vsock: fix stale sk_err handling after a failed connect
Subsystems (4)
v6 cover vsock: fix stale sk_err handling after a failed connect
A socket whose connect() failed keeps sk_err set. If that socket is later reused as a listener, vsock_accept() rejects an unrelated incoming connection, and on virtio/hyperv the resulting child socket is leaked.
Patch 1 removes the listener's sk_err check from vsock_accept(), since no vsock transport ever sets sk_err on a TCP_LISTEN socket. This will fix what the syzbot reported.
Patch 2 removes vsock_sock.rejected, now unreachable after patch 1.
Patch 3 is a related but separate fix: vsock_connect() now consumes sk_err via sock_error() once it has been returned to userspace, so a failed blocking connect() doesn't keep reporting the same error a second time.
---
Changes in v6:
- Reorder if/else in vsock_accept() to put the smaller block first.
- Add historical context to patch 2
v5: https://lore.kernel.org/netdev/[email protected]/
- Split into a series
- Remove the now-unused rejected flag from vsock_sock
v4: https://lore.kernel.org/netdev/[email protected]/
- Remove sk_err checks from vsock_accept()
v3: https://lore.kernel.org/netdev/[email protected]/
- Fix truncated title and add annotations to reproducer steps.
v2: https://lore.kernel.org/netdev/[email protected]/
- Add reproducer steps to commit message.
v1: https://lore.kernel.org/netdev/[email protected]/
Nguyen Dinh Phi (3):
vsock: don't check the listener's sk_err in vsock_accept()
vsock: remove the now-unused rejected flag
vsock: use sock_error() to consume sk_err after a failed connect
include/net/af_vsock.h | 5 +--- net/vmw_vsock/af_vsock.c | 62 +++++++++++++--------------------------- 2 files changed, 21 insertions(+), 46 deletions(-)
Signature · 1 line
2.53.0
PATCH v6 1/3 vsock: don't check the listener's sk_err in vsock_accept()
Syzbot reported an issue which can be reproduced with these steps:
r0 = socket(AF_VSOCK, SOCK_STREAM, 0)
bind(r0, {VMADDR_CID_ANY, PORT})
connect(r0, {VMADDR_CID_LOCAL, PORT}) -> -1, EPROTO (self-connect)
listen(r0, backlog) -> 0
r1 = socket(AF_VSOCK, SOCK_STREAM, 0)
connect(r1, {VMADDR_CID_LOCAL, PORT}) -> 0
accept(r0) -> -1, EPROTO (stale sk_err)Basically, it creates a socket (r0) and triggers a self-connect after binding it. This self-connect fails with EPROTO because it loops back to r0 while the socket is still in the TCP_SYN_SENT state, causing it to be incorrectly dispatched to the connecting-client path. The unexpected packet type encountered there sets sk_err to EPROTO.
After that, it invokes a listen() call on the same socket. This listen() call succeeds because the kernel's listening path never inspects or clears sk_err. Then, a new socket (r1) is created as a normal client and connects to r0. However, vsock_accept() rejects this incoming connection because the listener's sk_err still holds the EPROTO error from the earlier failed self-connect.
This rejection causes the child socket created for r1's connection to never be freed on virtio or hyperv transports; only the VMCI transport implements pending_work to revisit and clean up a rejected socket.
For a non-blocking connect(), vsock_connect() may return -EINPROGRESS immediately, and vsock_connect_timeout() can later set sk->sk_err asynchronously.
Since no vsock transport ever sets sk_err on a socket while it is in TCP_LISTEN state, checking it in vsock_accept() serves no purpose and only carries forward errors left behind by earlier, unrelated connection attempts on the same socket. Remove the checks so accept() no longer rejects valid incoming connections because of a stale error, which also avoids the resource leak described above.
