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

8 files changed, 336 insertions, 0 deletions

diff --git a/samples/check-exec/.gitignore b/samples/check-exec/.gitignore
new file mode 100644
index 000000000..cd759a19d
--- /dev/null
+++ b/samples/check-exec/.gitignore
@@ -0,0 +1,2 @@
+/inc
+/set-exec
diff --git a/samples/check-exec/Makefile b/samples/check-exec/Makefile
new file mode 100644
index 000000000..c4f08ad0f
--- /dev/null
+++ b/samples/check-exec/Makefile
@@ -0,0 +1,15 @@
+# SPDX-License-Identifier: BSD-3-Clause
+
+userprogs-always-y := \
+	inc \
+	set-exec
+
+userccflags += -I usr/include
+
+.PHONY: all clean
+
+all:
+	$(MAKE) -C ../.. samples/check-exec/
+
+clean:
+	$(MAKE) -C ../.. M=samples/check-exec/ clean
diff --git a/samples/check-exec/inc.c b/samples/check-exec/inc.c
new file mode 100644
index 000000000..7f6ef06a2
--- /dev/null
+++ b/samples/check-exec/inc.c
@@ -0,0 +1,212 @@
+// SPDX-License-Identifier: BSD-3-Clause
+/*
+ * Very simple script interpreter that can evaluate two different commands (one
+ * per line):
+ * - "?" to initialize a counter from user's input;
+ * - "+" to increment the counter (which is set to 0 by default).
+ *
+ * See tools/testing/selftests/exec/check-exec-tests.sh and
+ * Documentation/userspace-api/check_exec.rst
+ *
+ * Copyright © 2024 Microsoft Corporation
+ */
+
+#define _GNU_SOURCE
+#include <errno.h>
+#include <linux/fcntl.h>
+#include <linux/prctl.h>
+#include <linux/securebits.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/prctl.h>
+#include <sys/syscall.h>
+#include <unistd.h>
+
+static int sys_execveat(int dirfd, const char *pathname, char *const argv[],
+			char *const envp[], int flags)
+{
+	return syscall(__NR_execveat, dirfd, pathname, argv, envp, flags);
+}
+
+/* Returns 1 on error, 0 otherwise. */
+static int interpret_buffer(char *buffer, size_t buffer_size)
+{
+	char *line, *saveptr = NULL;
+	long long number = 0;
+
+	/* Each command is the first character of a line. */
+	saveptr = NULL;
+	line = strtok_r(buffer, "\n", &saveptr);
+	while (line) {
+		if (*line != '#' && strlen(line) != 1) {
+			fprintf(stderr, "# ERROR: Unknown string\n");
+			return 1;
+		}
+		switch (*line) {
+		case '#':
+			/* Skips shebang and comments. */
+			break;
+		case '+':
+			/* Increments and prints the number. */
+			number++;
+			printf("%lld\n", number);
+			break;
+		case '?':
+			/* Reads integer from stdin. */
+			fprintf(stderr, "> Enter new number: \n");
+			if (scanf("%lld", &number) != 1) {
+				fprintf(stderr,
+					"# WARNING: Failed to read number from stdin\n");
+			}
+			break;
+		default:
+			fprintf(stderr, "# ERROR: Unknown character '%c'\n",
+				*line);
+			return 1;
+		}
+		line = strtok_r(NULL, "\n", &saveptr);
+	}
+	return 0;
+}
+
+/* Returns 1 on error, 0 otherwise. */
+static int interpret_stream(FILE *script, char *const script_name,
+			    char *const *const envp, const bool restrict_stream)
+{
+	int err;
+	char *const script_argv[] = { script_name, NULL };
+	char buf[128] = {};
+	size_t buf_size = sizeof(buf);
+
+	/*
+	 * We pass a valid argv and envp to the kernel to emulate a native
+	 * script execution.  We must use the script file descriptor instead of
+	 * the script path name to avoid race conditions.
