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, 110 insertions, 0 deletions

diff --git a/net/ipv4/tcp_plb.c b/net/ipv4/tcp_plb.c
new file mode 100644
index 000000000..bcc2f0add
--- /dev/null
+++ b/net/ipv4/tcp_plb.c
@@ -0,0 +1,110 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Protective Load Balancing (PLB)
+ *
+ * PLB was designed to reduce link load imbalance across datacenter
+ * switches. PLB is a host-based optimization; it leverages congestion
+ * signals from the transport layer to randomly change the path of the
+ * connection experiencing sustained congestion. PLB prefers to repath
+ * after idle periods to minimize packet reordering. It repaths by
+ * changing the IPv6 Flow Label on the packets of a connection, which
+ * datacenter switches include as part of ECMP/WCMP hashing.
+ *
+ * PLB is described in detail in:
+ *
+ *	Mubashir Adnan Qureshi, Yuchung Cheng, Qianwen Yin, Qiaobin Fu,
+ *	Gautam Kumar, Masoud Moshref, Junhua Yan, Van Jacobson,
+ *	David Wetherall,Abdul Kabbani:
+ *	"PLB: Congestion Signals are Simple and Effective for
+ *	 Network Load Balancing"
+ *	In ACM SIGCOMM 2022, Amsterdam Netherlands.
+ *
+ */
+
+#include <net/tcp.h>
+
+/* Called once per round-trip to update PLB state for a connection. */
+void tcp_plb_update_state(const struct sock *sk, struct tcp_plb_state *plb,
+			  const int cong_ratio)
+{
+	struct net *net = sock_net(sk);
+
+	if (!READ_ONCE(net->ipv4.sysctl_tcp_plb_enabled))
+		return;
+
+	if (cong_ratio >= 0) {
+		if (cong_ratio < READ_ONCE(net->ipv4.sysctl_tcp_plb_cong_thresh))
+			plb->consec_cong_rounds = 0;
+		else if (plb->consec_cong_rounds <
+			 READ_ONCE(net->ipv4.sysctl_tcp_plb_rehash_rounds))
+			plb->consec_cong_rounds++;
+	}
+}
+EXPORT_SYMBOL_GPL(tcp_plb_update_state);
+
+/* Check whether recent congestion has been persistent enough to warrant
+ * a load balancing decision that switches the connection to another path.
+ */
+void tcp_plb_check_rehash(struct sock *sk, struct tcp_plb_state *plb)
+{
+	struct net *net = sock_net(sk);
+	u32 max_suspend;
+	bool forced_rehash = false, idle_rehash = false;
+
+	if (!READ_ONCE(net->ipv4.sysctl_tcp_plb_enabled))
+		return;
+
+	forced_rehash = plb->consec_cong_rounds >=
+			READ_ONCE(net->ipv4.sysctl_tcp_plb_rehash_rounds);
+	/* If sender goes idle then we check whether to rehash. */
+	idle_rehash = READ_ONCE(net->ipv4.sysctl_tcp_plb_idle_rehash_rounds) &&
+		      !tcp_sk(sk)->packets_out &&
+		      plb->consec_cong_rounds >=
+		      READ_ONCE(net->ipv4.sysctl_tcp_plb_idle_rehash_rounds);
+
+	if (!forced_rehash && !idle_rehash)
+		return;
+
+	/* Note that tcp_jiffies32 can wrap; we detect wraps by checking for
+	 * cases where the max suspension end is before the actual suspension
+	 * end. We clear pause_until to 0 to indicate there is no recent
+	 * RTO event that constrains PLB rehashing.
+	 */
+	max_suspend = 2 * READ_ONCE(net->ipv4.sysctl_tcp_plb_suspend_rto_sec) * HZ;
+	if (plb->pause_until &&
+	    (!before(tcp_jiffies32, plb->pause_until) ||
+	     before(tcp_jiffies32 + max_suspend, plb->pause_until)))
+		plb->pause_until = 0;
+
+	if (plb->pause_until)
+		return;
+
+	__sk_rethink_txhash_reset_dst(sk);
+	plb->consec_cong_rounds = 0;
+	WRITE_ONCE(tcp_sk(sk)->plb_rehash, tcp_sk(sk)->plb_rehash + 1);
+	NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPPLBREHASH);
+}
+EXPORT_SYMBOL_GPL(tcp_plb_check_rehash);
+
+/* Upon RTO, disallow load balancing for a while, to avoid having load
+ * balancing decisions switch traffic to a black-holed path that was
+ * previously avoided with a sk_rethink_txhash() call at RTO time.
+ */
+void tcp_plb_update_state_upon_rto(struct sock *sk, struct tcp_plb_state *plb)
+{
+	struct net *net = sock_net(sk);
+	u32 pause;
+
+	if (!READ_ONCE(net->ipv4.sysctl_tcp_plb_enabled))
+		return;
+
+	pause = READ_ONCE(net->ipv4.sysctl_tcp_plb_suspend_rto_sec) * HZ;
+	pause += get_random_u32_below(pause);
+	plb->pause_until = tcp_jiffies32 + pause;
+
+	/* Reset PLB state upon RTO, since an RTO causes a sk_rethink_txhash() call
+	 * that may switch this connection to a path with completely different
+	 * congestion characteristics.
+	 */
+	plb->consec_cong_rounds = 0;
+}
+EXPORT_SYMBOL_GPL(tcp_plb_update_state_upon_rto);