mirror of
https://github.com/amnezia-vpn/amneziawg-linux-kernel-module.git
synced 2026-10-13 21:34:34 +00:00
[GH-ISSUE #222] kmod v3.1.20260812 as client: handshake ignored by amneziawg-go v3.1.20260813+ servers with HeaderProtectionKey (go client works) #127
Labels
No labels
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
amnezia-vpn/amneziawg-linux-kernel-module#127
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @Fedoseew on GitHub (Aug 26, 2026).
Original GitHub issue: https://github.com/amnezia-vpn/amneziawg-linux-kernel-module/issues/222
Originally assigned to: @ygurov on GitHub.
Environment
kmod-amneziawg 3.1.20260812+amneziawg-tools 3.1.20260812(built from the release tags), netifd proto with full 3.1 option supportSummary
The kernel module at tag
v3.1.20260812cannot complete a handshake with a server running current amneziawg-go, while amneziawg-go as a client connects fine with the same config from the same network. amneziawg-go received the wire-format fixfix: wrong trailerLen for HandshakeCookie message(2026-08-13) after the last kmod tag, so the two implementations appear to have diverged.Config (client side, from Amnezia Premium export; keys/endpoint omitted)
Jc=6, Jmin=10, Jmax=80, S1=612, S2=311, S3=1204, S4=12, H1–H4=1–4,
HeaderProtectionKey set, I1 set (signature packet), ContentPaddingAddition=10-100,
RekeyAfterTime=100-120, RekeyTimeout=3-8, RejectAfterTime=150-180,
KeepaliveTimeout=7-13, MaxHandshakeAttempts=15-20, PersistentKeepalive=25-35, PSK set.
awg showconfirms every parameter above is applied to the interface (includingheader protection keyand ranges).Symptom
transfer: 0 B received, 4.40 KiB sentand growing), zero bytes ever received; no handshake within 60 s.Cross-checks (why this points at the kmod)
Ask
A kmod release aligned with amneziawg-go ≥ v3.1.20260813 (or a pointer if client-side 3.1 against current premium servers needs something else).
@press-vm commented on GitHub (Sep 9, 2026):
The newest kmod tag (v3.1.20260906, via #232) fixes the I1-I5/junk-packet trailer issue but doesn't touch the cookie-reply path — wg_socket_send_buffer_as_reply_to_skb() is untouched by that PR. #189 ("fix: wrong cookie-reply size constant, UB shift in bogus_endpoints") looks like it might be the missing piece for the trailerLen-for-HandshakeCookie parity with amneziawg-go's #179. Still reproducible with the latest tag as far as I can tell.