[PATCH net-next 0/3] misc cleanups for xdp

[PATCH net-next 0/3] misc cleanups for xdp

[Brenden Blanco]

This addresses several of the non-blocking comments left over from the
xdp patch set. See individual patches for details.

Brenden Blanco (3):
  net/mlx4_en: use READ_ONCE when freeing xdp_prog
  rtnl: protect do_setlink from IFLA_XDP_ATTACHED
  bpf: make xdp sample variable names more meaningful

 drivers/net/ethernet/mellanox/mlx4/en_rx.c |  6 ++++--
 net/core/rtnetlink.c                       |  4 ++++
 samples/bpf/xdp1_kern.c                    | 12 ++++++------
 samples/bpf/xdp2_kern.c                    | 14 +++++++-------
 4 files changed, 21 insertions(+), 15 deletions(-)

-- 
2.8.2

[PATCH net-next 1/3] net/mlx4_en: use READ_ONCE when freeing xdp_prog

[Brenden Blanco]

For consistency, and in order to hint at the synchronous nature of the
xdp_prog field, use READ_ONCE in the destroy path of the ring. All
occurrences should now use either READ_ONCE or xchg.

Signed-off-by: Brenden Blanco <bblanco@plumgrid.com>
---
 drivers/net/ethernet/mellanox/mlx4/en_rx.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx4/en_rx.c b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
index 11d88c8..a02dec6 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
@@ -535,9 +535,11 @@ void mlx4_en_destroy_rx_ring(struct mlx4_en_priv *priv,
 {
 	struct mlx4_en_dev *mdev = priv->mdev;
 	struct mlx4_en_rx_ring *ring = *pring;
+	struct bpf_prog *old_prog;
 
-	if (ring->xdp_prog)
-		bpf_prog_put(ring->xdp_prog);
+	old_prog = READ_ONCE(ring->xdp_prog);
+	if (old_prog)
+		bpf_prog_put(old_prog);
 	mlx4_free_hwq_res(mdev->dev, &ring->wqres, size * stride + TXBB_SIZE);
 	vfree(ring->rx_info);
 	ring->rx_info = NULL;
-- 
2.8.2

[PATCH net-next 2/3] rtnl: protect do_setlink from IFLA_XDP_ATTACHED

[Brenden Blanco]

The IFLA_XDP_ATTACHED nested attribute is meant for read-only, and while
do_setlink properly ignores it, it should be more paranoid and reject
commands that try to set it.

Signed-off-by: Brenden Blanco <bblanco@plumgrid.com>
---
 net/core/rtnetlink.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index eba2b82..189cc78 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -2109,6 +2109,10 @@ static int do_setlink(const struct sk_buff *skb,
 		if (err < 0)
 			goto errout;
 
+		if (xdp[IFLA_XDP_ATTACHED]) {
+			err = -EINVAL;
+			goto errout;
+		}
 		if (xdp[IFLA_XDP_FD]) {
 			err = dev_change_xdp_fd(dev,
 						nla_get_s32(xdp[IFLA_XDP_FD]));
-- 
2.8.2

[PATCH net-next 3/3] bpf: make xdp sample variable names more meaningful

[Brenden Blanco]

The naming choice of index is not terribly descriptive, and dropcnt is
in fact incorrect for xdp2. Pick better names for these: ipproto and
rxcnt.

Signed-off-by: Brenden Blanco <bblanco@plumgrid.com>
---
 samples/bpf/xdp1_kern.c | 12 ++++++------
 samples/bpf/xdp2_kern.c | 14 +++++++-------
 2 files changed, 13 insertions(+), 13 deletions(-)

diff --git a/samples/bpf/xdp1_kern.c b/samples/bpf/xdp1_kern.c
index e7dd8ac..2197421 100644
--- a/samples/bpf/xdp1_kern.c
+++ b/samples/bpf/xdp1_kern.c
@@ -14,7 +14,7 @@
 #include <linux/ipv6.h>
 #include "bpf_helpers.h"
 
-struct bpf_map_def SEC("maps") dropcnt = {
+struct bpf_map_def SEC("maps") rxcnt = {
 	.type = BPF_MAP_TYPE_PERCPU_ARRAY,
 	.key_size = sizeof(u32),
 	.value_size = sizeof(long),
@@ -49,7 +49,7 @@ int xdp_prog1(struct xdp_md *ctx)
 	long *value;
 	u16 h_proto;
 	u64 nh_off;
-	u32 index;
+	u32 ipproto;
 
 	nh_off = sizeof(*eth);
 	if (data + nh_off > data_end)
@@ -77,13 +77,13 @@ int xdp_prog1(struct xdp_md *ctx)
 	}
 
 	if (h_proto == htons(ETH_P_IP))
-		index = parse_ipv4(data, nh_off, data_end);
+		ipproto = parse_ipv4(data, nh_off, data_end);
 	else if (h_proto == htons(ETH_P_IPV6))
-		index = parse_ipv6(data, nh_off, data_end);
+		ipproto = parse_ipv6(data, nh_off, data_end);
 	else
-		index = 0;
+		ipproto = 0;
 
