From: Jakub Kicinski <kuba@kernel.org>
Date: Thu, 27 Jan 2022 10:42:59 -0800
Subject: ethtool: add header/data split indication
Patch-mainline: v5.18-rc1
Git-commit: 9690ae60429020f38e4aa2540c306f27eb021bc0
References: jsc#PED-1495

For applications running on a mix of platforms it's useful
to have a clear indication whether host's NIC supports the
geometry requirements of TCP zero-copy. TCP zero-copy Rx
requires data to be neatly placed into memory pages.
Most NICs can't do that.

This patch is adding GET support only, since the NICs
I work with either always have the feature enabled or
enable it whenever MTU is set to jumbo. In other words
I don't need SET. But adding set should be trivial.
(The only note on SET is that we will likely want
the setting to be "sticky" and use 0 / `unknown`
to reset it back to driver default.)

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
 Documentation/networking/ethtool-netlink.rst |    8 ++++++++
 include/linux/ethtool.h                      |    2 ++
 include/uapi/linux/ethtool_netlink.h         |    7 +++++++
 net/ethtool/rings.c                          |   15 ++++++++++-----
 4 files changed, 27 insertions(+), 5 deletions(-)

--- a/Documentation/networking/ethtool-netlink.rst
+++ b/Documentation/networking/ethtool-netlink.rst
@@ -834,8 +834,16 @@ Kernel response contents:
   ``ETHTOOL_A_RINGS_RX_JUMBO``          u32     size of RX jumbo ring
   ``ETHTOOL_A_RINGS_TX``                u32     size of TX ring
   ``ETHTOOL_A_RINGS_RX_BUF_LEN``        u32     size of buffers on the ring
+  ``ETHTOOL_A_RINGS_TCP_DATA_SPLIT``    u8      TCP header / data split
   ====================================  ======  ===========================
 
+``ETHTOOL_A_RINGS_TCP_DATA_SPLIT`` indicates whether the device is usable with
+page-flipping TCP zero-copy receive (``getsockopt(TCP_ZEROCOPY_RECEIVE)``).
+If enabled the device is configured to place frame headers and data into
+separate buffers. The device configuration must make it possible to receive
+full memory pages of data, for example because MTU is high enough or through
+HW-GRO.
+
 
 RINGS_SET
 =========
--- a/include/linux/ethtool.h
+++ b/include/linux/ethtool.h
@@ -70,9 +70,11 @@ enum {
 /**
  * struct kernel_ethtool_ringparam - RX/TX ring configuration
  * @rx_buf_len: Current length of buffers on the rx ring.
+ * @tcp_data_split: Scatter packet headers and data to separate buffers
  */
 struct kernel_ethtool_ringparam {
 	u32	rx_buf_len;
+	u8	tcp_data_split;
 };
 
 /**
--- a/include/uapi/linux/ethtool_netlink.h
+++ b/include/uapi/linux/ethtool_netlink.h
@@ -315,6 +315,12 @@ enum {
 /* RINGS */
 
 enum {
+	ETHTOOL_TCP_DATA_SPLIT_UNKNOWN = 0,
+	ETHTOOL_TCP_DATA_SPLIT_DISABLED,
+	ETHTOOL_TCP_DATA_SPLIT_ENABLED,
+};
+
+enum {
 	ETHTOOL_A_RINGS_UNSPEC,
 	ETHTOOL_A_RINGS_HEADER,				/* nest - _A_HEADER_* */
 	ETHTOOL_A_RINGS_RX_MAX,				/* u32 */
@@ -326,6 +332,7 @@ enum {
 	ETHTOOL_A_RINGS_RX_JUMBO,			/* u32 */
 	ETHTOOL_A_RINGS_TX,				/* u32 */
 	ETHTOOL_A_RINGS_RX_BUF_LEN,                     /* u32 */
+	ETHTOOL_A_RINGS_TCP_DATA_SPLIT,			/* u8 */
 
 	/* add new constants above here */
 	__ETHTOOL_A_RINGS_CNT,
--- a/net/ethtool/rings.c
+++ b/net/ethtool/rings.c
@@ -53,7 +53,8 @@ static int rings_reply_size(const struct
 	       nla_total_size(sizeof(u32)) +	/* _RINGS_RX_MINI */
 	       nla_total_size(sizeof(u32)) +	/* _RINGS_RX_JUMBO */
 	       nla_total_size(sizeof(u32)) +	/* _RINGS_TX */
-	       nla_total_size(sizeof(u32));     /* _RINGS_RX_BUF_LEN */
+	       nla_total_size(sizeof(u32)) +	/* _RINGS_RX_BUF_LEN */
+	       nla_total_size(sizeof(u8));	/* _RINGS_TCP_DATA_SPLIT */
 }
 
 static int rings_fill_reply(struct sk_buff *skb,
@@ -61,9 +62,11 @@ static int rings_fill_reply(struct sk_bu
 			    const struct ethnl_reply_data *reply_base)
 {
 	const struct rings_reply_data *data = RINGS_REPDATA(reply_base);
-	const struct kernel_ethtool_ringparam *kernel_ringparam = &data->kernel_ringparam;
+	const struct kernel_ethtool_ringparam *kr = &data->kernel_ringparam;
 	const struct ethtool_ringparam *ringparam = &data->ringparam;
 
+	WARN_ON(kr->tcp_data_split > ETHTOOL_TCP_DATA_SPLIT_ENABLED);
+
 	if ((ringparam->rx_max_pending &&
 	     (nla_put_u32(skb, ETHTOOL_A_RINGS_RX_MAX,
 			  ringparam->rx_max_pending) ||
@@ -84,9 +87,11 @@ static int rings_fill_reply(struct sk_bu
 			  ringparam->tx_max_pending) ||
 	      nla_put_u32(skb, ETHTOOL_A_RINGS_TX,
 			  ringparam->tx_pending)))  ||
-	    (kernel_ringparam->rx_buf_len &&
-	     (nla_put_u32(skb, ETHTOOL_A_RINGS_RX_BUF_LEN,
-			  kernel_ringparam->rx_buf_len))))
+	    (kr->rx_buf_len &&
+	     (nla_put_u32(skb, ETHTOOL_A_RINGS_RX_BUF_LEN, kr->rx_buf_len))) ||
+	    (kr->tcp_data_split &&
+	     (nla_put_u8(skb, ETHTOOL_A_RINGS_TCP_DATA_SPLIT,
+			 kr->tcp_data_split))))
 		return -EMSGSIZE;
 
 	return 0;