From a5f7a5bb3b7f28ba7e4fa246775b29a0e5537255 Mon Sep 17 00:00:00 2001 From: Jinke Han Date: Tue, 8 Sep 2026 15:37:42 +0800 Subject: x86/kprobes: Fix crash when probing CS CALL instructions 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) Signed-off-by: Jinke Han Signed-off-by: Ingo Molnar Reviewed-by: Masami Hiramatsu (Google) Acked-by: Yafang Shao Acked-by: Borislav Petkov Cc: Peter Zijlstra Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000 --- net/ipv4/xfrm4_input.c | 229 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 229 insertions(+) create mode 100644 net/ipv4/xfrm4_input.c (limited to 'net/ipv4/xfrm4_input.c') diff --git a/net/ipv4/xfrm4_input.c b/net/ipv4/xfrm4_input.c new file mode 100644 index 000000000..f6f2a8ef3 --- /dev/null +++ b/net/ipv4/xfrm4_input.c @@ -0,0 +1,229 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * xfrm4_input.c + * + * Changes: + * YOSHIFUJI Hideaki @USAGI + * Split up af-specific portion + * Derek Atkins + * Add Encapsulation support + * + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +static int xfrm4_rcv_encap_finish2(struct net *net, struct sock *sk, + struct sk_buff *skb) +{ + return dst_input(skb); +} + +static inline int xfrm4_rcv_encap_finish(struct net *net, struct sock *sk, + struct sk_buff *skb) +{ + if (!skb_dst(skb)) { + const struct iphdr *iph = ip_hdr(skb); + + if (ip_route_input_noref(skb, iph->daddr, iph->saddr, + ip4h_dscp(iph), skb->dev)) + goto drop; + } + + if (xfrm_trans_queue(skb, xfrm4_rcv_encap_finish2)) + goto drop; + + return 0; +drop: + kfree_skb(skb); + return NET_RX_DROP; +} + +int xfrm4_transport_finish(struct sk_buff *skb, int async) +{ + struct xfrm_offload *xo = xfrm_offload(skb); + struct iphdr *iph = ip_hdr(skb); + struct net_device *dev = skb->dev; + + iph->protocol = XFRM_MODE_SKB_CB(skb)->protocol; + +#ifndef CONFIG_NETFILTER + if (!async) + return -iph->protocol; +#endif + + __skb_push(skb, -skb_network_offset(skb)); + iph->tot_len = htons(skb->len); + ip_send_check(iph); + + if (xo && (xo->flags & XFRM_GRO)) { + /* The full l2 header needs to be preserved so that re-injecting the packet at l2 + * works correctly in the presence of vlan tags. + */ + skb_mac_header_rebuild_full(skb, xo->orig_mac_len); + skb_reset_network_header(skb); + skb_reset_transport_header(skb); + return 0; + } + + NF_HOOK(NFPROTO_IPV4, NF_INET_PRE_ROUTING, + dev_net(dev), NULL, skb, dev, NULL, + xfrm4_rcv_encap_finish); + return 0; +} + +static int __xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb, bool pull) +{ + struct udp_sock *up = udp_sk(sk); + struct udphdr *uh; + struct iphdr *iph; + int iphlen, len; + __u8 *udpdata; + __be32 *udpdata32; + u16 encap_type; + + encap_type = READ_ONCE(up->encap_type); + /* if this is not encapsulated socket, then just return now */ + if (!encap_type) + return 1; + + /* If this is a paged skb, make sure we pull up + * whatever data we need to look at. */ + len = skb->len - sizeof(struct udphdr); + if (!pskb_may_pull(skb, sizeof(struct udphdr) + min(len, 8))) + return 1; + + /* Now we can get the pointers */ + uh = udp_hdr(skb); + udpdata = (__u8 *)uh + sizeof(struct udphdr); + udpdata32 = (__be32 *)udpdata; + + switch (encap_type) { + default: + case UDP_ENCAP_ESPINUDP: + /* Check if this is a keepalive packet. If so, eat it. */ + if (len == 1 && udpdata[0] == 0xff) { + return -EINVAL; + } else if (len > sizeof(struct ip_esp_hdr) && udpdata32[0] != 0) { + /* ESP Packet without Non-ESP header */ + len = sizeof(struct udphdr); + } else + /* Must be an IKE packet.. pass it through */ + return 1; + break; + } + + /* At this point we are sure that this is an ESPinUDP packet, + * so we need to remove 'len' bytes from the packet (the UDP + * header and optional ESP marker bytes) and then modify the + * protocol to ESP, and then call into the transform receiver. + */ + if (skb_unclone(skb, GFP_ATOMIC)) + return -EINVAL; + + /* Now we can update and verify the packet length... */ + iph = ip_hdr(skb); + iphlen = iph->ihl << 2; + iph->tot_len = htons(ntohs(iph->tot_len) - len); + if (skb->len < iphlen + len) { + /* packet is too small!?! */ + return -EINVAL; + } + + /* pull the data buffer up to the ESP header and set the + * transport header to point to ESP. Keep UDP on the stack + * for later. + */ + if (pull) { + __skb_pull(skb, len); + skb_reset_transport_header(skb); + } else { + skb_set_transport_header(skb, len); + } + + /* process ESP */ + return 0; +} + +/* If it's a keepalive packet, then just eat it. + * If it's an encapsulated packet, then pass it to the + * IPsec xfrm input. + * Returns 0 if skb passed to xfrm or was dropped. + * Returns >0 if skb should be passed to UDP. + * Returns <0 if skb should be resubmitted (-ret is protocol) + */ +int xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb) +{ + int ret; + + ret = __xfrm4_udp_encap_rcv(sk, skb, true); + if (!ret) + return xfrm4_rcv_encap(skb, IPPROTO_ESP, 0, + udp_sk(sk)->encap_type); + + if (ret < 0) { + kfree_skb(skb); + return 0; + } + + return ret; +} +EXPORT_SYMBOL(xfrm4_udp_encap_rcv); + +struct sk_buff *xfrm4_gro_udp_encap_rcv(struct sock *sk, struct list_head *head, + struct sk_buff *skb) +{ + int offset = skb_gro_offset(skb); + const struct net_offload *ops; + struct sk_buff *pp = NULL; + int len, dlen; + __u8 *udpdata; + __be32 *udpdata32; + + len = skb->len - offset; + dlen = offset + min(len, 8); + udpdata = skb_gro_header(skb, dlen, offset); + udpdata32 = (__be32 *)udpdata; + if (unlikely(!udpdata)) + return NULL; + + rcu_read_lock(); + ops = rcu_dereference(inet_offloads[IPPROTO_ESP]); + if (!ops || !ops->callbacks.gro_receive) + goto out; + + /* check if it is a keepalive or IKE packet */ + if (len <= sizeof(struct ip_esp_hdr) || udpdata32[0] == 0) + goto out; + + /* set the transport header to ESP */ + skb_set_transport_header(skb, offset); + + NAPI_GRO_CB(skb)->proto = IPPROTO_UDP; + + pp = call_gro_receive(ops->callbacks.gro_receive, head, skb); + rcu_read_unlock(); + + return pp; + +out: + rcu_read_unlock(); + NAPI_GRO_CB(skb)->same_flow = 0; + NAPI_GRO_CB(skb)->flush = 1; + + return NULL; +} +EXPORT_SYMBOL(xfrm4_gro_udp_encap_rcv); + +int xfrm4_rcv(struct sk_buff *skb) +{ + return xfrm4_rcv_spi(skb, ip_hdr(skb)->protocol, 0); +} +EXPORT_SYMBOL(xfrm4_rcv); -- cgit v1.3.1