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, 250 insertions, 0 deletions

diff --git a/net/sched/act_simple.c b/net/sched/act_simple.c
new file mode 100644
index 000000000..8e69a919b
--- /dev/null
+++ b/net/sched/act_simple.c
@@ -0,0 +1,250 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * net/sched/act_simple.c	Simple example of an action
+ *
+ * Authors:	Jamal Hadi Salim (2005-8)
+ */
+
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/skbuff.h>
+#include <linux/rtnetlink.h>
+#include <net/netlink.h>
+#include <net/pkt_sched.h>
+#include <net/pkt_cls.h>
+#include <net/tc_wrapper.h>
+
+#include <linux/tc_act/tc_defact.h>
+#include <net/tc_act/tc_defact.h>
+
+static struct tc_action_ops act_simp_ops;
+
+#define SIMP_MAX_DATA	32
+TC_INDIRECT_SCOPE int tcf_simp_act(struct sk_buff *skb,
+				   const struct tc_action *a,
+				   struct tcf_result *res)
+{
+	struct tcf_defact *d = to_defact(a);
+
+	spin_lock(&d->tcf_lock);
+	tcf_lastuse_update(&d->tcf_tm);
+	bstats_update(&d->tcf_bstats, skb);
+
+	/* print policy string followed by _ then packet count
+	 * Example if this was the 3rd packet and the string was "hello"
+	 * then it would look like "hello_3" (without quotes)
+	 */
+	pr_info("simple: %s_%llu\n",
+		(char *)d->tcfd_defdata,
+		u64_stats_read(&d->tcf_bstats.packets));
+	spin_unlock(&d->tcf_lock);
+	return d->tcf_action;
+}
+
+static void tcf_simp_release(struct tc_action *a)
+{
+	struct tcf_defact *d = to_defact(a);
+	kfree(d->tcfd_defdata);
+}
+
+static int alloc_defdata(struct tcf_defact *d, const struct nlattr *defdata)
+{
+	d->tcfd_defdata = kzalloc(SIMP_MAX_DATA, GFP_KERNEL);
+	if (unlikely(!d->tcfd_defdata))
+		return -ENOMEM;
+	nla_strscpy(d->tcfd_defdata, defdata, SIMP_MAX_DATA);
+	return 0;
+}
+
+static int reset_policy(struct tc_action *a, const struct nlattr *defdata,
+			struct tc_defact *p, struct tcf_proto *tp,
+			struct netlink_ext_ack *extack)
+{
+	struct tcf_chain *goto_ch = NULL;
+	struct tcf_defact *d;
+	int err;
+
+	err = tcf_action_check_ctrlact(p->action, tp, &goto_ch, extack);
+	if (err < 0)
+		return err;
+	d = to_defact(a);
+	spin_lock_bh(&d->tcf_lock);
+	goto_ch = tcf_action_set_ctrlact(a, p->action, goto_ch);
+	nla_strscpy(d->tcfd_defdata, defdata, SIMP_MAX_DATA);
+	spin_unlock_bh(&d->tcf_lock);
+	if (goto_ch)
+		tcf_chain_put_by_act(goto_ch);
+	return 0;
+}
+
+static const struct nla_policy simple_policy[TCA_DEF_MAX + 1] = {
+	[TCA_DEF_PARMS]	= { .len = sizeof(struct tc_defact) },
+	[TCA_DEF_DATA]	= { .type = NLA_STRING, .len = SIMP_MAX_DATA },
+};
+
+static int tcf_simp_init(struct net *net, struct nlattr *nla,
+			 struct nlattr *est, struct tc_action **a,
+			 struct tcf_proto *tp, u32 flags,
+			 struct netlink_ext_ack *extack)
+{
+	struct tc_action_net *tn = net_generic(net, act_simp_ops.net_id);
+	bool bind = flags & TCA_ACT_FLAGS_BIND;
+	struct nlattr *tb[TCA_DEF_MAX + 1];
+	struct tcf_chain *goto_ch = NULL;
+	struct tc_defact *parm;
+	struct tcf_defact *d;
+	bool exists = false;
+	int ret = 0, err;
+	u32 index;
+
+	if (nla == NULL)
+		return -EINVAL;
+
+	err = nla_parse_nested_deprecated(tb, TCA_DEF_MAX, nla, simple_policy,
+					  NULL);
+	if (err < 0)
+		return err;
+
+	if (tb[TCA_DEF_PARMS] == NULL)
+		return -EINVAL;
+