--- net/vmw_vsock/af_vsock.c | 10 +++------- 1 file changed, 3 insertions(+), 7 deletions(-)
net/vmw_vsock/af_vsock.c +3 -7
@@ -1893,7 +1893,7 @@ static int vsock_accept(struct socket *sock, struct socket *newsock, timeout =sock_rcvtimeo (listener, arg->flags &O_NONBLOCK );while ((connected =vsock_dequeue_accept (listener)) ==NULL && - listener->sk_err ==0 && timeout !=0 ) { + timeout !=0 ) {prepare_to_wait (sk_sleep (listener), &wait,TASK_INTERRUPTIBLE );release_sock (listener); timeout =schedule_timeout (timeout);
@@ -1906,13 +1906,9 @@ static int vsock_accept(struct socket *sock, struct socket *newsock, } } -if (listener->sk_err) { - err = -listener->sk_err; - }else if (!connected) { +if (!connected) { err = -EAGAIN ; - } - -if (connected) { + }else {sk_acceptq_removed (listener);lock_sock_nested (connected,SINGLE_DEPTH_NESTING );
Signature · 1 line
2.53.0
Re: [PATCH v6 1/3] vsock: don't check the listener's sk_err in vsock_accept()
Quoted 51 lines · Nguyen Dinh Phi
Syzbot reported an issue which can be reproduced with these steps:
r0 = socket(AF_VSOCK, SOCK_STREAM, 0) bind(r0, {VMADDR_CID_ANY, PORT}) connect(r0, {VMADDR_CID_LOCAL, PORT}) -> -1, EPROTO (self-connect) listen(r0, backlog) -> 0 r1 = socket(AF_VSOCK, SOCK_STREAM, 0) connect(r1, {VMADDR_CID_LOCAL, PORT}) -> 0 accept(r0) -> -1, EPROTO (stale sk_err)Basically, it creates a socket (r0) and triggers a self-connect after binding it. This self-connect fails with EPROTO because it loops back to r0 while the socket is still in the
TCP_SYN_SENTstate, causing it to be incorrectly dispatched to the connecting-client path. The unexpected packet type encountered there setssk_errto EPROTO.After that, it invokes a
listen()call on the same socket. Thislisten()call succeeds because the kernel's listening path never inspects or clearssk_err. Then, a new socket (r1) is created as a normal client and connects to r0. However,vsock_accept()rejects this incoming connection because the listener'ssk_errstill holds the EPROTO error from the earlier failed self-connect.This rejection causes the child socket created for r1's connection to never be freed on virtio or hyperv transports; only the VMCI transport implements
pending_workto revisit and clean up a rejected socket.For a non-blocking
connect(),vsock_connect()may return-EINPROGRESSimmediately, andvsock_connect_timeout()can later setsk->sk_errasynchronously.Since no vsock transport ever sets
sk_erron a socket while it is inTCP_LISTENstate, checking it invsock_accept()serves no purpose and only carries forward errors left behind by earlier, unrelated connection attempts on the same socket. Remove the checks soaccept()no longer rejects valid incoming connections because of a stale error, which also avoids the resource leak described above.Fixes d021c344051a ("VSOCK: Introduce VM Sockets")Reported-by [email protected]Suggested-by Michal Luczaj <[email protected]>Signed-off-by Nguyen Dinh Phi <[email protected]>---
net/vmw_vsock/af_vsock.c| 10 +++-------1 file changed, 3 insertions(+), 7 deletions(-)net/vmw_vsock/af_vsock.c +0
-0
net/vmw_vsock/af_vsock.c static int vsock_accept(struct socket *sock, struct socket *newsock, · line 1896
Quoted 5 lines
net/vmw_vsock/af_vsock.c +1
-1@@ -1893,7 +1893,7 @@ static int vsock_accept(struct socket *sock, struct socket *newsock, timeout =sock_rcvtimeo (listener, arg->flags &O_NONBLOCK );while ((connected =vsock_dequeue_accept (listener)) ==NULL && - listener->sk_err ==0 && timeout !=0 ) { + timeout !=0 ) {
^
nit: I think the alignment is wrong, maybe if you need to respin you can fix it
Anyway, this LGTM
PATCH v6 2/3 vsock: remove the now-unused rejected flag
In reply to Nguyen Dinh PhiAfter previous patch, the branch marking a socket rejected in vsock_accept() is unreachable, and nothing ever sets vsk->rejected elsewhere.
In fact, since commit d021c344051a ("VSOCK: Introduce VM Sockets"), where rejected was introduced, there has never been a path that sets sk_err on a listening socket, so that branch has been dead code since the beginning.
Therefore, we can remove the rejected field from vsock_sock structure.
