Skip to contentSkip to conversation

netfilter: flowtable: consolidate net_device field in nft_forward_info struct

PATCH Superseded Pablo Neira Ayuso 2 participants 7 messages Jul 27

Subsystems (7)

v2 of 5 Superseded Latest is v5 →

lore
Share & tools
lore

Pablo Neira Ayuso

author maintainer
#0 lore raw

PATCH v2 1/5 netfilter: flowtable: consolidate net_device field in nft_forward_info struct

info->indev and info->outdev refer to the same device, a single info->dev field is sufficient.

While at it, remove unused router parameter from the flowtable path discovery function.

---
v2: no changes

 net/netfilter/nf_flow_table_path.c | 20 ++++++++------------
 1 file changed, 8 insertions(+), 12 deletions(-)
net/netfilter/nf_flow_table_path.c +8 -12
index 98c03b487f52..261bb44d08eb 100644
@@ -42,8 +42,7 @@ static bool nft_is_valid_ether_device(const struct net_device *dev)
 	return true;
 }
 
-static int nft_dev_fill_forward_path(const struct nf_flow_route *route,
-				     const struct dst_entry *dst_cache,
+static int nft_dev_fill_forward_path(const struct dst_entry *dst_cache,
 				     const struct nf_conn *ct,
 				     enum ip_conntrack_dir dir, u8 *ha,
 				     struct net_device_path_stack *stack)
@@ -76,8 +75,7 @@ static int nft_dev_fill_forward_path(const struct nf_flow_route *route,
 }
 
 struct nft_forward_info {
-	const struct net_device *indev;
-	const struct net_device *outdev;
+	const struct net_device *dev;
 	struct id {
 		__u16	id;
 		__be16	proto;
@@ -109,7 +107,7 @@ static int nft_dev_path_info(const struct net_device_path_stack *stack,
 		case DEV_PATH_VLAN:
 		case DEV_PATH_PPPOE:
 		case DEV_PATH_TUN:
-			info->indev = path->dev;
+			info->dev = path->dev;
 			if (is_zero_ether_addr(info->h_source))
 				memcpy(info->h_source, path->dev->dev_addr, ETH_ALEN);
 
@@ -179,10 +177,9 @@ static int nft_dev_path_info(const struct net_device_path_stack *stack,
 			return -1;
 		}
 	}
-	info->outdev = info->indev;
 
 	if (nf_flowtable_hw_offload(flowtable) &&
-	    nft_is_valid_ether_device(info->indev))
+	    nft_is_valid_ether_device(info->dev))
 		info->xmit_type = FLOW_OFFLOAD_XMIT_DIRECT;
 
 	return 0;
@@ -255,17 +252,16 @@ static int nft_dev_forward_path(const struct nft_pktinfo *pkt,
 	unsigned char ha[ETH_ALEN];
 	int i;
 
-	if (nft_dev_fill_forward_path(route, dst, ct, dir, ha, &stack) < 0 ||
+	if (nft_dev_fill_forward_path(dst, ct, dir, ha, &stack) < 0 ||
 	    nft_dev_path_info(&stack, &info, ha, &ft->data) < 0)
 		return -ENOENT;
 
-	if (!nft_flowtable_find_dev(info.indev, ft))
+	if (!nft_flowtable_find_dev(info.dev, ft))
 		return -ENOENT;
 
-	if (info.outdev)
-		route->tuple[dir].out.ifindex = info.outdev->ifindex;
+	route->tuple[!dir].in.ifindex = info.dev->ifindex;
+	route->tuple[dir].out.ifindex = info.dev->ifindex;
 
-	route->tuple[!dir].in.ifindex = info.indev->ifindex;
 	for (i = 0; i < info.num_encaps; i++) {
 		route->tuple[!dir].in.encap[i].id = info.encap[i].id;
 		route->tuple[!dir].in.encap[i].proto = info.encap[i].proto;
Signature · 1 line
2.47.3
author maintainer
#1 lore raw

PATCH v2 2/5 netfilter: flowtable: consolidate flowtable device check

Check that device belongs to the flowtable right after the flowtable discovery path. This is a preparation patch to obtain the dst entry from the .fill_forward_path in tunnels.