+	 */
+	err = sys_execveat(fileno(script), "", script_argv, envp,
+			   AT_EMPTY_PATH | AT_EXECVE_CHECK);
+	if (err && restrict_stream) {
+		perror("ERROR: Script execution check");
+		return 1;
+	}
+
+	/* Reads script. */
+	buf_size = fread(buf, 1, buf_size - 1, script);
+	return interpret_buffer(buf, buf_size);
+}
+
+static void print_usage(const char *argv0)
+{
+	fprintf(stderr, "usage: %s <script.inc> | -i | -c <command>\n\n",
+		argv0);
+	fprintf(stderr, "Example:\n");
+	fprintf(stderr, "  ./set-exec -fi -- ./inc -i < script-exec.inc\n");
+}
+
+int main(const int argc, char *const argv[], char *const *const envp)
+{
+	int opt;
+	char *cmd = NULL;
+	char *script_name = NULL;
+	bool interpret_stdin = false;
+	FILE *script_file = NULL;
+	int secbits;
+	bool deny_interactive, restrict_file;
+	size_t arg_nb;
+
+	secbits = prctl(PR_GET_SECUREBITS);
+	if (secbits == -1) {
+		/*
+		 * This should never happen, except with a buggy seccomp
+		 * filter.
+		 */
+		perror("ERROR: Failed to get securebits");
+		return 1;
+	}
+
+	deny_interactive = !!(secbits & SECBIT_EXEC_DENY_INTERACTIVE);
+	restrict_file = !!(secbits & SECBIT_EXEC_RESTRICT_FILE);
+
+	while ((opt = getopt(argc, argv, "c:i")) != -1) {
+		switch (opt) {
+		case 'c':
+			if (cmd) {
+				fprintf(stderr, "ERROR: Command already set");
+				return 1;
+			}
+			cmd = optarg;
+			break;
+		case 'i':
+			interpret_stdin = true;
+			break;
+		default:
+			print_usage(argv[0]);
+			return 1;
+		}
+	}
+
+	/* Checks that only one argument is used, or read stdin. */
+	arg_nb = !!cmd + !!interpret_stdin;
+	if (arg_nb == 0 && argc == 2) {
+		script_name = argv[1];
+	} else if (arg_nb != 1) {
+		print_usage(argv[0]);
+		return 1;
+	}
+
+	if (cmd) {
+		/*
+		 * Other kind of interactive interpretations should be denied
+		 * as well (e.g. CLI arguments passing script snippets,
+		 * environment variables interpreted as script).  However, any
+		 * way to pass script files should only be restricted according
+		 * to restrict_file.
+		 */
+		if (deny_interactive) {
+			fprintf(stderr,
+				"ERROR: Interactive interpretation denied.\n");
+			return 1;
+		}
+
+		return interpret_buffer(cmd, strlen(cmd));
+	}
+
+	if (interpret_stdin && !script_name) {
+		script_file = stdin;
+		/*
+		 * As for any execve(2) call, this path may be logged by the
+		 * kernel.
+		 */
+		script_name = "/proc/self/fd/0";
+		/*
+		 * When stdin is used, it can point to a regular file or a
+		 * pipe.  Restrict stdin execution according to
+		 * SECBIT_EXEC_DENY_INTERACTIVE but always allow executable
+		 * files (which are not considered as interactive inputs).
+		 */
+		return interpret_stream(script_file, script_name, envp,
+					deny_interactive);
+	} else if (script_name && !interpret_stdin) {
+		/*
+		 * In this sample, we don't pass any argument to scripts, but
+		 * otherwise we would have to forge an argv with such
+		 * arguments.
+		 */
+		script_file = fopen(script_name, "r");
+		if (!script_file) {
+			perror("ERROR: Failed to open script");
+			return 1;
+		}
+		/*
+		 * Restricts file execution according to
+		 * SECBIT_EXEC_RESTRICT_FILE.
+		 */
+		return interpret_stream(script_file, script_name, envp,
+					restrict_file);
+	}
+
+	print_usage(argv[0]);
+	return 1;
+}
diff --git a/samples/check-exec/run-script-ask.sh b/samples/check-exec/run-script-ask.sh
new file mode 100755
index 000000000..8ef0fdc37
--- /dev/null
+++ b/samples/check-exec/run-script-ask.sh
@@ -0,0 +1,9 @@
+#!/usr/bin/env sh
+# SPDX-License-Identifier: BSD-3-Clause
+
+DIR="$(dirname -- "$0")"
+
+PATH="${PATH}:${DIR}"
+
+set -x
+"${DIR}/script-ask.inc"
diff --git a/samples/check-exec/script-ask.inc b/samples/check-exec/script-ask.inc
new file mode 100755
index 000000000..720a8e649
--- /dev/null
+++ b/samples/check-exec/script-ask.inc
@@ -0,0 +1,5 @@
+#!/usr/bin/env inc
+# SPDX-License-Identifier: BSD-3-Clause
+
+?
