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, 85 insertions, 0 deletions

diff --git a/lib/crypto/x86/aes.h b/lib/crypto/x86/aes.h
new file mode 100644
index 000000000..b047dee94
--- /dev/null
+++ b/lib/crypto/x86/aes.h
@@ -0,0 +1,85 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * AES block cipher using AES-NI instructions
+ *
+ * Copyright 2026 Google LLC
+ */
+
+#include <asm/fpu/api.h>
+
+static __ro_after_init DEFINE_STATIC_KEY_FALSE(have_aes);
+
+void aes128_expandkey_aesni(u32 rndkeys[], u32 *inv_rndkeys,
+			    const u8 in_key[AES_KEYSIZE_128]);
+void aes256_expandkey_aesni(u32 rndkeys[], u32 *inv_rndkeys,
+			    const u8 in_key[AES_KEYSIZE_256]);
+void aes_encrypt_aesni(const u32 rndkeys[], int nrounds,
+		       u8 out[AES_BLOCK_SIZE], const u8 in[AES_BLOCK_SIZE]);
+void aes_decrypt_aesni(const u32 inv_rndkeys[], int nrounds,
+		       u8 out[AES_BLOCK_SIZE], const u8 in[AES_BLOCK_SIZE]);
+
+/*
+ * Expand an AES key using AES-NI if supported and usable or generic code
+ * otherwise.  The expanded key format is compatible between the two cases.  The
+ * outputs are @k->rndkeys (required) and @inv_k->inv_rndkeys (optional).
+ *
+ * We could just always use the generic key expansion code.  AES key expansion
+ * is usually less performance-critical than AES en/decryption.  However,
+ * there's still *some* value in speed here, as well as in non-key-dependent
+ * execution time which AES-NI provides.  So, do use AES-NI to expand AES-128
+ * and AES-256 keys.  (Don't bother with AES-192, as it's almost never used.)
+ */
+static void aes_preparekey_arch(union aes_enckey_arch *k,
+				union aes_invkey_arch *inv_k,
+				const u8 *in_key, int key_len, int nrounds)
+{
+	u32 *rndkeys = k->rndkeys;
+	u32 *inv_rndkeys = inv_k ? inv_k->inv_rndkeys : NULL;
+
+	if (static_branch_likely(&have_aes) && key_len != AES_KEYSIZE_192 &&
+	    irq_fpu_usable()) {
+		kernel_fpu_begin();
+		if (key_len == AES_KEYSIZE_128)
+			aes128_expandkey_aesni(rndkeys, inv_rndkeys, in_key);
+		else
+			aes256_expandkey_aesni(rndkeys, inv_rndkeys, in_key);
+		kernel_fpu_end();
+	} else {
+		aes_expandkey_generic(rndkeys, inv_rndkeys, in_key, key_len);
+	}
+}
+
+static void aes_encrypt_arch(const struct aes_enckey *key,
+			     u8 out[AES_BLOCK_SIZE],
+			     const u8 in[AES_BLOCK_SIZE])
+{
+	if (static_branch_likely(&have_aes) && irq_fpu_usable()) {
+		kernel_fpu_begin();
+		aes_encrypt_aesni(key->k.rndkeys, key->nrounds, out, in);
+		kernel_fpu_end();
+	} else {
+		aes_encrypt_generic(key->k.rndkeys, key->nrounds, out, in);
+	}
+}
+
+static void aes_decrypt_arch(const struct aes_key *key,
+			     u8 out[AES_BLOCK_SIZE],
+			     const u8 in[AES_BLOCK_SIZE])
+{
+	if (static_branch_likely(&have_aes) && irq_fpu_usable()) {
+		kernel_fpu_begin();
+		aes_decrypt_aesni(key->inv_k.inv_rndkeys, key->nrounds,
+				  out, in);
+		kernel_fpu_end();
+	} else {
+		aes_decrypt_generic(key->inv_k.inv_rndkeys, key->nrounds,
+				    out, in);
+	}
+}
+
+#define aes_mod_init_arch aes_mod_init_arch
+static void aes_mod_init_arch(void)
+{
+	if (boot_cpu_has(X86_FEATURE_AES))
+		static_branch_enable(&have_aes);
+}