No functional changes are intended.

---
v2: no changes

 net/netfilter/nf_flow_table_path.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)
net/netfilter/nf_flow_table_path.c +9 -6
index 261bb44d08eb..8f04a4487897 100644
@@ -90,9 +90,12 @@ struct nft_forward_info {
 	enum flow_offload_xmit_type xmit_type;
 };
 
+static bool nft_flowtable_find_dev(const struct net_device *dev,
+				   struct nft_flowtable *ft);
+
 static int nft_dev_path_info(const struct net_device_path_stack *stack,
 			     struct nft_forward_info *info,
-			     unsigned char *ha, struct nf_flowtable *flowtable)
+			     unsigned char *ha, struct nft_flowtable *ft)
 {
 	const struct net_device_path *path;
 	int i;
@@ -178,10 +181,13 @@ static int nft_dev_path_info(const struct net_device_path_stack *stack,
 		}
 	}
 
-	if (nf_flowtable_hw_offload(flowtable) &&
+	if (nf_flowtable_hw_offload(&ft->data) &&
 	    nft_is_valid_ether_device(info->dev))
 		info->xmit_type = FLOW_OFFLOAD_XMIT_DIRECT;
 
+	if (!nft_flowtable_find_dev(info->dev, ft))
+		return -1;
+
 	return 0;
 }
 
@@ -253,10 +259,7 @@ static int nft_dev_forward_path(const struct nft_pktinfo *pkt,
 	int i;
 
 	if (nft_dev_fill_forward_path(dst, ct, dir, ha, &stack) < 0 ||
-	    nft_dev_path_info(&stack, &info, ha, &ft->data) < 0)
-		return -ENOENT;
-
-	if (!nft_flowtable_find_dev(info.dev, ft))
+	    nft_dev_path_info(&stack, &info, ha, ft) < 0)
 		return -ENOENT;
 
 	route->tuple[!dir].in.ifindex = info.dev->ifindex;
Signature · 1 line
2.47.3
author maintainer
#2 lore raw

PATCH v2 3/5 net: do not advance stack index from dev_fwd_path()

In reply to Pablo Neira Ayuso

Update stack index from dev_fill_forward_path() instead, once the forward path slot has been populated.

Adjust existing drivers that recycle dev_fill_forward_path() as a way to reach the mtk SoC WDMA info.

---
v2: new in this series to address sashiko report on possible access
to uninitialized path in stack in case tunnel route reference need to be dropped.

 drivers/net/ethernet/airoha/airoha_ppe.c        | 2 +-
 drivers/net/ethernet/mediatek/mtk_ppe_offload.c | 2 +-
 net/core/dev.c                                  | 6 ++++--
 3 files changed, 6 insertions(+), 4 deletions(-)
drivers/net/ethernet/airoha/airoha_ppe.c +1 -1
index e4c3644dd6ec..72755ca974fa 100644
@@ -292,7 +292,7 @@ static int airoha_ppe_get_wdma_info(struct net_device *dev, const u8 *addr,
 	if (err)
 		return err;
 
-	path = &stack.path[stack.num_paths - 1];
+	path = &stack.path[stack.num_paths];
 	if (path->type != DEV_PATH_MTK_WDMA)
 		return -EINVAL;
 
drivers/net/ethernet/mediatek/mtk_ppe_offload.c +1 -1
index cc8c4ef8038f..7dab360b8be3 100644
@@ -107,7 +107,7 @@ mtk_flow_get_wdma_info(struct net_device *dev, const u8 *addr, struct mtk_wdma_i
 	if (err)
 		return err;
 