+	parm = nla_data(tb[TCA_DEF_PARMS]);
+	index = parm->index;
+	err = tcf_idr_check_alloc(tn, &index, a, bind);
+	if (err < 0)
+		return err;
+	exists = err;
+	if (exists && bind)
+		return ACT_P_BOUND;
+
+	if (tb[TCA_DEF_DATA] == NULL) {
+		if (exists)
+			tcf_idr_release(*a, bind);
+		else
+			tcf_idr_cleanup(tn, index);
+		return -EINVAL;
+	}
+
+	if (!exists) {
+		ret = tcf_idr_create(tn, index, est, a,
+				     &act_simp_ops, bind, false, flags);
+		if (ret) {
+			tcf_idr_cleanup(tn, index);
+			return ret;
+		}
+
+		d = to_defact(*a);
+		err = tcf_action_check_ctrlact(parm->action, tp, &goto_ch,
+					       extack);
+		if (err < 0)
+			goto release_idr;
+
+		err = alloc_defdata(d, tb[TCA_DEF_DATA]);
+		if (err < 0)
+			goto put_chain;
+
+		tcf_action_set_ctrlact(*a, parm->action, goto_ch);
+		ret = ACT_P_CREATED;
+	} else {
+		if (!(flags & TCA_ACT_FLAGS_REPLACE)) {
+			err = -EEXIST;
+			goto release_idr;
+		}
+
+		err = reset_policy(*a, tb[TCA_DEF_DATA], parm, tp, extack);
+		if (err)
+			goto release_idr;
+	}
+
+	return ret;
+put_chain:
+	if (goto_ch)
+		tcf_chain_put_by_act(goto_ch);
+release_idr:
+	tcf_idr_release(*a, bind);
+	return err;
+}
+
+static int tcf_simp_dump(struct sk_buff *skb, struct tc_action *a,
+			 int bind, int ref)
+{
+	unsigned char *b = skb_tail_pointer(skb);
+	struct tcf_defact *d = to_defact(a);
+	struct tc_defact opt = {
+		.index   = d->tcf_index,
+		.refcnt  = refcount_read(&d->tcf_refcnt) - ref,
+		.bindcnt = atomic_read(&d->tcf_bindcnt) - bind,
+	};
+	struct tcf_t t;
+
+	spin_lock_bh(&d->tcf_lock);
+	opt.action = d->tcf_action;
+	if (nla_put(skb, TCA_DEF_PARMS, sizeof(opt), &opt) ||
+	    nla_put_string(skb, TCA_DEF_DATA, d->tcfd_defdata))
+		goto nla_put_failure;
+
+	tcf_tm_dump(&t, &d->tcf_tm);
+	if (nla_put_64bit(skb, TCA_DEF_TM, sizeof(t), &t, TCA_DEF_PAD))
+		goto nla_put_failure;
+	spin_unlock_bh(&d->tcf_lock);
+
+	return skb->len;
+
+nla_put_failure:
+	spin_unlock_bh(&d->tcf_lock);
+	nlmsg_trim(skb, b);
+	return -1;
+}
+
+static struct tc_action_ops act_simp_ops = {
+	.kind		=	"simple",
+	.id		=	TCA_ID_SIMP,
+	.owner		=	THIS_MODULE,
+	.act		=	tcf_simp_act,
+	.dump		=	tcf_simp_dump,
+	.cleanup	=	tcf_simp_release,
+	.init		=	tcf_simp_init,
+	.size		=	sizeof(struct tcf_defact),
+};
+MODULE_ALIAS_NET_ACT("simple");
+
+static __net_init int simp_init_net(struct net *net)
+{
+	struct tc_action_net *tn = net_generic(net, act_simp_ops.net_id);
+
+	return tc_action_net_init(net, tn, &act_simp_ops);
+}
+
+static void __net_exit simp_exit_net(struct list_head *net_list)
+{
+	tc_action_net_exit(net_list, act_simp_ops.net_id);
+}
+
+static struct pernet_operations simp_net_ops = {
+	.init = simp_init_net,
+	.exit_batch = simp_exit_net,
+	.id   = &act_simp_ops.net_id,
+	.size = sizeof(struct tc_action_net),
+};
+
+MODULE_AUTHOR("Jamal Hadi Salim(2005)");
+MODULE_DESCRIPTION("Simple example action");
+MODULE_LICENSE("GPL");
+
+static int __init simp_init_module(void)
+{
+	int ret = tcf_register_action(&act_simp_ops, &simp_net_ops);
+	if (!ret)
+		pr_info("Simple TC action Loaded\n");
+	return ret;
+}
+
+static void __exit simp_cleanup_module(void)
+{
+	tcf_unregister_action(&act_simp_ops, &simp_net_ops);
+}
+
+module_init(simp_init_module);
+module_exit(simp_cleanup_module);