--- include/net/af_vsock.h | 5 +---- net/vmw_vsock/af_vsock.c | 46 +++++++++++++--------------------------- 2 files changed, 16 insertions(+), 35 deletions(-)
include/net/af_vsock.h +1 -4
@@ -52,13 +52,10 @@ struct vsock_sock {* The listening socket is the head for both lists. Sockets created * for connection requests are placed in the pending list until they * are connected, at which point they are put in the accept queue list -* so they can be accepted in accept(). If accept() cannot accept the -* connection, it is marked as rejected so the cleanup function knows -* to clean up the socket. +* so they can be accepted in accept(). */ struct list_head pending_links;struct list_head accept_queue; -bool rejected;struct delayed_work connect_work;struct delayed_work pending_work;struct delayed_work close_work;
net/vmw_vsock/af_vsock.c +15 -31
78 lines not on this page · raw message
Signature · 1 line
2.53.0
Re: [PATCH v6 2/3] vsock: remove the now-unused rejected flag
Quoted 9 lines · Nguyen Dinh Phi
After previous patch, the branch marking a socket rejected in
vsock_accept()is unreachable, and nothing ever setsvsk->rejectedelsewhere.In fact, since commit
d021c344051a("VSOCK: Introduce VM Sockets"), whererejectedwas introduced, there has never been a path that setssk_erron a listening socket, so that branch has been dead code since the beginning.
Quoted 8 lines
Therefore, we can remove the
rejectedfield fromvsock_sockstructure.Suggested-by Stefano Garzarella <[email protected]>Signed-off-by Nguyen Dinh Phi <[email protected]>---
include/net/af_vsock.h| 5 +----net/vmw_vsock/af_vsock.c| 46 +++++++++++++---------------------------2 files changed, 16 insertions(+), 35 deletions(-)
Thanks for improving the commit description!
PATCH v6 3/3 vsock: use sock_error() to consume sk_err after a failed connect
In reply to Nguyen Dinh Phivsock_connect() returns sk_err to userspace but does not clear it:
if (sk->sk_err) {
err = -sk->sk_err;For a blocking connect() the error has already been delivered as connect()'s return value, so leaving it set causes subsequent operations like poll()/epoll() to keep reporting POLLERR even though the connect failure was already delivered.
The error should be consumed once it has been returned to userspace. Switch to sock_error(), which reads and clears sk_err atomically, matching the behavior of other protocol implementations such as __inet_stream_connect().
--- net/vmw_vsock/af_vsock.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-)
net/vmw_vsock/af_vsock.c +2 -4
@@ -1842,12 +1842,10 @@ static int vsock_connect(struct socket *sock, struct sockaddr_unsized *addr,prepare_to_wait (sk_sleep (sk), &wait,TASK_INTERRUPTIBLE ); } -if (sk->sk_err) { - err = -sk->sk_err; + err =sock_error (sk); +if (err) { sk->sk_state =TCP_CLOSE ; sock->state =SS_UNCONNECTED ; - }else { - err =0 ; } out_wait:
Signature · 1 line
2.53.0
This series was applied to netdev/net.git (main) by Paolo Abeni <[email protected]>:
Hello:
This series was applied to netdev/net.git (main)
by Paolo Abeni <[email protected]>:
Quoted 10 lines · On Fri, 14 Aug 2026 01:30:17 +0800 you wrote:
A socket whose
connect()failed keepssk_errset. If that socket is later reused as a listener,vsock_accept()rejects an unrelated incoming connection, and on virtio/hyperv the resulting child socket is leaked.Patch 1 removes the listener's
sk_errcheck fromvsock_accept(), since no vsock transport ever setssk_erron aTCP_LISTENsocket. This will fix what the syzbot reported.[...]
Here is the summary with links:
- [v6,1/3] vsock: don't check the listener's sk_err in vsock_accept()
https://git.kernel.org/netdev/net/c/b8c899cf5e7b - [v6,2/3] vsock: remove the now-unused rejected flag https://git.kernel.org/netdev/net/c/81fc0f369637 - [v6,3/3] vsock: use sock_error() to consume sk_err after a failed connect https://git.kernel.org/netdev/net/c/96cbf8999309
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html