-	path = &stack.path[stack.num_paths - 1];
+	path = &stack.path[stack.num_paths];
 	if (path->type != DEV_PATH_MTK_WDMA)
 		return -1;
 
net/core/dev.c +4 -2
index c1c1be1a6962..ca65d1dcd604 100644
@@ -742,9 +742,9 @@ EXPORT_SYMBOL_GPL(dev_fill_metadata_dst);
 
 static struct net_device_path *dev_fwd_path(struct net_device_path_stack *stack)
 {
-	int k = stack->num_paths++;
+	int k = stack->num_paths;
 
-	if (k >= NET_DEVICE_PATH_STACK_MAX)
+	if (k + 1 >= NET_DEVICE_PATH_STACK_MAX)
 		return NULL;
 
 	return &stack->path[k];
@@ -773,6 +773,7 @@ int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
 		if (ret < 0)
 			return -1;
 
+		stack->num_paths++;
 		if (WARN_ON_ONCE(last_dev == ctx.dev))
 			return -1;
 	}
@@ -785,6 +786,7 @@ int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
 		return -1;
 	path->type = DEV_PATH_ETHERNET;
 	path->dev = ctx.dev;
+	stack->num_paths++;
 
 	return ret;
 }
Signature · 1 line
2.47.3
maintainer Acked-by on v2 3/5
#3 lore raw

Re: [PATCH nf-next,v2 3/5] net: do not advance stack index from dev_fwd_path()

Quoted 7 lines

Update stack index from dev_fill_forward_path() instead, once the forward path slot has been populated.

Adjust existing drivers that recycle dev_fill_forward_path() as a way to reach the mtk SoC WDMA info.

Quoted 41 lines

---
v2: new in this series to address sashiko report on possible access
to uninitialized path in stack in case tunnel route reference need to be dropped.

 drivers/net/ethernet/airoha/airoha_ppe.c        | 2 +-
 drivers/net/ethernet/mediatek/mtk_ppe_offload.c | 2 +-
 net/core/dev.c                                  | 6 ++++--
 3 files changed, 6 insertions(+), 4 deletions(-)
drivers/net/ethernet/airoha/airoha_ppe.c +1 -1
index e4c3644dd6ec..72755ca974fa 100644
@@ -292,7 +292,7 @@ static int airoha_ppe_get_wdma_info(struct net_device *dev, const u8 *addr,
 	if (err)
 		return err;
 
-	path = &stack.path[stack.num_paths - 1];
+	path = &stack.path[stack.num_paths];
 	if (path->type != DEV_PATH_MTK_WDMA)
 		return -EINVAL;
 
drivers/net/ethernet/mediatek/mtk_ppe_offload.c +1 -1
index cc8c4ef8038f..7dab360b8be3 100644
@@ -107,7 +107,7 @@ mtk_flow_get_wdma_info(struct net_device *dev, const u8 *addr, struct mtk_wdma_i
 	if (err)
 		return err;
 
-	path = &stack.path[stack.num_paths - 1];
+	path = &stack.path[stack.num_paths];
 	if (path->type != DEV_PATH_MTK_WDMA)
 		return -1;
 
net/core/dev.c +0 -0
index c1c1be1a6962..ca65d1dcd604 100644

net/core/dev.c EXPORT_SYMBOL_GPL(dev_fill_metadata_dst); · line 747

Quoted 8 lines
net/core/dev.c +2 -2
@@ -742,9 +742,9 @@ EXPORT_SYMBOL_GPL(dev_fill_metadata_dst);
 
 static struct net_device_path *dev_fwd_path(struct net_device_path_stack *stack)
 {
-	int k = stack->num_paths++;
+	int k = stack->num_paths;
 
-	if (k >= NET_DEVICE_PATH_STACK_MAX)
+	if (k + 1 >= NET_DEVICE_PATH_STACK_MAX)

nit: do we really need k in this case? (ignore the comment if you do not need to repost).

