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, 355 insertions, 0 deletions

diff --git a/fs/nfs/callback.c b/fs/nfs/callback.c
new file mode 100644
index 000000000..bc282b744
--- /dev/null
+++ b/fs/nfs/callback.c
@@ -0,0 +1,355 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * linux/fs/nfs/callback.c
+ *
+ * Copyright (C) 2004 Trond Myklebust
+ *
+ * NFSv4 callback handling
+ */
+
+#include <linux/completion.h>
+#include <linux/ip.h>
+#include <linux/module.h>
+#include <linux/sched/signal.h>
+#include <linux/sunrpc/svc.h>
+#include <linux/sunrpc/svcsock.h>
+#include <linux/nfs_fs.h>
+#include <linux/errno.h>
+#include <linux/mutex.h>
+#include <linux/freezer.h>
+#include <linux/sunrpc/svcauth_gss.h>
+#include <linux/sunrpc/bc_xprt.h>
+
+#include <net/inet_sock.h>
+
+#include "nfs4_fs.h"
+#include "callback.h"
+#include "internal.h"
+#include "netns.h"
+
+#define NFSDBG_FACILITY NFSDBG_CALLBACK
+
+struct nfs_callback_data {
+	unsigned int users;
+	struct svc_serv *serv;
+};
+
+static struct nfs_callback_data nfs_callback_info[NFS4_MAX_MINOR_VERSION + 1];
+static DEFINE_MUTEX(nfs_callback_mutex);
+static struct svc_program nfs4_callback_program;
+
+static int nfs4_callback_up_net(struct svc_serv *serv, struct net *net)
+{
+	const struct cred *cred = current_cred();
+	int ret;
+	struct nfs_net *nn = net_generic(net, nfs_net_id);
+
+	ret = svc_xprt_create(serv, "tcp", net, PF_INET,
+			      nfs_callback_set_tcpport, SVC_SOCK_ANONYMOUS,
+			      cred);
+	if (ret <= 0)
+		goto out_err;
+	nn->nfs_callback_tcpport = ret;
+	dprintk("NFS: Callback listener port = %u (af %u, net %x)\n",
+		nn->nfs_callback_tcpport, PF_INET, net->ns.inum);
+
+	ret = svc_xprt_create(serv, "tcp", net, PF_INET6,
+			      nfs_callback_set_tcpport, SVC_SOCK_ANONYMOUS,
+			      cred);
+	if (ret > 0) {
+		nn->nfs_callback_tcpport6 = ret;
+		dprintk("NFS: Callback listener port = %u (af %u, net %x)\n",
+			nn->nfs_callback_tcpport6, PF_INET6, net->ns.inum);
+	} else if (ret != -EAFNOSUPPORT)
+		goto out_err;
+	return 0;
+
+out_err:
+	return (ret) ? ret : -ENOMEM;
+}
+
+/*
+ * This is the NFSv4 callback kernel thread.
+ */
+static int
+nfs4_callback_svc(void *vrqstp)
+{
+	struct svc_rqst *rqstp = vrqstp;
+
+	svc_thread_init_status(rqstp, 0);
+
+	set_freezable();
+
+	while (!svc_thread_should_stop(rqstp))
+		svc_recv(rqstp, 0);
+
+	svc_exit_thread(rqstp);
+	return 0;
+}
+
+static inline void nfs_callback_bc_serv(u32 minorversion, struct rpc_xprt *xprt,
+		struct svc_serv *serv)
+{
+	if (minorversion)
+		/*
+		 * Save the svc_serv in the transport so that it can
+		 * be referenced when the session backchannel is initialized
+		 */
+		xprt->bc_serv = serv;
+}
+
+static int nfs_callback_start_svc(int minorversion, struct rpc_xprt *xprt,
+				  struct svc_serv *serv)
+{
+	int nrservs = nfs_callback_nr_threads;
+	int ret;
+
+	nfs_callback_bc_serv(minorversion, xprt, serv);
+
+	if (nrservs < NFS4_MIN_NR_CALLBACK_THREADS)
+		nrservs = NFS4_MIN_NR_CALLBACK_THREADS;
+
+	if (serv->sv_nrthreads == nrservs)
+		return 0;
+
+	ret = svc_set_num_threads(serv, 0, nrservs);
+	if (ret) {
+		svc_set_num_threads(serv, 0, 0);
+		return ret;
+	}
