x86/kprobes: Fix crash when probing CS CALL instructionsgrafted

When using eBPF to probe CS CALL instructions within a function,
a crash can be triggered.

The eBPF tool probes offset 257 of the __hrtimer_run_queues()
function:

<__hrtimer_run_queues+249>:  nopl   0x0(%rax,%rax,1)
<__hrtimer_run_queues+254>:  mov    %r14,%rdi
<__hrtimer_run_queues+257>:  cs call <__x86_indirect_thunk_r12>
<__hrtimer_run_queues+263>:  mov    %eax,%r12d
<__hrtimer_run_queues+266>:  xchg   %ax,%ax
<__hrtimer_run_queues+268>:  mov    %r13,%rdi

Which triggers this crash:

  BUG: unable to handle page fault for address: 00000000000f41c9
  #PF: supervisor write access in kernel mode
  #PF: error_code(0x0002) - not-present page
  PGD 0 P4D 0
  Oops: 0002 [#1] SMP NOPTI
  CPU: 1 PID: 0 Comm: swapper/1 Kdump: loaded Tainted: P
  RIP: 0010:__hrtimer_run_queues+0x106/0x230

Note that __hrtimer_run_queues+0x106 is __hrtimer_run_queues+262, which is
at the 6th byte of the above CS CALL instruction. Since the CS CALL
instruction occupies 6 bytes, the exception occurred in the middle of that
call instruction.

The root cause is that when using eBPF tools to probe in the middle of a
function, a kprobe with INT3 is used as the underlying implementation.

During single-step emulation of the original CALL instruction,
int3_emulate_call() assumes that the probed CALL instruction is 5 bytes
long. However, the actual CS-prefixed CALL instruction occupies 6 bytes,
so it constructs an incorrect exception return address. When the CPU
returns from the kprobe handler, the next instruction to be executed is at
the address of the last byte of that CS CALL instruction. Coincidentally,
starting from that address, the CPU fetches and decodes a completely
different instruction, which ultimately triggers a kernel crash.

Fix the issue by using the actual instruction length obtained from
the instruction decoder when constructing the exception return
address, rather than relying on the hardcoded CALL_INSN_SIZE macro.

[ mingo: Refined the changelog ]

Fixes: 6256e668b7af ("x86/kprobes: Use int3 instead of debug trap for single-step")
Suggested-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Jinke Han <jinkehan@didiglobal.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Acked-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000

