feat(net): add netlink route support by Anekoique · Pull Request #25 · Starry-OS/arceos

Anekoique · 2025-12-04T17:46:16Z

This PR introduces support for Netlink Route, referencing the implementations of Asterinas and DragonOS.

Some Linux apps rely on simple Netlink mechanisms to retrieve link or address information. This support is crucial for adapting FastDDS and ROS 2 to Starry.

Now, Starry can run ip link show and ip addr show normally:

starry:~# ip link show
1: lo: <LOOPBACK,UP> mtu 1500 qlen 65536
    link/loopback
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qlen 65536
    link/ether
starry:~# ip addr show
1: lo: <LOOPBACK,UP> mtu 1500 qlen 65536
    link/loopback
    inet 127.0.0.1/8 scope host lo
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qlen 65536
    link/ether
    inet 10.0.2.15/24 scope host eth0

Additional Fix (FS):

The original function is_valid checked file open flags based on std::fs::OpenOptions, which does not conform to POSIX semantics (e.g., it was unable to create read-only files), leading to failures in some programs.
I have renamed is_valid to check_options to perform simple flag checks. More comprehensive checks may need to be added later, referencing https://man7.org/linux/man-pages/man2/open.2.html.

Copilot

Pull request overview

This PR adds Netlink Route support to enable Linux applications to retrieve network interface and address information through the Netlink protocol. The implementation references Asterinas and DragonOS designs and is essential for adapting FastDDS and ROS 2 to Starry.

Key changes:

Implements Netlink Route protocol with support for ip link show and ip addr show commands
Adds comprehensive message parsing and attribute handling infrastructure
Fixes filesystem open flag validation to conform to POSIX semantics

Reviewed changes

Copilot reviewed 31 out of 32 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
modules/axnet/src/socket.rs	Adds Netlink socket variant and address type to socket infrastructure
modules/axnet/src/service.rs	Exposes device iterator for Netlink route queries
modules/axnet/src/netlink/table.rs	Implements bind table for managing Netlink socket bindings and multicast groups
modules/axnet/src/netlink/route/mod.rs	Implements route transport layer with send/recv operations
modules/axnet/src/netlink/route/message/segment/*.rs	Defines route message segment structures for links and addresses
modules/axnet/src/netlink/route/message/attr/*.rs	Implements attribute parsing for link and address information
modules/axnet/src/netlink/route/handle.rs	Handles GetLink and GetAddr requests with device filtering
modules/axnet/src/netlink/receiver.rs	Implements message queue and receiver for async message handling
modules/axnet/src/netlink/mod.rs	Main Netlink socket implementation with bind/connect/send/recv operations
modules/axnet/src/netlink/message/*.rs	Core message parsing infrastructure with segments and attributes
modules/axnet/src/netlink/addr.rs	Netlink socket addressing with port and multicast group support
modules/axnet/src/lib.rs	Adds device index counter and Netlink module integration
modules/axnet/src/device/*.rs	Extends device trait with type, flags, index, and IP information
modules/axnet/Cargo.toml	Adds dependencies for Netlink feature
modules/axfs/src/highlevel/file.rs	Refactors file open options validation to fix POSIX compliance
api/axfeat/Cargo.toml	Adds netlink feature flag

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2025-12-04T17:47:44Z

modules/axnet/src/netlink/message/segment/mod.rs

+        align_up(payload_len, NLMSG_ALIGN) - payload_len
+    }
+
+    fn lecacy_padding_len() -> usize {


Corrected spelling of 'lecacy' to 'legacy'.

Suggested change

fn lecacy_padding_len() -> usize {

fn legacy_padding_len() -> usize {

Copilot · 2025-12-04T17:47:44Z

modules/axnet/src/netlink/route/message/segment/addr.rs

+        const DEPRECATED	 = 0x20;
+        const TENTATIVE		 = 0x40;
+        const PERMANENT		 = 0x80;
+        const MANAGETEMPADDR = 0x100;
+        const NOPREFIXROUTE	 = 0x200;
+        const MCAUTOJOIN	 = 0x400;
+        const STABLE_PRIVACY = 0x800;


Inconsistent whitespace in flag definitions; use consistent spacing between flag name and value.

Suggested change

const DEPRECATED = 0x20;

const TENTATIVE = 0x40;

const PERMANENT = 0x80;

const MANAGETEMPADDR = 0x100;

const NOPREFIXROUTE = 0x200;

const MCAUTOJOIN = 0x400;

const STABLE_PRIVACY = 0x800;

const DEPRECATED = 0x20;

const TENTATIVE = 0x40;

const PERMANENT = 0x80;

const MANAGETEMPADDR = 0x100;

const NOPREFIXROUTE = 0x200;

const MCAUTOJOIN = 0x400;

const STABLE_PRIVACY = 0x800;

Copilot · 2025-12-04T17:47:45Z

modules/axnet/src/netlink/route/handle.rs

+    },
+};
+
+/// An helper function to access the route protocol bind table.


