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, 151 insertions, 0 deletions

diff --git a/lib/crypto/s390/sha3.h b/lib/crypto/s390/sha3.h
new file mode 100644
index 000000000..854714047
--- /dev/null
+++ b/lib/crypto/s390/sha3.h
@@ -0,0 +1,151 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * SHA-3 optimized using the CP Assist for Cryptographic Functions (CPACF)
+ *
+ * Copyright 2025 Google LLC
+ */
+#include <asm/cpacf.h>
+#include <linux/cpufeature.h>
+
+static __ro_after_init DEFINE_STATIC_KEY_FALSE(have_sha3);
+static __ro_after_init DEFINE_STATIC_KEY_FALSE(have_sha3_init_optim);
+
+static void sha3_absorb_blocks(struct sha3_state *state, const u8 *data,
+			       size_t nblocks, size_t block_size)
+{
+	if (static_branch_likely(&have_sha3)) {
+		/*
+		 * Note that KIMD assumes little-endian order of the state
+		 * words.  sha3_state already uses that order, though, so
+		 * there's no need for a byteswap.
+		 */
+		switch (block_size) {
+		case SHA3_224_BLOCK_SIZE:
+			cpacf_kimd(CPACF_KIMD_SHA3_224, state,
+				   data, nblocks * block_size);
+			return;
+		case SHA3_256_BLOCK_SIZE:
+			/*
+			 * This case handles both SHA3-256 and SHAKE256, since
+			 * they have the same block size.
+			 */
+			cpacf_kimd(CPACF_KIMD_SHA3_256, state,
+				   data, nblocks * block_size);
+			return;
+		case SHA3_384_BLOCK_SIZE:
+			cpacf_kimd(CPACF_KIMD_SHA3_384, state,
+				   data, nblocks * block_size);
+			return;
+		case SHA3_512_BLOCK_SIZE:
+			cpacf_kimd(CPACF_KIMD_SHA3_512, state,
+				   data, nblocks * block_size);
+			return;
+		}
+	}
+	sha3_absorb_blocks_generic(state, data, nblocks, block_size);
+}
+
+static void sha3_keccakf(struct sha3_state *state)
+{
+	if (static_branch_likely(&have_sha3)) {
+		/*
+		 * Passing zeroes into any of CPACF_KIMD_SHA3_* gives the plain
+		 * Keccak-f permutation, which is what we want here.  Use
+		 * SHA3-512 since it has the smallest block size.
+		 */
+		static const u8 zeroes[SHA3_512_BLOCK_SIZE];
+
+		cpacf_kimd(CPACF_KIMD_SHA3_512, state, zeroes, sizeof(zeroes));
+	} else {
+		sha3_keccakf_generic(state);
+	}
+}
+
+static inline bool s390_sha3(int func, const u8 *in, size_t in_len,
+			     u8 *out, size_t out_len)
+{
+	struct sha3_state state;
+
+	if (!static_branch_likely(&have_sha3))
+		return false;
+
+	if (static_branch_likely(&have_sha3_init_optim))
+		func |= CPACF_KLMD_NIP | CPACF_KLMD_DUFOP;
+	else
+		memset(&state, 0, sizeof(state));
+
+	cpacf_klmd(func, &state, in, in_len);
+
+	if (static_branch_likely(&have_sha3_init_optim))
+		kmsan_unpoison_memory(&state, out_len);
+
+	memcpy(out, &state, out_len);
+	memzero_explicit(&state, sizeof(state));
+	return true;
+}
+
+#define sha3_224_arch sha3_224_arch
+static bool sha3_224_arch(const u8 *in, size_t in_len,
+			  u8 out[SHA3_224_DIGEST_SIZE])
+{
+	return s390_sha3(CPACF_KLMD_SHA3_224, in, in_len,
+			 out, SHA3_224_DIGEST_SIZE);
+}
+
+#define sha3_256_arch sha3_256_arch
+static bool sha3_256_arch(const u8 *in, size_t in_len,
+			  u8 out[SHA3_256_DIGEST_SIZE])
+{
+	return s390_sha3(CPACF_KLMD_SHA3_256, in, in_len,
+			 out, SHA3_256_DIGEST_SIZE);
+}
+
+#define sha3_384_arch sha3_384_arch
+static bool sha3_384_arch(const u8 *in, size_t in_len,
+			  u8 out[SHA3_384_DIGEST_SIZE])
+{
+	return s390_sha3(CPACF_KLMD_SHA3_384, in, in_len,
+			 out, SHA3_384_DIGEST_SIZE);
+}
+
+#define sha3_512_arch sha3_512_arch
+static bool sha3_512_arch(const u8 *in, size_t in_len,
+			  u8 out[SHA3_512_DIGEST_SIZE])
+{
+	return s390_sha3(CPACF_KLMD_SHA3_512, in, in_len,
+			 out, SHA3_512_DIGEST_SIZE);
+}
+
+#define sha3_mod_init_arch sha3_mod_init_arch
+static void sha3_mod_init_arch(void)
+{
+	int num_present = 0;
+	int num_possible = 0;
+
+	if (!cpu_have_feature(S390_CPU_FEATURE_MSA))
+		return;
+	/*
+	 * Since all the SHA-3 functions are in Message-Security-Assist
+	 * Extension 6, just treat them as all or nothing.  This way we need
+	 * only one static_key.
+	 */
+#define QUERY(opcode, func) \
+	({ num_present += !!cpacf_query_func(opcode, func); num_possible++; })
+	QUERY(CPACF_KIMD, CPACF_KIMD_SHA3_224);
+	QUERY(CPACF_KIMD, CPACF_KIMD_SHA3_256);
+	QUERY(CPACF_KIMD, CPACF_KIMD_SHA3_384);
+	QUERY(CPACF_KIMD, CPACF_KIMD_SHA3_512);
+	QUERY(CPACF_KLMD, CPACF_KLMD_SHA3_224);
+	QUERY(CPACF_KLMD, CPACF_KLMD_SHA3_256);
+	QUERY(CPACF_KLMD, CPACF_KLMD_SHA3_384);
+	QUERY(CPACF_KLMD, CPACF_KLMD_SHA3_512);
+#undef QUERY
+
+	if (num_present == num_possible) {
+		static_branch_enable(&have_sha3);
+		if (test_facility(86))
+			static_branch_enable(&have_sha3_init_optim);
+	} else if (num_present != 0) {
+		pr_warn("Unsupported combination of SHA-3 facilities\n");
+	}
+}