Passthrough.Flush hands its whole lane to a new tio.BatchWriter in one
WriteBatch call when the queue implements it, and NewSingleQueue
exposes WriteBatch when its source does. On darwin, tun.WriteBatch
writes up to 64 packets per sendmsg_x on the utun control socket,
which is a connected datagram socket, so xnu takes the batched
sendit_x path rather than one sendit per message. utun registers no
send_list handler, so the kernel still passes each packet to utun on
its own: this saves syscalls and fd locking, not per-packet kernel work.
Error handling follows what sendmsg_x reports:
- A short count with no error (a packet past the send buffer
mid-call, or a later send failing with ENOBUFS): the remainder goes
out in the next sendmsg_x.
- EMSGSIZE: only returned when the first packet is too large, before
the kernel takes any, so that packet alone goes through Write, which
reports it, and the rest are batched again.
- ENOSYS, EPERM, EOPNOTSUPP: sendmsg_x is private API, so WriteBatch
logs once and falls back to writev per packet for good.
- Anything else: sendmsg_x returns the error without the count it
took, so the call is reported rather than retried, which could
deliver packets twice.
One loss mode has no userspace fix. sendit_x counts each packet as
sent once it is copied into an mbuf; if a later allocation fails, it
frees the copied prefix unsent, returns its length, and sendmsg_x
drops the ENOBUFS because the count is nonzero. That looks exactly
like a real short count, so under mbuf exhaustion WriteBatch silently
drops packets that writev would have delivered.
The iOS tun path (tun_ios.go) is unchanged.
Inside a macOS Network Extension, download goes from 640 to 832 Mbps
against the recvmsg_x change alone in three paired rounds, at 15% less
CPU per Mbps; the likely cause is that its utun charges more per write
than a raw one. Raw dnclient already runs download near line rate,
and there the CPU difference is within noise.
TestTunWriteBatch* round-trip through a connected AF_UNIX datagram
socketpair, which takes sendmsg_x the same way, over both the batched
and the fallback path: order, AF prefixes, skipped invalid packets,
and an oversize packet mid-call and first-in-call. Dropping Namelen
from msghdrX fails them. Mutation testing with gremlins leaves only
equivalent survivors in WriteBatch, plus which error is reported first
when several fail.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
newTunFromFd was a stub on darwin, so an embedder handed a utun by the OS had
no way to use it. A darwin network extension is given exactly that:
NEPacketTunnelProvider creates the device and applies addressing, mtu and
routes from its network settings, and a sandboxed extension cannot change them
afterwards. Adopting that fd means everything about moving packets is shared
with newTun and only the setup differs, so the tun tracks whether the host owns
it and leaves the system routing table and interface config alone when it does.
Co-authored-by: Nate Brown <nbrown.us@gmail.com>
* upgrade to yaml.v3
The main nice fix here is that maps unmarshal into `map[string]any`
instead of `map[any]any`, so it cleans things up a bit.
* add config.AsBool
Since yaml.v3 doesn't automatically convert yes to bool now, for
backwards compat
* use type aliases for m
* more cleanup
* more cleanup
* more cleanup
* go mod cleanup
With Nebula 1.4.0, if you create an unsafe_route that has a collision with an existing route on the system, the unsafe_route will be silently dropped (and the existing system route remains).
With Nebula 1.5.0, this same situation will cause Nebula to fail to start with an error (EEXIST).
This change restores the Nebula 1.4.0 behavior (but with a WARN log as well).
With the previous implementation, we check if route.MTU is greater than zero,
but it will always be because we set it to the default MTU in
parseUnsafeRoutes. This change leaves it as zero in parseUnsafeRoutes so
it can be examined later.
