public inbox for virtualization@lists.linux-foundation.org
 help / color / mirror / Atom feed
From: Alexander Graf <graf@amazon.com>
To: <virtualization@lists.linux.dev>
Cc: <linux-kernel@vger.kernel.org>, <netdev@vger.kernel.org>,
	<kvm@vger.kernel.org>, <eperezma@redhat.com>,
	Jason Wang <jasowang@redhat.com>, <mst@redhat.com>,
	Stefano Garzarella <sgarzare@redhat.com>,
	Stefan Hajnoczi <stefanha@redhat.com>,
	<bcm-kernel-feedback-list@broadcom.com>,
	Arnd Bergmann <arnd@arndb.de>,
	"Greg Kroah-Hartman" <gregkh@linuxfoundation.org>,
	Jonathan Corbet <corbet@lwn.net>,
	Bryan Tan <bryan-bt.tan@broadcom.com>,
	Vishnu Dasa <vishnu.dasa@broadcom.com>,
	<nh-open-source@amazon.com>
Subject: [PATCH v2] vsock: add G2H fallback for CIDs not owned by H2G transport
Date: Mon, 2 Mar 2026 19:49:26 +0000	[thread overview]
Message-ID: <20260302194926.90378-1-graf@amazon.com> (raw)

Vsock maintains a single CID number space which can be used to
communicate to the host (G2H) or to a child-VM (H2G). The current logic
assumes that G2H is only relevant for CID <= 2 because these target the
hypervisor. However, in environments like Nitro Enclaves, an instance
that hosts vhost_vsock powered VMs may still want to communicate to
Enclaves that are reachable at higher CIDs through virtio-vsock-pci.

Vsock introduced VMADDR_FLAG_TO_HOST to allow user space applications
to clearly express a desire to talk to the host instead of a guest via
the passed target CID. However, users may not actually know which one
they want to talk to and the application ecosystem has not picked up a
way for users to specify that desire.

Instead, make it easy for users and introduce a G2H fallback mechanism:
when user space attempts to connect to a CID and the H2G transport
(vhost-vsock / VMCI) does not own it, automatically route the connection
through the G2H transport. This provides a single unified CID address
space where vhost-registered CIDs go to nested VMs and all other CIDs
are routed to the hypervisor.

To give user space at least a hint that the kernel applied this logic,
automatically set the VMADDR_FLAG_TO_HOST on the remote address so it
can determine the path taken via getpeername().

To force the system back into old behavior, provide a sysctl
(net.vsock.g2h_fallback, defaults to 1).

Signed-off-by: Alexander Graf <graf@amazon.com>

---

v1 -> v2:

  - Rebase on 7.0, include namespace support
  - Add net.vsock.g2h_fallback sysctl
  - Rework description
  - Set VMADDR_FLAG_TO_HOST automatically
  - Add VMCI support
  - Update vsock_assign_transport() comment
---
 Documentation/admin-guide/sysctl/net.rst | 22 ++++++++++++++++++++++
 drivers/misc/vmw_vmci/vmci_context.c     |  1 +
 drivers/vhost/vsock.c                    | 13 +++++++++++++
 include/linux/vmw_vmci_api.h             |  1 +
 include/net/af_vsock.h                   |  3 +++
 net/vmw_vsock/af_vsock.c                 | 20 +++++++++++++++++++-
 net/vmw_vsock/vmci_transport.c           |  6 ++++++
 7 files changed, 65 insertions(+), 1 deletion(-)

diff --git a/Documentation/admin-guide/sysctl/net.rst b/Documentation/admin-guide/sysctl/net.rst
index 3b2ad61995d4..cc364baa9021 100644
--- a/Documentation/admin-guide/sysctl/net.rst
+++ b/Documentation/admin-guide/sysctl/net.rst
@@ -602,3 +602,25 @@ it does not modify the current namespace or any existing children.
 
