Skip to content

Fix socketpair fd-table race when the table is full #1187

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
Weiteng Chen (CvvT) wants to merge 2 commits into
base: weiteng/fix_dup
Choose a base branch
from
+62 −23
Draft

Fix socketpair fd-table race when the table is full #1187

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
635d0a6
Fix socketpair fd-table race when the table is full
CvvT Aug 17, 2026
68bf6ee
fix rebase
CvvT Aug 17, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 4 additions & 12 deletions litebox_shim_linux/src/syscalls/file.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -107,7 +107,7 @@ impl<Platform: ShimPlatform, FS: ShimFS> FilesState<Platform, FS> {
self.insert_raw_fd_locked(&mut rds, typed_fd)
}

fn insert_raw_fd_locked<Subsystem: FdEnabledSubsystem>(
pub(super) fn insert_raw_fd_locked<Subsystem: FdEnabledSubsystem>(
&self,
rds: &mut litebox::fd::RawDescriptorStorage,
typed_fd: TypedFd<Subsystem>,
Expand Down Expand Up @@ -778,7 +778,7 @@ impl<Platform: ShimPlatform, FS: ShimFS> Task<Platform, FS> {
self.do_close_and_replace::<FS>(raw_fd, None)
}

fn remove_and_drop_descriptor<S: FdEnabledSubsystem>(&self, fd: &TypedFd<S>) {
pub(super) fn remove_and_drop_descriptor<S: FdEnabledSubsystem>(&self, fd: &TypedFd<S>) {
let entry = {
let mut dt = self.global.litebox.descriptor_table_mut();
dt.remove(fd)
Expand Down Expand Up @@ -1800,11 +1800,7 @@ impl<Platform: ShimPlatform, FS: ShimFS> Task<Platform, FS> {
drop(dt);
let files = self.files.borrow();
let raw_fd = files.insert_raw_fd(typed).map_err(|typed| {
self.global
.litebox
.descriptor_table_mut()
.remove(&typed)
.unwrap();
self.remove_and_drop_descriptor(&typed);
Errno::EMFILE
})?;
Ok(raw_fd.try_into().unwrap())
Expand Down Expand Up @@ -2042,11 +2038,7 @@ impl<Platform: ShimPlatform, FS: ShimFS> Task<Platform, FS> {
drop(dt);
let files = self.files.borrow();
let raw_fd = files.insert_raw_fd(typed).map_err(|typed| {
self.global
.litebox
.descriptor_table_mut()
.remove(&typed)
.unwrap();
self.remove_and_drop_descriptor(&typed);
Errno::EMFILE
})?;
Ok(raw_fd.try_into().unwrap())
Expand Down
69 changes: 58 additions & 11 deletions litebox_shim_linux/src/syscalls/net.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1016,7 +1016,7 @@ impl<Platform: ShimPlatform, FS: ShimFS> Task<Platform, FS> {
}

files.insert_raw_fd(typed).map_err(|typed| {
let _ = self.global.litebox.descriptor_table_mut().remove(&typed);
self.remove_and_drop_descriptor(&typed);
Errno::EMFILE
})?
}
Expand Down Expand Up @@ -1047,6 +1047,7 @@ impl<Platform: ShimPlatform, FS: ShimFS> Task<Platform, FS> {
.ok_or(Errno::EFAULT)?;
Ok(())
}

fn do_socketpair(
&self,
domain: AddressFamily,
Expand All @@ -1072,15 +1073,36 @@ impl<Platform: ShimPlatform, FS: ShimFS> Task<Platform, FS> {
assert!(old.is_none());
}
drop(dt);
let raw_fd1 = files.insert_raw_fd(typed1).map_err(|typed| {
let _ = self.global.litebox.descriptor_table_mut().remove(&typed);
Errno::EMFILE
})?;
let raw_fd2 = files.insert_raw_fd(typed2).map_err(|typed| {
self.do_close(raw_fd1).unwrap();
let _ = self.global.litebox.descriptor_table_mut().remove(&typed);
Errno::EMFILE
})?;
// Both inserts and the rollback of the first one must happen under a single
// acquisition of the raw descriptor store lock: otherwise a concurrent `close`
// could free the first socket's slot and another thread could take it over,
// making the rollback remove an unrelated file descriptor.
let mut rds = files.raw_descriptor_store.write();
let raw_fd1 = match files.insert_raw_fd_locked(&mut rds, typed1) {
Ok(raw_fd) => raw_fd,
Err(typed1) => {
drop(rds);
self.remove_and_drop_descriptor(&typed1);
self.remove_and_drop_descriptor(&typed2);
return Err(Errno::EMFILE);
}
};
let raw_fd2 = match files.insert_raw_fd_locked(&mut rds, typed2) {
Ok(raw_fd) => raw_fd,
Err(typed2) => {
let typed1 = rds
.fd_consume_raw_integer::<crate::syscalls::unix::UnixSocketSubsystem<
Platform,
FS,
>>(raw_fd1)
.unwrap();
drop(rds);
self.remove_and_drop_descriptor(&typed1);
self.remove_and_drop_descriptor(&typed2);
return Err(Errno::EMFILE);
}
};
drop(rds);
(raw_fd1, raw_fd2)
}
AddressFamily::INET | AddressFamily::INET6 | AddressFamily::NETLINK => {
Expand Down Expand Up @@ -1302,7 +1324,7 @@ impl<Platform: ShimPlatform, FS: ShimFS> Task<Platform, FS> {
}
drop(dt);
let raw_fd = files.insert_raw_fd(typed).map_err(|typed| {
let _ = self.global.litebox.descriptor_table_mut().remove(&typed);
self.remove_and_drop_descriptor(&typed);
Errno::EMFILE
})?;
Ok((raw_fd, peer_addr))
Expand Down Expand Up @@ -3341,6 +3363,31 @@ mod unix_tests {
unix_socketpair_bidirectional(SockType::Datagram, true);
}

#[test]
fn test_socketpair_race_with_concurrent_close() {
let task = init_platform(None);
task.files.borrow().set_fd_limit(3);

let stop = alloc::sync::Arc::new(core::sync::atomic::AtomicBool::new(false));
let stop_closer = stop.clone();
let closer = task.spawn_clone_for_test(move |task| {
while !stop_closer.load(core::sync::atomic::Ordering::Relaxed) {
let _ = task.sys_close(3);
}
});

for iter in 0..50_000 {
assert_eq!(
task.do_socketpair(AddressFamily::UNIX, SockType::Stream, SockFlags::empty(), 0),
Err(Errno::EMFILE),
"failed at iteration {iter}"
);
}

stop.store(true, core::sync::atomic::Ordering::Relaxed);
closer.join().unwrap();
}

fn unix_socket_recv_timeout(ty: SockType) {
let task = init_platform(None);
let (sock1, _sock2) = task
Expand Down