buf_num > HNS3_MAX_BD_PER_FRAG)) {
+		buf_num = (skb->len + HNS3_MAX_BD_SIZE - 1) / HNS3_MAX_BD_SIZE;
+		if (ring_space(ring) < buf_num)
+			return -EBUSY;
+		/* manual split the send packet */
+		new_skb = skb_copy(skb, GFP_ATOMIC);
+		if (!new_skb)
+			return -ENOMEM;
+		dev_kfree_skb_any(skb);
+		*out_skb = new_skb;
+	}
+
+	if (unlikely(ring_space(ring) < buf_num))
 		return -EBUSY;
 
 	*bnum = buf_num;
@@ -1166,11 +1179,24 @@ static int hns3_nic_maybe_stop_tx(struct sk_buff **out_skb, int *bnum,
 				  struct hns3_enet_ring *ring)
 {
 	struct sk_buff *skb = *out_skb;
+	struct sk_buff *new_skb = NULL;
 	int buf_num;
 
 	/* No. of segments (plus a header) */
 	buf_num = skb_shinfo(skb)->nr_frags + 1;
 
+	if (unlikely(buf_num > HNS3_MAX_BD_PER_FRAG)) {
+		buf_num = (skb->len + HNS3_MAX_BD_SIZE - 1) / HNS3_MAX_BD_SIZE;
+		if (ring_space(ring) < buf_num)
+			return -EBUSY;
+		/* manual split the send packet */
+		new_skb = skb_copy(skb, GFP_ATOMIC);
+		if (!new_skb)
+			return -ENOMEM;
+		dev_kfree_skb_any(skb);
+		*out_skb = new_skb;
+	}
+
 	if (unlikely(ring_space(ring) < buf_num))
 		return -EBUSY;
 
-- 
2.20.1