+	dprintk("nfs_callback_up: service started\n");
+	return 0;
+}
+
+static void nfs_callback_down_net(u32 minorversion, struct svc_serv *serv, struct net *net)
+{
+	struct nfs_net *nn = net_generic(net, nfs_net_id);
+
+	if (--nn->cb_users[minorversion])
+		return;
+
+	dprintk("NFS: destroy per-net callback data; net=%x\n", net->ns.inum);
+	svc_xprt_destroy_all(serv, net, false);
+}
+
+static int nfs_callback_up_net(int minorversion, struct svc_serv *serv,
+			       struct net *net, struct rpc_xprt *xprt)
+{
+	struct nfs_net *nn = net_generic(net, nfs_net_id);
+	int ret;
+
+	if (nn->cb_users[minorversion]++)
+		return 0;
+
+	dprintk("NFS: create per-net callback data; net=%x\n", net->ns.inum);
+
+	ret = svc_bind(serv, net);
+	if (ret < 0) {
+		printk(KERN_WARNING "NFS: bind callback service failed\n");
+		goto err;
+	}
+
+	ret = 0;
+	if (minorversion == 0)
+		ret = nfs4_callback_up_net(serv, net);
+	else if (xprt->ops->bc_setup)
+		set_bc_enabled(serv);
+	else
+		ret = -EPROTONOSUPPORT;
+
+	if (ret < 0) {
+		printk(KERN_ERR "NFS: callback service start failed\n");
+		goto err;
+	}
+	return 0;
+
+err:
+	nn->cb_users[minorversion]--;
+	dprintk("NFS: Couldn't create callback socket: err = %d; "
+			"net = %x\n", ret, net->ns.inum);
+	return ret;
+}
+
+static struct svc_serv *nfs_callback_create_svc(int minorversion)
+{
+	struct nfs_callback_data *cb_info = &nfs_callback_info[minorversion];
+	int (*threadfn)(void *data);
+	struct svc_serv *serv;
+
+	/*
+	 * Check whether we're already up and running.
+	 */
+	if (cb_info->serv)
+		return cb_info->serv;
+
+	/*
+	 * Sanity check: if there's no task,
+	 * we should be the first user ...
+	 */
+	if (cb_info->users)
+		printk(KERN_WARNING "nfs_callback_create_svc: no kthread, %d users??\n",
+			cb_info->users);
+
+	threadfn = nfs4_callback_svc;
+	serv = svc_create(&nfs4_callback_program, NFS4_CALLBACK_BUFSIZE,
+			  threadfn);
+	if (!serv) {
+		printk(KERN_ERR "nfs_callback_create_svc: create service failed\n");
+		return ERR_PTR(-ENOMEM);
+	}
+	cb_info->serv = serv;
+	dprintk("nfs_callback_create_svc: service created\n");
+	return serv;
+}
+
+/*
+ * Bring up the callback thread if it is not already up.
+ */
+int nfs_callback_up(u32 minorversion, struct rpc_xprt *xprt)
+{
+	struct svc_serv *serv;
+	struct nfs_callback_data *cb_info = &nfs_callback_info[minorversion];
+	int ret;
+	struct net *net = xprt->xprt_net;
+
+	mutex_lock(&nfs_callback_mutex);
+
+	serv = nfs_callback_create_svc(minorversion);
+	if (IS_ERR(serv)) {
+		ret = PTR_ERR(serv);
+		goto err_create;
+	}
+
+	ret = nfs_callback_up_net(minorversion, serv, net, xprt);
+	if (ret < 0)
+		goto err_net;
+
+	ret = nfs_callback_start_svc(minorversion, xprt, serv);
+	if (ret < 0)
+		goto err_start;
+
+	cb_info->users++;
+err_net:
+	if (!cb_info->users) {
+		xprt_svc_shutdown_bc(xprt);
+		svc_set_num_threads(cb_info->serv, 0, 0);
+		xprt_svc_destroy_nullify_bc(xprt, &cb_info->serv);
+	}
+err_create:
+	mutex_unlock(&nfs_callback_mutex);
+	return ret;
+
+err_start:
+	nfs_callback_down_net(minorversion, serv, net);
+	dprintk("NFS: Couldn't create server thread; err = %d\n", ret);
+	goto err_net;
+}
+
+/*
+ * Kill the callback thread if it's no longer being used.