-	value = bpf_map_lookup_elem(&dropcnt, &index);
+	value = bpf_map_lookup_elem(&rxcnt, &ipproto);
 	if (value)
 		*value += 1;
 
diff --git a/samples/bpf/xdp2_kern.c b/samples/bpf/xdp2_kern.c
index 38fe7e1..e012888 100644
--- a/samples/bpf/xdp2_kern.c
+++ b/samples/bpf/xdp2_kern.c
@@ -14,7 +14,7 @@
 #include <linux/ipv6.h>
 #include "bpf_helpers.h"
 
-struct bpf_map_def SEC("maps") dropcnt = {
+struct bpf_map_def SEC("maps") rxcnt = {
 	.type = BPF_MAP_TYPE_PERCPU_ARRAY,
 	.key_size = sizeof(u32),
 	.value_size = sizeof(long),
@@ -65,7 +65,7 @@ int xdp_prog1(struct xdp_md *ctx)
 	long *value;
 	u16 h_proto;
 	u64 nh_off;
-	u32 index;
+	u32 ipproto;
 
 	nh_off = sizeof(*eth);
 	if (data + nh_off > data_end)
@@ -93,17 +93,17 @@ int xdp_prog1(struct xdp_md *ctx)
 	}
 
 	if (h_proto == htons(ETH_P_IP))
-		index = parse_ipv4(data, nh_off, data_end);
+		ipproto = parse_ipv4(data, nh_off, data_end);
 	else if (h_proto == htons(ETH_P_IPV6))
-		index = parse_ipv6(data, nh_off, data_end);
+		ipproto = parse_ipv6(data, nh_off, data_end);
 	else
-		index = 0;
+		ipproto = 0;
 
-	value = bpf_map_lookup_elem(&dropcnt, &index);
+	value = bpf_map_lookup_elem(&rxcnt, &ipproto);
 	if (value)
 		*value += 1;
 
-	if (index == 17) {
+	if (ipproto == IPPROTO_UDP) {
 		swap_src_dst_mac(data);
 		rc = XDP_TX;
 	}
-- 
2.8.2

Re: [PATCH net-next 1/3] net/mlx4_en: use READ_ONCE when freeing xdp_prog

[Alexei Starovoitov]

On Wed, Jul 20, 2016 at 05:22:33PM -0700, Brenden Blanco wrote:
> For consistency, and in order to hint at the synchronous nature of the
> xdp_prog field, use READ_ONCE in the destroy path of the ring. All
> occurrences should now use either READ_ONCE or xchg.
> 
> Signed-off-by: Brenden Blanco <bblanco@plumgrid.com>

Suggested-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>

Re: [PATCH net-next 2/3] rtnl: protect do_setlink from IFLA_XDP_ATTACHED

[Alexei Starovoitov]

On Wed, Jul 20, 2016 at 05:22:34PM -0700, Brenden Blanco wrote:
> The IFLA_XDP_ATTACHED nested attribute is meant for read-only, and while
> do_setlink properly ignores it, it should be more paranoid and reject
> commands that try to set it.
> 
> Signed-off-by: Brenden Blanco <bblanco@plumgrid.com>

Suggested-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>

Re: [PATCH net-next 3/3] bpf: make xdp sample variable names more meaningful

[Alexei Starovoitov]

On Wed, Jul 20, 2016 at 05:22:35PM -0700, Brenden Blanco wrote:
> The naming choice of index is not terribly descriptive, and dropcnt is
> in fact incorrect for xdp2. Pick better names for these: ipproto and
> rxcnt.
> 
> Signed-off-by: Brenden Blanco <bblanco@plumgrid.com>

Thanks!
Acked-by: Alexei Starovoitov <ast@kernel.org>

Re: [PATCH net-next 0/3] misc cleanups for xdp

[David Miller]

From: Brenden Blanco <bblanco@plumgrid.com>
Date: Wed, 20 Jul 2016 17:22:32 -0700

> This addresses several of the non-blocking comments left over from the
> xdp patch set. See individual patches for details.

Series applied, thanks.