Regards,
Lorenzo

Quoted 22 lines
net/core/dev.c +2 -0
 		return NULL;
 
 	return &stack->path[k];
@@ -773,6 +773,7 @@ int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
 		if (ret < 0)
 			return -1;
 
+		stack->num_paths++;
 		if (WARN_ON_ONCE(last_dev == ctx.dev))
 			return -1;
 	}
@@ -785,6 +786,7 @@ int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
 		return -1;
 	path->type = DEV_PATH_ETHERNET;
 	path->dev = ctx.dev;
+	stack->num_paths++;
 
 	return ret;
 }
Signature · 1 line
2.47.3
author maintainer
#4 lore raw

Re: [PATCH nf-next,v2 3/5] net: do not advance stack index from dev_fwd_path()

On Mon, Jul 27, 2026 at 02:42:19PM +0200, Lorenzo Bianconi wrote:
[...]

Quoted 11 lines
Quoted 11 lines
net/core/dev.c +1 -1
index c1c1be1a6962..ca65d1dcd604 100644
@@ -742,9 +742,9 @@ EXPORT_SYMBOL_GPL(dev_fill_metadata_dst);
 
 static struct net_device_path *dev_fwd_path(struct net_device_path_stack *stack)
 {
-	int k = stack->num_paths++;
+	int k = stack->num_paths;
 
Quoted 5 lines

net/core/dev.c EXPORT_SYMBOL_GPL(dev_fill_metadata_dst); · line 747

Quoted 2 lines
net/core/dev.c +1 -1
-	if (k >= NET_DEVICE_PATH_STACK_MAX)
+	if (k + 1 >= NET_DEVICE_PATH_STACK_MAX)

nit: do we really need k in this case? (ignore the comment if you do not need to repost).

Yes, I can do that. I will have to repost anyway, sashiko did not kick in.

author maintainer
#5 lore raw

PATCH v2 4/5 net: pass dst via net_device_path in dev_fill_forward_path()

In reply to Pablo Neira Ayuso

Add dst_entry to tunnel device path, this will allow us to remove a duplicated route lookup.

This is a preparation patch to retrieve the tunnel route directly from the .fill_forward_path. This new dst_entry in the tunnel will be used by a follow up patch.

Since dst_release() works fine on NULL interface, this is still noop until the flowtable starts using this.

Add a new dev_fill_forward_path_release() function to drop the refcount on the tunnel device route, in case of error out. Export it so it can be use in case of error out, to drop the refcount on the tunnel route at a later stage.

---
v2: - move dst field in net_device_path at the beginning of tun field per Lorenzo
- fix dev_fill_forward_path_release() initial index value
- remove useless ret assignment per Lorenzo

 include/linux/netdevice.h |  2 ++
 net/core/dev.c            | 36 +++++++++++++++++++++++++++++++-----
 2 files changed, 33 insertions(+), 5 deletions(-)
include/linux/netdevice.h +2 -0
index 8db25b79573e..62cfad7e6b79 100644
@@ -892,6 +892,7 @@ struct net_device_path {
 			u8		h_dest[ETH_ALEN];
 		} encap;
 		struct {
+			struct dst_entry *dst;
 			union {
 				struct in_addr	src_v4;
 				struct in6_addr	src_v6;
@@ -3427,6 +3428,7 @@ int dev_get_iflink(const struct net_device *dev);
 int dev_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb);
 int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
 			  struct net_device_path_stack *stack);
+void dev_fill_forward_path_release(struct net_device_path_stack *stack);
 struct net_device *dev_get_by_name(struct net *net, const char *name);
 struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
 struct net_device *__dev_get_by_name(struct net *net, const char *name);
net/core/dev.c +31 -5
index ca65d1dcd604..b6e3700b674f 100644
@@ -750,6 +750,27 @@ static struct net_device_path *dev_fwd_path(struct net_device_path_stack *stack)
 	return &stack->path[k];
 }
 
