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, 327 insertions, 0 deletions

diff --git a/lib/tests/test_fprobe.c b/lib/tests/test_fprobe.c
new file mode 100644
index 000000000..108c7aa33
--- /dev/null
+++ b/lib/tests/test_fprobe.c
@@ -0,0 +1,327 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * test_fprobe.c - simple sanity test for fprobe
+ */
+
+#include <linux/kernel.h>
+#include <linux/fprobe.h>
+#include <linux/random.h>
+#include <kunit/test.h>
+
+#define div_factor 3
+
+static struct kunit *current_test;
+
+static u32 rand1, entry_only_val, entry_val, exit_val;
+static u32 entry_only_count, entry_count, exit_count;
+
+/* Use indirect calls to avoid inlining the target functions */
+static u32 (*target)(u32 value);
+static u32 (*target2)(u32 value);
+static unsigned long target_ip;
+static unsigned long target2_ip;
+static int entry_return_value;
+
+static noinline u32 fprobe_selftest_target(u32 value)
+{
+	return (value / div_factor);
+}
+
+static noinline u32 fprobe_selftest_target2(u32 value)
+{
+	return (value / div_factor) + 1;
+}
+
+static notrace int fp_entry_handler(struct fprobe *fp, unsigned long ip,
+				    unsigned long ret_ip,
+				    struct ftrace_regs *fregs, void *data)
+{
+	KUNIT_EXPECT_FALSE(current_test, preemptible());
+	/* This can be called on the fprobe_selftest_target and the fprobe_selftest_target2 */
+	if (ip != target_ip)
+		KUNIT_EXPECT_EQ(current_test, ip, target2_ip);
+	entry_val = (rand1 / div_factor);
+	if (fp->entry_data_size) {
+		KUNIT_EXPECT_NOT_NULL(current_test, data);
+		if (data)
+			*(u32 *)data = entry_val;
+	} else
+		KUNIT_EXPECT_NULL(current_test, data);
+
+	return entry_return_value;
+}
+
+static notrace void fp_exit_handler(struct fprobe *fp, unsigned long ip,
+				    unsigned long ret_ip,
+				    struct ftrace_regs *fregs, void *data)
+{
+	unsigned long ret = ftrace_regs_get_return_value(fregs);
+
+	KUNIT_EXPECT_FALSE(current_test, preemptible());
+	if (ip != target_ip) {
+		KUNIT_EXPECT_EQ(current_test, ip, target2_ip);
+		KUNIT_EXPECT_EQ(current_test, ret, (rand1 / div_factor) + 1);
+	} else
+		KUNIT_EXPECT_EQ(current_test, ret, (rand1 / div_factor));
+	KUNIT_EXPECT_EQ(current_test, entry_val, (rand1 / div_factor));
+	exit_val = entry_val + div_factor;
+	if (fp->entry_data_size) {
+		KUNIT_EXPECT_NOT_NULL(current_test, data);
+		if (data)
+			KUNIT_EXPECT_EQ(current_test, *(u32 *)data, entry_val);
+	} else
+		KUNIT_EXPECT_NULL(current_test, data);
+}
+
+/* Test entry only (no rethook) */
+static void test_fprobe_entry(struct kunit *test)
+{
+	struct fprobe fp_entry = {
+		.entry_handler = fp_entry_handler,
+	};
+
+	current_test = test;
+
+	/* Before register, unregister should be failed. */
+	KUNIT_EXPECT_NE(test, 0, unregister_fprobe(&fp_entry));
+	KUNIT_EXPECT_EQ(test, 0, register_fprobe(&fp_entry, "fprobe_selftest_target*", NULL));
+
+	entry_val = 0;
+	exit_val = 0;
+	target(rand1);
+	KUNIT_EXPECT_NE(test, 0, entry_val);
+	KUNIT_EXPECT_EQ(test, 0, exit_val);
+
+	entry_val = 0;
+	exit_val = 0;
+	target2(rand1);
+	KUNIT_EXPECT_NE(test, 0, entry_val);