+ */
+void nfs_callback_down(int minorversion, struct net *net, struct rpc_xprt *xprt)
+{
+	struct nfs_callback_data *cb_info = &nfs_callback_info[minorversion];
+	struct svc_serv *serv;
+
+	mutex_lock(&nfs_callback_mutex);
+	serv = cb_info->serv;
+	xprt_svc_shutdown_bc(xprt);
+	nfs_callback_down_net(minorversion, serv, net);
+	cb_info->users--;
+	if (cb_info->users == 0) {
+		svc_set_num_threads(serv, 0, 0);
+		dprintk("nfs_callback_down: service destroyed\n");
+		xprt_svc_destroy_nullify_bc(xprt, &cb_info->serv);
+	}
+	mutex_unlock(&nfs_callback_mutex);
+}
+
+/* Boolean check of RPC_AUTH_GSS principal */
+int
+check_gss_callback_principal(struct nfs_client *clp, struct svc_rqst *rqstp)
+{
+	char *p = rqstp->rq_cred.cr_principal;
+
+	if (rqstp->rq_authop->flavour != RPC_AUTH_GSS)
+		return 1;
+
+	/* No RPC_AUTH_GSS on NFSv4.1 back channel yet */
+	if (clp->cl_minorversion != 0)
+		return 0;
+	/*
+	 * It might just be a normal user principal, in which case
+	 * userspace won't bother to tell us the name at all.
+	 */
+	if (p == NULL)
+		return 0;
+
+	/*
+	 * Did we get the acceptor from userland during the SETCLIENID
+	 * negotiation?
+	 */
+	if (clp->cl_acceptor)
+		return !strcmp(p, clp->cl_acceptor);
+
+	/*
+	 * Otherwise try to verify it using the cl_hostname. Note that this
+	 * doesn't work if a non-canonical hostname was used in the devname.
+	 */
+
+	/* Expect a GSS_C_NT_HOSTBASED_NAME like "nfs@serverhostname" */
+
+	if (memcmp(p, "nfs@", 4) != 0)
+		return 0;
+	p += 4;
+	if (strcmp(p, clp->cl_hostname) != 0)
+		return 0;
+	return 1;
+}
+
+/*
+ * pg_authenticate method for nfsv4 callback threads.
+ *
+ * The authflavor has been negotiated, so an incorrect flavor is a server
+ * bug. Deny packets with incorrect authflavor.
+ *
+ * All other checking done after NFS decoding where the nfs_client can be
+ * found in nfs4_callback_compound
+ */
+static enum svc_auth_status nfs_callback_authenticate(struct svc_rqst *rqstp)
+{
+	rqstp->rq_auth_stat = rpc_autherr_badcred;
+
+	switch (rqstp->rq_authop->flavour) {
+	case RPC_AUTH_NULL:
+		if (rqstp->rq_proc != CB_NULL)
+			return SVC_DENIED;
+		break;
+	case RPC_AUTH_GSS:
+		/* No RPC_AUTH_GSS support yet in NFSv4.1 */
+		 if (svc_is_backchannel(rqstp))
+			return SVC_DENIED;
+	}
+
+	rqstp->rq_auth_stat = rpc_auth_ok;
+	return SVC_OK;
+}
+
+/*
+ * Define NFS4 callback program
+ */
+static const struct svc_version *nfs4_callback_version[] = {
+	[1] = &nfs4_callback_version1,
+	[4] = &nfs4_callback_version4,
+};
+
+static struct svc_program nfs4_callback_program = {
+	.pg_prog = NFS4_CALLBACK,			/* RPC service number */
+	.pg_nvers = ARRAY_SIZE(nfs4_callback_version),	/* Number of entries */
+	.pg_vers = nfs4_callback_version,		/* version table */
+	.pg_name = "NFSv4 callback",			/* service name */
+	.pg_class = "nfs",				/* authentication class */
+	.pg_authenticate = nfs_callback_authenticate,
+	.pg_init_request = svc_generic_init_request,
+	.pg_rpcbind_set	= svc_generic_rpcbind_set,
+};