+void dev_fill_forward_path_release(struct net_device_path_stack *stack)
+{
+	struct net_device_path *path;
+	int k;
+
+	if (stack->num_paths == 0)
+		return;
+
+	for (k = stack->num_paths - 1; k >= 0; k--) {
+		path = &stack->path[k];
+		switch (path->type) {
+		case DEV_PATH_TUN:
+			dst_release(path->tun.dst);
+			break;
+		default:
+			break;
+		}
+	}
+}
+EXPORT_SYMBOL_GPL(dev_fill_forward_path_release);
+
 int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
 			  struct net_device_path_stack *stack)
 {
@@ -766,16 +787,16 @@ int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
 		last_dev = ctx.dev;
 		path = dev_fwd_path(stack);
 		if (!path)
-			return -1;
+			goto err_out;
 
 		memset(path, 0, sizeof(struct net_device_path));
 		ret = ctx.dev->netdev_ops->ndo_fill_forward_path(&ctx, path);
 		if (ret < 0)
-			return -1;
+			goto err_out;
 
 		stack->num_paths++;
 		if (WARN_ON_ONCE(last_dev == ctx.dev))
-			return -1;
+			goto err_out;
 	}
 
 	if (!ctx.dev)
@@ -783,12 +804,17 @@ int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
 
 	path = dev_fwd_path(stack);
 	if (!path)
-		return -1;
+		goto err_out;
+
 	path->type = DEV_PATH_ETHERNET;
 	path->dev = ctx.dev;
 	stack->num_paths++;
 
-	return ret;
+	return 0;
+err_out:
+	dev_fill_forward_path_release(stack);
+
+	return -1;
 }
 EXPORT_SYMBOL_GPL(dev_fill_forward_path);
 
Signature · 1 line
2.47.3
author maintainer
#6 lore raw

PATCH v2 5/5 netfilter: flowtable: release tunnel route on error when building forward path

In reply to Pablo Neira Ayuso

nft_flow_tunnel_update_route() can lazy fail, leaving an incomplete forward path set ip. The route lookup also happens twice, once from dev_fill_forward_path() and again in this aforementioned function.

Update ipip and ip6ip6 not to release the dst_entry and pass it on via the tunnel forward path information.

In case of failure when setting up the forwarding path, release the tunnel dst that was provided via dev_fill_forward_path().

---
v2: - fix missing dst_release() call in v6.
- use dev_fill_forward_path_release() in err_out path.

 net/ipv4/ipip.c                    |  2 +-
 net/ipv6/ip6_tunnel.c              |  4 +-
 net/netfilter/nf_flow_table_path.c | 65 ++++++++----------------------
 3 files changed, 21 insertions(+), 50 deletions(-)
net/ipv4/ipip.c +1 -1
index d1aa048a6099..0a4fa1351dd7 100644
@@ -370,10 +370,10 @@ static int ipip_fill_forward_path(struct net_device_path_ctx *ctx,
 	path->tun.src_v4.s_addr = tiph->saddr;
 	path->tun.dst_v4.s_addr = tiph->daddr;
 	path->tun.l3_proto = IPPROTO_IPIP;
+	path->tun.dst = &rt->dst;
 	path->dev = ctx->dev;
 
 	ctx->dev = rt->dst.dev;
--	ip_rt_put(rt);
 
 	return 0;
 }
net/ipv6/ip6_tunnel.c +3 -1
index bf8e40af60b0..82eeed183001 100644
@@ -1864,12 +1864,14 @@ static int ip6_tnl_fill_forward_path(struct net_device_path_ctx *ctx,
 		path->tun.src_v6 = t->parms.laddr;
 		path->tun.dst_v6 = t->parms.raddr;
 		path->tun.l3_proto = IPPROTO_IPV6;
+		path->tun.dst = dst;
 		path->dev = ctx->dev;
 		ctx->dev = dst->dev;
 	}
 
 	err = dst->error;
-	dst_release(dst);
+	if (err)
+		dst_release(dst);
 
 	return err;
 }
net/netfilter/nf_flow_table_path.c +17 -48

137 lines not on this page · raw message

Signature · 1 line
2.47.3