 A namespace with ``ns_mode`` set to ``local`` cannot change
 ``child_ns_mode`` to ``global`` (returns ``-EPERM``).
+
+g2h_fallback
+------------
+
+Controls whether connections to CIDs not owned by the host-to-guest (H2G)
+transport automatically fall back to the guest-to-host (G2H) transport.
+
+When enabled, if a connect targets a CID that the H2G transport (e.g.
+vhost-vsock) does not serve, the connection is routed via the G2H transport
+(e.g. virtio-vsock) instead. This allows a host running both nested VMs
+(via vhost-vsock) and sibling VMs reachable through the hypervisor (e.g.
+Nitro Enclaves) to address both using a single CID space, without requiring
+applications to set ``VMADDR_FLAG_TO_HOST``.
+
+When the fallback is taken, ``VMADDR_FLAG_TO_HOST`` is automatically set on
+the remote address so that userspace can determine the path via
+``getpeername()``.
+
+Values:
+
+	- 0 - Connections to CIDs < 3 get handled by G2H, others by H2G.
+	- 1 - Connections to CIDs not owned by H2G fall back to G2H. (default)
diff --git a/drivers/misc/vmw_vmci/vmci_context.c b/drivers/misc/vmw_vmci/vmci_context.c
index 19ca00feed6e..577296784df5 100644
--- a/drivers/misc/vmw_vmci/vmci_context.c
+++ b/drivers/misc/vmw_vmci/vmci_context.c
@@ -364,6 +364,7 @@ bool vmci_ctx_exists(u32 cid)
 	rcu_read_unlock();
 	return exists;
 }
+EXPORT_SYMBOL_GPL(vmci_ctx_exists);
 
 /*
  * Retrieves VMCI context corresponding to the given cid.
diff --git a/drivers/vhost/vsock.c b/drivers/vhost/vsock.c
index 054f7a718f50..319e3a690108 100644
--- a/drivers/vhost/vsock.c
+++ b/drivers/vhost/vsock.c
@@ -91,6 +91,18 @@ static struct vhost_vsock *vhost_vsock_get(u32 guest_cid, struct net *net)
 	return NULL;
 }
 
+static bool vhost_transport_has_remote_cid(struct vsock_sock *vsk, u32 cid)
+{
+	struct sock *sk = sk_vsock(vsk);
+	struct net *net = sock_net(sk);
+	bool found;
+
+	rcu_read_lock();
+	found = vhost_vsock_get(cid, net) != NULL;
+	rcu_read_unlock();
+	return found;
+}
+
 static void
 vhost_transport_do_send_pkt(struct vhost_vsock *vsock,
 			    struct vhost_virtqueue *vq)
@@ -424,6 +436,7 @@ static struct virtio_transport vhost_transport = {
 		.module                   = THIS_MODULE,
 
 		.get_local_cid            = vhost_transport_get_local_cid,
+		.has_remote_cid           = vhost_transport_has_remote_cid,
 
 		.init                     = virtio_transport_do_socket_init,
 		.destruct                 = virtio_transport_destruct,
diff --git a/include/linux/vmw_vmci_api.h b/include/linux/vmw_vmci_api.h
index 41764a684423..c412d17c572f 100644
--- a/include/linux/vmw_vmci_api.h
+++ b/include/linux/vmw_vmci_api.h
@@ -37,6 +37,7 @@ int vmci_doorbell_create(struct vmci_handle *handle, u32 flags,
 int vmci_doorbell_destroy(struct vmci_handle handle);
 u32 vmci_get_context_id(void);
 bool vmci_is_context_owner(u32 context_id, kuid_t uid);
+bool vmci_ctx_exists(u32 cid);
 int vmci_register_vsock_callback(vmci_vsock_cb callback);
 
 int vmci_event_subscribe(u32 event,
diff --git a/include/net/af_vsock.h b/include/net/af_vsock.h
index 533d8e75f7bb..0aeb25642827 100644
--- a/include/net/af_vsock.h
+++ b/include/net/af_vsock.h
@@ -179,6 +179,9 @@ struct vsock_transport {
 	/* Addressing. */
 	u32 (*get_local_cid)(void);
 
+	/* Check if this transport serves a specific remote CID. */
+	bool (*has_remote_cid)(struct vsock_sock *vsk, u32 remote_cid);
+
 	/* Read a single skb */
 	int (*read_skb)(struct vsock_sock *, skb_read_actor_t);
 
diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c
index 2f7d94d682cb..b41bc734d6c0 100644
--- a/net/vmw_vsock/af_vsock.c
+++ b/net/vmw_vsock/af_vsock.c
@@ -210,6 +210,8 @@ static const struct vsock_transport *transport_dgram;
 static const struct vsock_transport *transport_local;
 static DEFINE_MUTEX(vsock_register_mutex);
 