Corrected grammar: 'An helper' should be 'A helper'.

Suggested change

/// An helper function to access the route protocol bind table.

/// A helper function to access the route protocol bind table.

Copilot · 2025-12-04T17:47:45Z

modules/axnet/src/device/loopback.rs

+        self.index
+    }
+
+    fn ipv4_addr(&self) -> Option<Ipv4Addr> {


Return type uses Ipv4Addr from std::net, but Device trait definition shows Option<Ipv4Address> from smoltcp. The trait definition should use the stdlib type for consistency, or this implementation should match the trait's smoltcp type.

Copilot · 2025-12-04T17:47:45Z

modules/axnet/src/device/ethernet.rs

+        self.index
+    }
+
+    fn ipv4_addr(&self) -> Option<Ipv4Address> {


Return type uses Ipv4Address from smoltcp, which is inconsistent with loopback device's use of Ipv4Addr from std::net. Both implementations should use the same type for consistency.

Copilot · 2025-12-04T17:47:45Z

modules/axnet/src/netlink/message/segment/header.rs

+    /// Reference: <https://elixir.bootlin.com/linux/v6.13/source/include/uapi/linux/netlink.h#L86>.
+    pub struct AckFlags: u16 {
+        const CAPPED = 0x100;
+        const ACK_TLVS = 0x100;


Two different flags have the same value 0x100. This appears to be a copy-paste error - these flags should have distinct values.

Suggested change

const ACK_TLVS = 0x100;

const ACK_TLVS = 0x200;

Copilot

Pull request overview

Copilot reviewed 33 out of 33 changed files in this pull request and generated 6 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-05T12:49:08Z

modules/axnet/src/netlink/route/mod.rs

+        let initial_remaining = src.remaining();
+        let mut reader = BufReader::new(src);
+        loop {
+            let mut segment = match RouteSegment::read_from(&mut reader)? {
+                ContinueRead::Parsed(seg) => seg,
+                ContinueRead::Skipped => continue,
+                ContinueRead::SkippedErr(_error_segment) => {
+                    // TODO: Should unicast `_error_segment` as NLMSG_ERROR with original seq/pid to `port` instead of silently continuing.
+                    continue;
+                }
+            };
+            let header = segment.header_mut();
+            // Set the pid to the sender's port if it's zero
+            if header.pid == 0 {
+                header.pid = port
+            }
+            self.handle_request(&segment, port);
+            return Ok(initial_remaining);
+        }


send() returns immediately after handling the first parsed segment and returns initial_remaining regardless of how many segments were actually consumed/handled. Netlink requests can contain multiple segments; the current logic will silently ignore trailing segments while still reporting success. Consider looping until the input buffer is fully consumed (handling each parsed segment), and return the actual number of bytes consumed.

Suggested change

let initial_remaining = src.remaining();

let mut reader = BufReader::new(src);

loop {

let mut segment = match RouteSegment::read_from(&mut reader)? {

ContinueRead::Parsed(seg) => seg,

ContinueRead::Skipped => continue,

ContinueRead::SkippedErr(_error_segment) => {

// TODO: Should unicast `_error_segment` as NLMSG_ERROR with original seq/pid to `port` instead of silently continuing.

continue;

}

};

let header = segment.header_mut();

// Set the pid to the sender's port if it's zero

if header.pid == 0 {

header.pid = port

}

self.handle_request(&segment, port);

return Ok(initial_remaining);

}

let mut reader = BufReader::new(src);

let mut total_consumed = 0usize;

loop {

let before = reader.remaining();

if before == 0 {

break;

}

let parsed = RouteSegment::read_from(&mut reader)?;

let after = reader.remaining();

total_consumed += before.saturating_sub(after);

match parsed {

ContinueRead::Parsed(mut segment) => {

let header = segment.header_mut();

// Set the pid to the sender's port if it's zero

if header.pid == 0 {

header.pid = port

}

self.handle_request(&segment, port);

}

ContinueRead::Skipped => {

continue;

}

ContinueRead::SkippedErr(_error_segment) => {

// TODO: Should unicast `_error_segment` as NLMSG_ERROR with original seq/pid to `port` instead of silently continuing.

continue;

}

}

}

Ok(total_consumed)

Copilot · 2026-03-05T12:49:09Z

modules/axfs/src/highlevel/file.rs

-                }
-            }
+        if self.create_new && !self.create {
+            return Err(VfsError::AlreadyExists);


create_new without create currently returns VfsError::AlreadyExists, which implies the path exists even though this is just an invalid/meaningless flag combination (POSIX treats O_EXCL as only meaningful with O_CREAT). Returning InvalidInput (or implicitly enabling create when create_new is set) would be more accurate.

Suggested change

return Err(VfsError::AlreadyExists);

return Err(VfsError::InvalidInput);

Copilot · 2026-03-05T12:49:09Z

modules/axnet/src/device/loopback.rs

+    fn ipv4_addr(&self) -> Option<Ipv4Addr> {
+        Some(Ipv4Addr::new(127, 0, 0, 1))
+    }


Device::ipv4_addr is defined to return smoltcp::wire::Ipv4Address (see device/mod.rs), but LoopbackDevice implements it as core::net::Ipv4Addr, which will not satisfy the trait and should fail to compile. Use the same Ipv4Address type as the trait (or change the trait to consistently use core::net::Ipv4Addr).

Copilot · 2026-03-05T12:49:09Z

modules/axnet/src/netlink/message/attr/noattr.rs

+use axio::BufRead;
+
+use super::{Attribute, AttrHeader};
+use crate::netlink::message::ContinueRead;


ContinueRead is not re-exported from crate::netlink::message, so use crate::netlink::message::ContinueRead; will not resolve. Import it from crate::netlink::message::result::ContinueRead (or re-export it in message/mod.rs).

Suggested change

use crate::netlink::message::ContinueRead;

use crate::netlink::message::result::ContinueRead;

Copilot · 2026-03-05T12:49:10Z

modules/axnet/src/netlink/message/attr/mod.rs

+use memory_addr::align_up;
+pub use noattr::NoAttr;
+
+use crate::netlink::message::{ContinueRead, NLMSG_ALIGN};


ContinueRead is not defined in crate::netlink::message (it lives in message::result), so this import will fail to compile. Import ContinueRead from super::result::ContinueRead or crate::netlink::message::result::ContinueRead instead.

Suggested change

use crate::netlink::message::{ContinueRead, NLMSG_ALIGN};

use crate::netlink::message::NLMSG_ALIGN;

use crate::netlink::message::result::ContinueRead;

Copilot · 2026-03-05T12:49:10Z

modules/axnet/src/netlink/message/segment/mod.rs

+    common::SegmentCommon,
+    header::*,
+};
+use super::{ContinueRead, NLMSG_ALIGN};


ContinueRead is not available as super::ContinueRead here (it is declared in message::result). This use super::{ContinueRead, NLMSG_ALIGN}; import should instead pull ContinueRead from super::result::ContinueRead (or re-export ContinueRead from message/mod.rs).

Suggested change

use super::{ContinueRead, NLMSG_ALIGN};

use super::NLMSG_ALIGN;

use super::result::ContinueRead;

AsakuraMizu · 2026-03-06T04:52:05Z

Hi @Anekoique,

We are migrating our development repository to https://github.com/arceos-org/arceos/tree/dev. If your development has stalled, could you rebase it there? Thanks!

Copilot AI review requested due to automatic review settings December 4, 2025 17:46

Copilot AI reviewed Dec 4, 2025

View reviewed changes

This was referenced Dec 4, 2025

feat: add netlink route support and misc fixes Starry-OS/StarryOS#71

Draft

feat: add BufWriter and extend buffered I/O traits and impls Starry-OS/axio#1

Closed

AsakuraMizu force-pushed the dev branch from de4c455 to 9253879 Compare December 8, 2025 09:19

AsakuraMizu force-pushed the dev branch from 8195846 to 9a39bb7 Compare December 16, 2025 06:37

Anekoique added 2 commits March 5, 2026 14:35

fix(fs): fix open option validation

5fa4958

feat(net): add netlink route

d109ed6

Anekoique force-pushed the feat/netlink branch from 6fa9b30 to d109ed6 Compare March 5, 2026 12:40

Anekoique marked this pull request as ready for review March 5, 2026 12:41

Copilot AI review requested due to automatic review settings March 5, 2026 12:41

Copilot started reviewing on behalf of Anekoique March 5, 2026 12:42 View session

Copilot AI reviewed Mar 5, 2026

View reviewed changes

	fn lecacy_padding_len() -> usize {
	fn legacy_padding_len() -> usize {

	/// An helper function to access the route protocol bind table.
	/// A helper function to access the route protocol bind table.

	return Err(VfsError::AlreadyExists);
	return Err(VfsError::InvalidInput);

	use crate::netlink::message::ContinueRead;
	use crate::netlink::message::result::ContinueRead;

	use crate::netlink::message::{ContinueRead, NLMSG_ALIGN};
	use crate::netlink::message::NLMSG_ALIGN;
	use crate::netlink::message::result::ContinueRead;

	use super::{ContinueRead, NLMSG_ALIGN};
	use super::NLMSG_ALIGN;
	use super::result::ContinueRead;

Conversation

Anekoique commented Dec 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Dec 4, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 4, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 4, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 4, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 4, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 4, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 5, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 5, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 5, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 5, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 5, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 5, 2026

Choose a reason for hiding this comment

Uh oh!

AsakuraMizu commented Mar 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Anekoique commented Dec 4, 2025 •

edited

Loading