+	KUNIT_EXPECT_EQ(test, 0, exit_val);
+
+	KUNIT_EXPECT_EQ(test, 0, unregister_fprobe(&fp_entry));
+}
+
+static void test_fprobe(struct kunit *test)
+{
+	struct fprobe fp = {
+		.entry_handler = fp_entry_handler,
+		.exit_handler = fp_exit_handler,
+	};
+
+	current_test = test;
+	KUNIT_EXPECT_EQ(test, 0, register_fprobe(&fp, "fprobe_selftest_target*", NULL));
+
+	entry_val = 0;
+	exit_val = 0;
+	target(rand1);
+	KUNIT_EXPECT_NE(test, 0, entry_val);
+	KUNIT_EXPECT_EQ(test, entry_val + div_factor, exit_val);
+
+	entry_val = 0;
+	exit_val = 0;
+	target2(rand1);
+	KUNIT_EXPECT_NE(test, 0, entry_val);
+	KUNIT_EXPECT_EQ(test, entry_val + div_factor, exit_val);
+
+	KUNIT_EXPECT_EQ(test, 0, unregister_fprobe(&fp));
+}
+
+static void test_fprobe_syms(struct kunit *test)
+{
+	static const char *syms[] = {"fprobe_selftest_target", "fprobe_selftest_target2"};
+	struct fprobe fp = {
+		.entry_handler = fp_entry_handler,
+		.exit_handler = fp_exit_handler,
+	};
+
+	current_test = test;
+	KUNIT_EXPECT_EQ(test, 0, register_fprobe_syms(&fp, syms, 2));
+
+	entry_val = 0;
+	exit_val = 0;
+	target(rand1);
+	KUNIT_EXPECT_NE(test, 0, entry_val);
+	KUNIT_EXPECT_EQ(test, entry_val + div_factor, exit_val);
+
+	entry_val = 0;
+	exit_val = 0;
+	target2(rand1);
+	KUNIT_EXPECT_NE(test, 0, entry_val);
+	KUNIT_EXPECT_EQ(test, entry_val + div_factor, exit_val);
+
+	KUNIT_EXPECT_EQ(test, 0, unregister_fprobe(&fp));
+}
+
+/* Test private entry_data */
+static void test_fprobe_data(struct kunit *test)
+{
+	struct fprobe fp = {
+		.entry_handler = fp_entry_handler,
+		.exit_handler = fp_exit_handler,
+		.entry_data_size = sizeof(u32),
+	};
+
+	current_test = test;
+	KUNIT_EXPECT_EQ(test, 0, register_fprobe(&fp, "fprobe_selftest_target", NULL));
+
+	target(rand1);
+
+	KUNIT_EXPECT_EQ(test, 0, unregister_fprobe(&fp));
+}
+
+static void test_fprobe_skip(struct kunit *test)
+{
+	struct fprobe fp = {
+		.entry_handler = fp_entry_handler,
+		.exit_handler = fp_exit_handler,
+	};
+
+	current_test = test;
+	KUNIT_EXPECT_EQ(test, 0, register_fprobe(&fp, "fprobe_selftest_target", NULL));
+
+	entry_return_value = 1;
+	entry_val = 0;
+	exit_val = 0;
+	target(rand1);
+	KUNIT_EXPECT_NE(test, 0, entry_val);
+	KUNIT_EXPECT_EQ(test, 0, exit_val);
+	KUNIT_EXPECT_EQ(test, 0, fp.nmissed);
+	entry_return_value = 0;
+
+	KUNIT_EXPECT_EQ(test, 0, unregister_fprobe(&fp));
+}
+
+/* Handler for fprobe entry only case */
+static notrace int entry_only_handler(struct fprobe *fp, unsigned long ip,
+				      unsigned long ret_ip,
+				      struct ftrace_regs *fregs, void *data)
+{
+	KUNIT_EXPECT_FALSE(current_test, preemptible());
+	KUNIT_EXPECT_EQ(current_test, ip, target_ip);
+
+	entry_only_count++;
+	entry_only_val = (rand1 / div_factor);
+
+	return 0;
+}
+
+static notrace int fprobe_entry_multi_handler(struct fprobe *fp, unsigned long ip,
+					      unsigned long ret_ip,
+					      struct ftrace_regs *fregs,
+					      void *data)
+{
+	KUNIT_EXPECT_FALSE(current_test, preemptible());
+	KUNIT_EXPECT_EQ(current_test, ip, target_ip);
+
+	entry_count++;
+	entry_val = (rand1 / div_factor);