+static int vsock_g2h_fallback = 1;
+
 /**** UTILS ****/
 
 /* Each bound VSocket is stored in the bind hash table and each connected
@@ -547,7 +549,8 @@ static void vsock_deassign_transport(struct vsock_sock *vsk)
  *    g2h is not loaded, will use local transport;
  *  - remote CID <= VMADDR_CID_HOST or h2g is not loaded or remote flags field
  *    includes VMADDR_FLAG_TO_HOST flag value, will use guest->host transport;
- *  - remote CID > VMADDR_CID_HOST will use host->guest transport;
+ *  - remote CID > VMADDR_CID_HOST will use host->guest transport if h2g has
+ *    registered that CID, otherwise will use guest->host transport (overlay);
  */
 int vsock_assign_transport(struct vsock_sock *vsk, struct vsock_sock *psk)
 {
@@ -584,6 +587,12 @@ int vsock_assign_transport(struct vsock_sock *vsk, struct vsock_sock *psk)
 		else if (remote_cid <= VMADDR_CID_HOST || !transport_h2g ||
 			 (remote_flags & VMADDR_FLAG_TO_HOST))
 			new_transport = transport_g2h;
+		else if (vsock_g2h_fallback &&
+			 transport_h2g->has_remote_cid &&
+			 !transport_h2g->has_remote_cid(vsk, remote_cid)) {
+			vsk->remote_addr.svm_flags |= VMADDR_FLAG_TO_HOST;
+			new_transport = transport_g2h;
+		}
 		else
 			new_transport = transport_h2g;
 		break;
@@ -2879,6 +2888,15 @@ static struct ctl_table vsock_table[] = {
 		.mode		= 0644,
 		.proc_handler	= vsock_net_child_mode_string
 	},
+	{
+		.procname	= "g2h_fallback",
+		.data		= &vsock_g2h_fallback,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= SYSCTL_ZERO,
+		.extra2		= SYSCTL_ONE,
+	},
 };
 
 static int __net_init vsock_sysctl_register(struct net *net)
diff --git a/net/vmw_vsock/vmci_transport.c b/net/vmw_vsock/vmci_transport.c
index 4296ca1183f1..de3dff52c566 100644
--- a/net/vmw_vsock/vmci_transport.c
+++ b/net/vmw_vsock/vmci_transport.c
@@ -2045,6 +2045,11 @@ static u32 vmci_transport_get_local_cid(void)
 	return vmci_get_context_id();
 }
 
+static bool vmci_transport_has_remote_cid(struct vsock_sock *vsk, u32 cid)
+{
+	return vmci_ctx_exists(cid);
+}
+
 static struct vsock_transport vmci_transport = {
 	.module = THIS_MODULE,
 	.init = vmci_transport_socket_init,
@@ -2074,6 +2079,7 @@ static struct vsock_transport vmci_transport = {
 	.notify_send_post_enqueue = vmci_transport_notify_send_post_enqueue,
 	.shutdown = vmci_transport_shutdown,
 	.get_local_cid = vmci_transport_get_local_cid,
+	.has_remote_cid = vmci_transport_has_remote_cid,
 };
 
 static bool vmci_check_transport(struct vsock_sock *vsk)
-- 
2.47.1




Amazon Web Services Development Center Germany GmbH
Tamara-Danz-Str. 13
10243 Berlin
Geschaeftsfuehrung: Christof Hellmis, Andreas Stieger
Eingetragen am Amtsgericht Charlottenburg unter HRB 257764 B
Sitz: Berlin
Ust-ID: DE 365 538 597


             reply	other threads:[~2026-03-02 19:49 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-03-02 19:49 Alexander Graf [this message]
2026-03-03  7:55 ` [syzbot ci] Re: vsock: add G2H fallback for CIDs not owned by H2G transport syzbot ci
2026-03-03 10:28 ` [PATCH v2] " Stefano Garzarella
2026-03-03 12:32   ` Stefano Garzarella
2026-03-03 11:59 ` Michael S. Tsirkin

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260302194926.90378-1-graf@amazon.com \
    --to=graf@amazon.com \
    --cc=arnd@arndb.de \
    --cc=bcm-kernel-feedback-list@broadcom.com \
    --cc=bryan-bt.tan@broadcom.com \
    --cc=corbet@lwn.net \
    --cc=eperezma@redhat.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=jasowang@redhat.com \
    --cc=kvm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mst@redhat.com \
    --cc=netdev@vger.kernel.org \
    --cc=nh-open-source@amazon.com \
    --cc=sgarzare@redhat.com \
    --cc=stefanha@redhat.com \
    --cc=virtualization@lists.linux.dev \
    --cc=vishnu.dasa@broadcom.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox