diff --git a/examples/subnet_nat.rs b/examples/subnet_nat.rs
new file mode 100644
index 0000000..4b4fdd3
--- /dev/null
+++ b/examples/subnet_nat.rs
@@ -0,0 +1,201 @@
+// SPDX-License-Identifier: MIT
+use std::env;
+use std::net::Ipv4Addr;
+
+use futures::stream::TryStreamExt;
+use netlink_packet_route::tc::{self, nlas::matchall, nlas::nat, Action};
+use rtnetlink::{new_connection, Error, Handle};
+
+#[tokio::main]
+async fn main() -> Result<(), ()> {
+    env_logger::init();
+
+    // Parse the command line
+    let args: Vec<String> = env::args().collect();
+    if args.len() != 4 {
+        usage();
+        return Ok(());
+    }
+
+    let (old_subnet, prefix_len) = match split_cidr(&args[2]) {
+        Ok(addrs) => addrs,
+        Err(s) => {
+            eprintln!("{}", s);
+            return Err(());
+        }
+    };
+
+    let (new_subnet, _) = match split_cidr(&args[3]) {
+        Ok(addrs) => addrs,
+        Err(s) => {
+            eprintln!("{}", s);
+            return Err(());
+        }
+    };
+
+    let (connection, handle, _) = new_connection().unwrap();
+    tokio::spawn(connection);
+    let link_index =
+        match get_link_index_by_name(handle.clone(), args[1].clone()).await {
+            Ok(i) => i,
+            Err(_) => {
+                eprintln!("Link: {} not found", args[1]);
+                return Err(());
+            }
+        };
+
+    // Create qdiscs on the interface.
+    create_ingress_qdisc(handle.clone(), link_index).await?;
+    create_egress_qdisc(&args[1]).await?;
+
+    // Add tc nat action filters
+
+    // First add the egress filter. This is equivalent to the following command:
+    // tc filter add dev $devname \
+    //     parent 10: protocol ip prio 10 \
+    //     matchall action nat egress $old_subnet $new_subnet
+    let nat_params = nat::Nla::Parms(
+        nat::TcNat::default()
+            .set_new_addr(new_subnet)
+            .set_old_addr(old_subnet)
+            .set_prefix(prefix_len)
+            .egress(),
+    );
+
+    let mut nat_act = Action::default();
+    nat_act.nlas.push(tc::ActNla::Kind(nat::KIND.to_string()));
+    nat_act
+        .nlas
+        .push(tc::ActNla::Options(vec![tc::ActOpt::Nat(nat_params)]));
+
+    let msg = handle
+        .traffic_filter(link_index as i32)
+        .add()
+        .parent(0x10 << 16)
+        .priority(10)
+        .protocol(0x0008)
+        .matchall(vec![matchall::Nla::Act(vec![nat_act])])
+        .unwrap();
+
+    if let Err(res) = msg.execute().await {
+        eprintln!("{}", res);
+        return Err(());
+    }
+
+    // Then add the ingress filter. This is equivalent to the following command:
+    // tc filter add dev $devname \
+    //     parent 10: protocol ip prio 10 \
+    //     matchall action nat ingress $new_subnet $old_subnet
+    let nat_params = nat::Nla::Parms(
+        nat::TcNat::default()
+            .set_new_addr(old_subnet)
+            .set_old_addr(new_subnet)
+            .set_prefix(prefix_len),
+    );
+
+    let mut nat_act = Action::default();
+    nat_act.nlas.push(tc::ActNla::Kind(nat::KIND.to_string()));
+    nat_act
+        .nlas
+        .push(tc::ActNla::Options(vec![tc::ActOpt::Nat(nat_params)]));
+
+    let msg = handle
+        .traffic_filter(link_index as i32)
+        .add()
+        .parent(0xffff << 16)
+        .priority(10)
+        .protocol(0x0008)
+        .matchall(vec![matchall::Nla::Act(vec![nat_act])])
+        .unwrap();
+
+    if let Err(res) = msg.execute().await {
+        eprintln!("{}", res);
+        return Err(());
+    }
+
+    Ok(())
+}
+
+// TODO: There is no code in netlink-packet-route for egress qisc types yet.
