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, 267 insertions, 0 deletions

diff --git a/net/packet/diag.c b/net/packet/diag.c
new file mode 100644
index 000000000..cee773f46
--- /dev/null
+++ b/net/packet/diag.c
@@ -0,0 +1,267 @@
+// SPDX-License-Identifier: GPL-2.0-only
+#include <linux/module.h>
+#include <linux/sock_diag.h>
+#include <linux/net.h>
+#include <linux/netdevice.h>
+#include <linux/packet_diag.h>
+#include <linux/percpu.h>
+#include <net/net_namespace.h>
+#include <net/sock.h>
+
+#include "internal.h"
+
+static int pdiag_put_info(const struct packet_sock *po, struct sk_buff *nlskb)
+{
+	struct packet_diag_info pinfo;
+
+	pinfo.pdi_index = po->ifindex;
+	pinfo.pdi_version = po->tp_version;
+	pinfo.pdi_reserve = po->tp_reserve;
+	pinfo.pdi_copy_thresh = READ_ONCE(po->copy_thresh);
+	pinfo.pdi_tstamp = READ_ONCE(po->tp_tstamp);
+
+	pinfo.pdi_flags = 0;
+	if (packet_sock_flag(po, PACKET_SOCK_RUNNING))
+		pinfo.pdi_flags |= PDI_RUNNING;
+	if (packet_sock_flag(po, PACKET_SOCK_AUXDATA))
+		pinfo.pdi_flags |= PDI_AUXDATA;
+	if (packet_sock_flag(po, PACKET_SOCK_ORIGDEV))
+		pinfo.pdi_flags |= PDI_ORIGDEV;
+	if (READ_ONCE(po->vnet_hdr_sz))
+		pinfo.pdi_flags |= PDI_VNETHDR;
+	if (packet_sock_flag(po, PACKET_SOCK_TP_LOSS))
+		pinfo.pdi_flags |= PDI_LOSS;
+
+	return nla_put(nlskb, PACKET_DIAG_INFO, sizeof(pinfo), &pinfo);
+}
+
+static int pdiag_put_mclist(const struct packet_sock *po, struct sk_buff *nlskb)
+{
+	struct nlattr *mca;
+	struct packet_mclist *ml;
+
+	mca = nla_nest_start_noflag(nlskb, PACKET_DIAG_MCLIST);
+	if (!mca)
+		return -EMSGSIZE;
+
+	rtnl_lock();
+	for (ml = po->mclist; ml; ml = ml->next) {
+		struct packet_diag_mclist *dml;
+
+		dml = nla_reserve_nohdr(nlskb, sizeof(*dml));
+		if (!dml) {
+			rtnl_unlock();
+			nla_nest_cancel(nlskb, mca);
+			return -EMSGSIZE;
+		}
+
+		dml->pdmc_index = ml->ifindex;
+		dml->pdmc_type = ml->type;
+		dml->pdmc_alen = ml->alen;
+		dml->pdmc_count = ml->count;
+		BUILD_BUG_ON(sizeof(dml->pdmc_addr) != sizeof(ml->addr));
+		memcpy(dml->pdmc_addr, ml->addr, sizeof(ml->addr));
+	}
+
+	rtnl_unlock();
+	nla_nest_end(nlskb, mca);
+
+	return 0;
+}
+
+static int pdiag_put_ring(struct packet_ring_buffer *ring, int ver, int nl_type,
+		struct sk_buff *nlskb)
+{
+	struct packet_diag_ring pdr;
+
+	if (!ring->pg_vec)
+		return 0;
+
+	pdr.pdr_block_size = ring->pg_vec_pages << PAGE_SHIFT;
+	pdr.pdr_block_nr = ring->pg_vec_len;
+	pdr.pdr_frame_size = ring->frame_size;
+	pdr.pdr_frame_nr = ring->frame_max + 1;
+
+	if (ver > TPACKET_V2) {
+		pdr.pdr_retire_tmo = ktime_to_ms(ring->prb_bdqc.interval_ktime);
+		pdr.pdr_sizeof_priv = ring->prb_bdqc.blk_sizeof_priv;
+		pdr.pdr_features = ring->prb_bdqc.feature_req_word;
+	} else {
+		pdr.pdr_retire_tmo = 0;
+		pdr.pdr_sizeof_priv = 0;
+		pdr.pdr_features = 0;
+	}
+
+	return nla_put(nlskb, nl_type, sizeof(pdr), &pdr);
+}
+
+static int pdiag_put_rings_cfg(struct packet_sock *po, struct sk_buff *skb)
+{
+	int ret;
+
+	mutex_lock(&po->pg_vec_lock);
+	ret = pdiag_put_ring(&po->rx_ring, po->tp_version,
+			PACKET_DIAG_RX_RING, skb);
+	if (!ret)
+		ret = pdiag_put_ring(&po->tx_ring, po->tp_version,
+				PACKET_DIAG_TX_RING, skb);
+	mutex_unlock(&po->pg_vec_lock);
+
+	return ret;
+}
+
+static int pdiag_put_fanout(struct packet_sock *po, struct sk_buff *nlskb)
+{
+	int ret = 0;
+
+	mutex_lock(&fanout_mutex);
+	if (po->fanout) {
+		u32 val;
+
+		val = (u32)po->fanout->id | ((u32)po->fanout->type << 16);
+		ret = nla_put_u32(nlskb, PACKET_DIAG_FANOUT, val);
+	}
+	mutex_unlock(&fanout_mutex);
+
+	return ret;
+}
+
+static int sk_diag_fill(struct sock *sk, struct sk_buff *skb,