++
diff --git a/samples/check-exec/script-exec.inc b/samples/check-exec/script-exec.inc
new file mode 100755
index 000000000..3245cb9d8
--- /dev/null
+++ b/samples/check-exec/script-exec.inc
@@ -0,0 +1,4 @@
+#!/usr/bin/env inc
+# SPDX-License-Identifier: BSD-3-Clause
+
++
diff --git a/samples/check-exec/script-noexec.inc b/samples/check-exec/script-noexec.inc
new file mode 100644
index 000000000..3245cb9d8
--- /dev/null
+++ b/samples/check-exec/script-noexec.inc
@@ -0,0 +1,4 @@
+#!/usr/bin/env inc
+# SPDX-License-Identifier: BSD-3-Clause
+
++
diff --git a/samples/check-exec/set-exec.c b/samples/check-exec/set-exec.c
new file mode 100644
index 000000000..ba86a60a2
--- /dev/null
+++ b/samples/check-exec/set-exec.c
@@ -0,0 +1,85 @@
+// SPDX-License-Identifier: BSD-3-Clause
+/*
+ * Simple tool to set SECBIT_EXEC_RESTRICT_FILE, SECBIT_EXEC_DENY_INTERACTIVE,
+ * before executing a command.
+ *
+ * Copyright © 2024 Microsoft Corporation
+ */
+
+#define _GNU_SOURCE
+#define __SANE_USERSPACE_TYPES__
+#include <errno.h>
+#include <linux/prctl.h>
+#include <linux/securebits.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/prctl.h>
+#include <unistd.h>
+
+static void print_usage(const char *argv0)
+{
+	fprintf(stderr, "usage: %s -f|-i -- <cmd> [args]...\n\n", argv0);
+	fprintf(stderr, "Execute a command with\n");
+	fprintf(stderr, "- SECBIT_EXEC_RESTRICT_FILE set: -f\n");
+	fprintf(stderr, "- SECBIT_EXEC_DENY_INTERACTIVE set: -i\n");
+}
+
+int main(const int argc, char *const argv[], char *const *const envp)
+{
+	const char *cmd_path;
+	char *const *cmd_argv;
+	int opt, secbits_cur, secbits_new;
+	bool has_policy = false;
+
+	secbits_cur = prctl(PR_GET_SECUREBITS);
+	if (secbits_cur == -1) {
+		/*
+		 * This should never happen, except with a buggy seccomp
+		 * filter.
+		 */
+		perror("ERROR: Failed to get securebits");
+		return 1;
+	}
+
+	secbits_new = secbits_cur;
+	while ((opt = getopt(argc, argv, "fi")) != -1) {
+		switch (opt) {
+		case 'f':
+			secbits_new |= SECBIT_EXEC_RESTRICT_FILE |
+				       SECBIT_EXEC_RESTRICT_FILE_LOCKED;
+			has_policy = true;
+			break;
+		case 'i':
+			secbits_new |= SECBIT_EXEC_DENY_INTERACTIVE |
+				       SECBIT_EXEC_DENY_INTERACTIVE_LOCKED;
+			has_policy = true;
+			break;
+		default:
+			print_usage(argv[0]);
+			return 1;
+		}
+	}
+
+	if (!argv[optind] || !has_policy) {
+		print_usage(argv[0]);
+		return 1;
+	}
+
+	if (secbits_cur != secbits_new &&
+	    prctl(PR_SET_SECUREBITS, secbits_new)) {
+		perror("Failed to set secure bit(s).");
+		fprintf(stderr,
+			"Hint: The running kernel may not support this feature.\n");
+		return 1;
+	}
+
+	cmd_path = argv[optind];
+	cmd_argv = argv + optind;
+	fprintf(stderr, "Executing command...\n");
+	execvpe(cmd_path, cmd_argv, envp);
+	fprintf(stderr, "Failed to execute \"%s\": %s\n", cmd_path,
+		strerror(errno));
+	return 1;
+}