+// This shells out to the `tc` command instead, and should be replaced when
+// the appropriate message types are available in netlink-packet-route.
+async fn create_egress_qdisc(devname: &str) -> Result<(), ()> {
+    match std::process::Command::new("tc")
+        .args(&[
+            "qdisc", "add", "dev", devname, "root", "handle", "10:", "htb",
+        ])
+        .output()
+    {
+        Err(e) => {
+            eprintln!("Error creating egress qdisc: {}", e);
+            Err(())
+        }
+        Ok(output) if output.status.success() => Ok(()),
+        Ok(_) => {
+            eprintln!("Error creating egress qdisc:");
+            Err(())
+        }
+    }
+}
+
+async fn create_ingress_qdisc(handle: Handle, index: u32) -> Result<(), ()> {
+    if let Err(e) = handle
+        .qdisc()
+        .add(index as i32)
+        .handle(0xffff, 0)
+        .ingress()
+        .execute()
+        .await
+    {
+        eprintln!("Error creating ingress qdisc: {e}");
+        return Err(());
+    }
+
+    Ok(())
+}
+
+async fn get_link_index_by_name(
+    handle: Handle,
+    name: String,
+) -> Result<u32, Error> {
+    let mut links = handle.link().get().match_name(name).execute();
+    let link = (links.try_next().await?).expect("Link not found");
+    Ok(link.header.index)
+}
+
+fn split_cidr(cidr_text: &str) -> Result<(Ipv4Addr, usize), String> {
+    let (prefix, len) = cidr_text
+        .split_once('/')
+        .ok_or(format!("'{}' is not a valid CIDR", cidr_text))?;
+    let address: Ipv4Addr = prefix.parse().map_err(|e| {
+        format!("'{}' cannot be parsed to an IP address: {}", prefix, e)
+    })?;
+    let prefix_len: usize = len
+        .parse()
+        .map_err(|_| format!("'{}' is not a valid prefix length", len))?;
+
+    Ok((address, prefix_len))
+}
+
+fn usage() {
+    eprintln!(
+        "usage:
+    cargo run --example subnet_nat -- <devname> <old_subnet> <new_subnet>
+
+This is will have the same effect as:
+    tc qdisc add dev $devname root handle 10: htb
+    tc qdisc add dev $devname ingress handle ffff
+
+    tc filter add dev $devname parent 10: protocol ip prio 10 matchall action nat egress $old_subnet $new_subnet
+    tc filter add dev $devname parent ffff: protocol ip prio 10 matchall action nat ingress $new_subnet $old_subnet
+
+Note that you need to run this program as root. Instead of running cargo as root,
+build the example normally:
+
+    cd rtnetlink ; cargo build --example add_tc_qdisc_ingress 
+
+Then find the binary in the target directory:
+
+    cd ../target/debug/example ; sudo ./add_tc_qdisc_ingress <index>"
+    );
+}
diff --git a/src/traffic_control/add_filter.rs b/src/traffic_control/add_filter.rs
index 73a5e6e..756079c 100644
--- a/src/traffic_control/add_filter.rs
+++ b/src/traffic_control/add_filter.rs
@@ -118,6 +118,32 @@ impl TrafficFilterNewRequest {
         self
     }
 
+    /// The matchall filter matches every packet and applies actions, etc
+    /// Equivalent to `tc filter ... matchall`
+    pub fn matchall(
+        mut self,
+        data: Vec<tc::matchall::Nla>,
+    ) -> Result<Self, Error> {
+        if self
+            .message
+            .nlas
+            .iter()
+            .any(|nla| matches!(nla, tc::Nla::Kind(_)))
+        {
+            return Err(Error::InvalidNla(
+                "message kind has already been set.".to_string(),
+            ));
+        }
+
+        self.message
+            .nlas
+            .push(tc::Nla::Kind(tc::matchall::KIND.to_string()));
+        self.message.nlas.push(tc::Nla::Options(
+            data.into_iter().map(tc::TcOpt::Matchall).collect(),
+        ));
+        Ok(self)
+    }
+
     /// The 32bit filter allows to match arbitrary bitfields in the packet.
     /// Equivalent to `tc filter ... u32`.
     pub fn u32(mut self, options: &[TcFilterU32Option]) -> Result<Self, Error> {