+			struct packet_diag_req *req,
+			bool may_report_filterinfo,
+			struct user_namespace *user_ns,
+			u32 portid, u32 seq, u32 flags, u64 sk_ino)
+{
+	struct nlmsghdr *nlh;
+	struct packet_diag_msg *rp;
+	struct packet_sock *po = pkt_sk(sk);
+
+	nlh = nlmsg_put(skb, portid, seq, SOCK_DIAG_BY_FAMILY, sizeof(*rp), flags);
+	if (!nlh)
+		return -EMSGSIZE;
+
+	rp = nlmsg_data(nlh);
+	rp->pdiag_family = AF_PACKET;
+	rp->pdiag_type = sk->sk_type;
+	rp->pdiag_num = ntohs(READ_ONCE(po->num));
+	rp->pdiag_ino = sk_ino;
+	sock_diag_save_cookie(sk, rp->pdiag_cookie);
+
+	if ((req->pdiag_show & PACKET_SHOW_INFO) &&
+			pdiag_put_info(po, skb))
+		goto out_nlmsg_trim;
+
+	if ((req->pdiag_show & PACKET_SHOW_INFO) &&
+	    nla_put_u32(skb, PACKET_DIAG_UID,
+			from_kuid_munged(user_ns, sk_uid(sk))))
+		goto out_nlmsg_trim;
+
+	if ((req->pdiag_show & PACKET_SHOW_MCLIST) &&
+			pdiag_put_mclist(po, skb))
+		goto out_nlmsg_trim;
+
+	if ((req->pdiag_show & PACKET_SHOW_RING_CFG) &&
+			pdiag_put_rings_cfg(po, skb))
+		goto out_nlmsg_trim;
+
+	if ((req->pdiag_show & PACKET_SHOW_FANOUT) &&
+			pdiag_put_fanout(po, skb))
+		goto out_nlmsg_trim;
+
+	if ((req->pdiag_show & PACKET_SHOW_MEMINFO) &&
+	    sock_diag_put_meminfo(sk, skb, PACKET_DIAG_MEMINFO))
+		goto out_nlmsg_trim;
+
+	if ((req->pdiag_show & PACKET_SHOW_FILTER) &&
+	    sock_diag_put_filterinfo(may_report_filterinfo, sk, skb,
+				     PACKET_DIAG_FILTER))
+		goto out_nlmsg_trim;
+
+	nlmsg_end(skb, nlh);
+	return 0;
+
+out_nlmsg_trim:
+	nlmsg_cancel(skb, nlh);
+	return -EMSGSIZE;
+}
+
+static int packet_diag_dump(struct sk_buff *skb, struct netlink_callback *cb)
+{
+	int num = 0, s_num = cb->args[0];
+	struct packet_diag_req *req;
+	struct net *net;
+	struct sock *sk;
+	bool may_report_filterinfo;
+
+	net = sock_net(skb->sk);
+	req = nlmsg_data(cb->nlh);
+	may_report_filterinfo = netlink_net_capable(cb->skb, CAP_NET_ADMIN);
+
+	mutex_lock(&net->packet.sklist_lock);
+	sk_for_each(sk, &net->packet.sklist) {
+		if (!net_eq(sock_net(sk), net))
+			continue;
+		if (num < s_num)
+			goto next;
+
+		if (sk_diag_fill(sk, skb, req,
+				 may_report_filterinfo,
+				 sk_user_ns(NETLINK_CB(cb->skb).sk),
+				 NETLINK_CB(cb->skb).portid,
+				 cb->nlh->nlmsg_seq, NLM_F_MULTI,
+				 sock_i_ino(sk)) < 0)
+			goto done;
+next:
+		num++;
+	}
+done:
+	mutex_unlock(&net->packet.sklist_lock);
+	cb->args[0] = num;
+
+	return skb->len;
+}
+
+static int packet_diag_handler_dump(struct sk_buff *skb, struct nlmsghdr *h)
+{
+	int hdrlen = sizeof(struct packet_diag_req);
+	struct net *net = sock_net(skb->sk);
+	struct packet_diag_req *req;
+
+	if (nlmsg_len(h) < hdrlen)
+		return -EINVAL;
+
+	req = nlmsg_data(h);
+	/* Make it possible to support protocol filtering later */
+	if (req->sdiag_protocol)
+		return -EINVAL;
+
+	if (h->nlmsg_flags & NLM_F_DUMP) {
+		struct netlink_dump_control c = {
+			.dump = packet_diag_dump,
+		};
+		return netlink_dump_start(net->diag_nlsk, skb, h, &c);
+	} else
+		return -EOPNOTSUPP;
+}
+
+static const struct sock_diag_handler packet_diag_handler = {
+	.owner = THIS_MODULE,
+	.family = AF_PACKET,
+	.dump = packet_diag_handler_dump,
+};
+
+static int __init packet_diag_init(void)
+{
+	return sock_diag_register(&packet_diag_handler);
+}
+
+static void __exit packet_diag_exit(void)
+{
+	sock_diag_unregister(&packet_diag_handler);
+}
+
+module_init(packet_diag_init);
+module_exit(packet_diag_exit);
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("PACKET socket monitoring via SOCK_DIAG");
+MODULE_ALIAS_NET_PF_PROTO_TYPE(PF_NETLINK, NETLINK_SOCK_DIAG, 17 /* AF_PACKET */);