+
+	return 0;
+}
+
+static notrace void fprobe_exit_multi_handler(struct fprobe *fp, unsigned long ip,
+					      unsigned long ret_ip,
+					      struct ftrace_regs *fregs,
+					      void *data)
+{
+	unsigned long ret = ftrace_regs_get_return_value(fregs);
+
+	KUNIT_EXPECT_FALSE(current_test, preemptible());
+	KUNIT_EXPECT_EQ(current_test, ip, target_ip);
+	KUNIT_EXPECT_EQ(current_test, ret, (rand1 / div_factor));
+
+	exit_count++;
+	exit_val = ret;
+}
+
+static void check_fprobe_multi(struct kunit *test)
+{
+	entry_only_count = entry_count = exit_count = 0;
+	entry_only_val = entry_val = exit_val = 0;
+
+	target(rand1);
+
+	/* Verify all handlers were called */
+	KUNIT_EXPECT_EQ(test, 1, entry_only_count);
+	KUNIT_EXPECT_EQ(test, 1, entry_count);
+	KUNIT_EXPECT_EQ(test, 1, exit_count);
+
+	/* Verify values are correct */
+	KUNIT_EXPECT_EQ(test, (rand1 / div_factor), entry_only_val);
+	KUNIT_EXPECT_EQ(test, (rand1 / div_factor), entry_val);
+	KUNIT_EXPECT_EQ(test, (rand1 / div_factor), exit_val);
+}
+
+/* Test multiple fprobes hooking the same target function */
+static void test_fprobe_multi(struct kunit *test)
+{
+	struct fprobe fp1 = {
+		.entry_handler = fprobe_entry_multi_handler,
+		.exit_handler = fprobe_exit_multi_handler,
+	};
+	struct fprobe fp2 = {
+		.entry_handler = entry_only_handler,
+	};
+
+	current_test = test;
+
+	/* Test Case 1: Register in order 1 -> 2 */
+	KUNIT_EXPECT_EQ(test, 0, register_fprobe(&fp1, "fprobe_selftest_target", NULL));
+	KUNIT_EXPECT_EQ(test, 0, register_fprobe(&fp2, "fprobe_selftest_target", NULL));
+
+	check_fprobe_multi(test);
+
+	/* Unregister all */
+	KUNIT_EXPECT_EQ(test, 0, unregister_fprobe(&fp1));
+	KUNIT_EXPECT_EQ(test, 0, unregister_fprobe(&fp2));
+
+	/* Test Case 2: Register in order 2 -> 1 */
+	KUNIT_EXPECT_EQ(test, 0, register_fprobe(&fp2, "fprobe_selftest_target", NULL));
+	KUNIT_EXPECT_EQ(test, 0, register_fprobe(&fp1, "fprobe_selftest_target", NULL));
+
+	check_fprobe_multi(test);
+
+	/* Unregister all */
+	KUNIT_EXPECT_EQ(test, 0, unregister_fprobe(&fp1));
+	KUNIT_EXPECT_EQ(test, 0, unregister_fprobe(&fp2));
+}
+
+static unsigned long get_ftrace_location(void *func)
+{
+	unsigned long size, addr = (unsigned long)func;
+
+	if (!kallsyms_lookup_size_offset(addr, &size, NULL) || !size)
+		return 0;
+
+	return ftrace_location_range(addr, addr + size - 1);
+}
+
+static int fprobe_test_init(struct kunit *test)
+{
+	rand1 = get_random_u32_above(div_factor);
+	target = fprobe_selftest_target;
+	target2 = fprobe_selftest_target2;
+	target_ip = get_ftrace_location(target);
+	target2_ip = get_ftrace_location(target2);
+
+	return 0;
+}
+
+static struct kunit_case fprobe_testcases[] = {
+	KUNIT_CASE(test_fprobe_entry),
+	KUNIT_CASE(test_fprobe),
+	KUNIT_CASE(test_fprobe_syms),
+	KUNIT_CASE(test_fprobe_data),
+	KUNIT_CASE(test_fprobe_skip),
+	KUNIT_CASE(test_fprobe_multi),
+	{}
+};
+
+static struct kunit_suite fprobe_test_suite = {
+	.name = "fprobe_test",
+	.init = fprobe_test_init,
+	.test_cases = fprobe_testcases,
+};
+
+kunit_test_suites(&fprobe_test_suite);
+