1 files changed, 229 insertions, 0 deletions

diff --git a/net/openvswitch/vport-netdev.c b/net/openvswitch/vport-netdev.c
new file mode 100644
index 000000000..44808cd0f
--- /dev/null
+++ b/net/openvswitch/vport-netdev.c
@@ -0,0 +1,229 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2007-2012 Nicira, Inc.
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/if_arp.h>
+#include <linux/if_bridge.h>
+#include <linux/if_vlan.h>
+#include <linux/kernel.h>
+#include <linux/llc.h>
+#include <linux/rtnetlink.h>
+#include <linux/skbuff.h>
+#include <linux/openvswitch.h>
+#include <linux/export.h>
+
+#include <net/ip_tunnels.h>
+#include <net/rtnetlink.h>
+
+#include "datapath.h"
+#include "vport.h"
+#include "vport-internal_dev.h"
+#include "vport-netdev.h"
+
+static struct vport_ops ovs_netdev_vport_ops;
+
+/* Must be called with rcu_read_lock. */
+static void netdev_port_receive(struct sk_buff *skb)
+{
+	struct vport *vport;
+
+	vport = ovs_netdev_get_vport(skb->dev);
+	if (unlikely(!vport))
+		goto error;
+
+	if (unlikely(skb_warn_if_lro(skb)))
+		goto error;
+
+	/* Make our own copy of the packet.  Otherwise we will mangle the
+	 * packet for anyone who came before us (e.g. tcpdump via AF_PACKET).
+	 */
+	skb = skb_share_check(skb, GFP_ATOMIC);
+	if (unlikely(!skb))
+		return;
+
+	if (skb->dev->type == ARPHRD_ETHER)
+		skb_push_rcsum(skb, ETH_HLEN);
+
+	ovs_vport_receive(vport, skb, skb_tunnel_info(skb));
+	return;
+error:
+	kfree_skb(skb);
+}
+
+/* Called with rcu_read_lock and bottom-halves disabled. */
+static rx_handler_result_t netdev_frame_hook(struct sk_buff **pskb)
+{
+	struct sk_buff *skb = *pskb;
+
+	if (unlikely(skb->pkt_type == PACKET_LOOPBACK))
+		return RX_HANDLER_PASS;
+
+	netdev_port_receive(skb);
+	return RX_HANDLER_CONSUMED;
+}
+
+static struct net_device *get_dpdev(const struct datapath *dp)
+{
+	struct vport *local;
+
+	local = ovs_vport_ovsl(dp, OVSP_LOCAL);
+	return local->dev;
+}
+
+static struct vport *ovs_netdev_link(struct vport *vport)
+{
+	int err;
+
+	if (WARN_ON_ONCE(!vport->dev)) {
+		err = -ENODEV;
+		goto error_free_vport;
+	}
+
+	rtnl_lock();
+	/* Do not link devices that are not registered to avoid a potential
+	 * race with the NETDEV_UNREGISTER notification in dp_device_event().
+	 */
+	if (vport->dev->reg_state != NETREG_REGISTERED) {
+		err = -ENODEV;
+		goto error_put_unlock;
+	}
+
+	err = netdev_master_upper_dev_link(vport->dev,
+					   get_dpdev(vport->dp),
+					   NULL, NULL, NULL);
+	if (err)
+		goto error_put_unlock;
+
+	err = netdev_rx_handler_register(vport->dev, netdev_frame_hook,
+					 vport);
+	if (err)
+		goto error_master_upper_dev_unlink;
+
+	dev_disable_lro(vport->dev);
+	dev_set_promiscuity(vport->dev, 1);
+	vport->dev->priv_flags |= IFF_OVS_DATAPATH;
+	rtnl_unlock();
+
+	return vport;
+
+error_master_upper_dev_unlink:
+	netdev_upper_dev_unlink(vport->dev, get_dpdev(vport->dp));
+error_put_unlock:
+	netdev_put(vport->dev, &vport->dev_tracker);
+	rtnl_unlock();
+error_free_vport:
+	ovs_vport_free(vport);
+	return ERR_PTR(err);
+}
+
+static struct vport *netdev_create(const struct vport_parms *parms)
+{
+	struct vport *vport;
+	int err;
+
+	vport = ovs_vport_alloc(0, &ovs_netdev_vport_ops, parms);
+	if (IS_ERR(vport))
+		return vport;
+
+	vport->dev = dev_get_by_name(ovs_dp_get_net(vport->dp), parms->name);
+	if (!vport->dev) {
+		err = -ENODEV;
+		goto error_free_vport;
+	}
+	netdev_tracker_alloc(vport->dev, &vport->dev_tracker, GFP_KERNEL);
+
+	/* Ensure that the provided name is not an alias. */
+	if (strcmp(parms->name, ovs_vport_name(vport))) {
+		err = -ENODEV;
+		goto error_put;
+	}
+
+	if (vport->dev->flags & IFF_LOOPBACK ||
+	    (vport->dev->type != ARPHRD_ETHER &&
+	     vport->dev->type != ARPHRD_NONE) ||
+	    ovs_is_internal_dev(vport->dev)) {
+		err = -EINVAL;
+		goto error_put;
+	}
+
+	return ovs_netdev_link(vport);
+error_put:
+	netdev_put(vport->dev, &vport->dev_tracker);
+error_free_vport:
+	ovs_vport_free(vport);
+	return ERR_PTR(err);
+}
+
+static void vport_netdev_free(struct rcu_head *rcu)
+{
+	struct vport *vport = container_of(rcu, struct vport, rcu);
+
+	netdev_put(vport->dev, &vport->dev_tracker);
+	ovs_vport_free(vport);
+}
+
+void ovs_netdev_detach_dev(struct vport *vport)
+{
+	ASSERT_RTNL();
+	netdev_rx_handler_unregister(vport->dev);
+	netdev_upper_dev_unlink(vport->dev,
+				netdev_master_upper_dev_get(vport->dev));
+	dev_set_promiscuity(vport->dev, -1);
+
+	/* paired with smp_mb() in netdev_destroy() */
+	smp_wmb();
+
+	vport->dev->priv_flags &= ~IFF_OVS_DATAPATH;
+}
+
+static void netdev_destroy(struct vport *vport)
+{
+	/* When called from ovs_db_notify_wq() after a dp_device_event(), the
+	 * port has already been detached, so we can avoid taking the RTNL by
+	 * checking this first.
+	 */
+	if (netif_is_ovs_port(vport->dev)) {
+		rtnl_lock();
+		/* Check again while holding the lock to ensure we don't race
+		 * with the netdev notifier and detach twice.
+		 */
+		if (netif_is_ovs_port(vport->dev))
+			ovs_netdev_detach_dev(vport);
+		rtnl_unlock();
+	}
+
+	/* paired with smp_wmb() in ovs_netdev_detach_dev() */
+	smp_mb();
+
+	call_rcu(&vport->rcu, vport_netdev_free);
+}
+
+/* Returns null if this device is not attached to a datapath. */
+struct vport *ovs_netdev_get_vport(struct net_device *dev)
+{
+	if (likely(netif_is_ovs_port(dev)))
+		return (struct vport *)
+			rcu_dereference_rtnl(dev->rx_handler_data);
+	else
+		return NULL;
+}
+
+static struct vport_ops ovs_netdev_vport_ops = {
+	.type		= OVS_VPORT_TYPE_NETDEV,
+	.create		= netdev_create,
+	.destroy	= netdev_destroy,
+	.send		= dev_queue_xmit,
+};
+
+int __init ovs_netdev_init(void)
+{
+	return ovs_vport_ops_register(&ovs_netdev_vport_ops);
+}
+
+void ovs_netdev_exit(void)
+{
+	ovs_vport_ops_unregister(&ovs_netdev_vport_ops);
+}