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

4 files changed, 376 insertions, 0 deletions

diff --git a/arch/sparc/boot/.gitignore b/arch/sparc/boot/.gitignore
new file mode 100644
index 000000000..f3d8569a2
--- /dev/null
+++ b/arch/sparc/boot/.gitignore
@@ -0,0 +1,9 @@
+# SPDX-License-Identifier: GPL-2.0-only
+btfix.S
+btfixupprep
+image
+zImage
+tftpboot.img
+vmlinux.aout
+piggyback
+
diff --git a/arch/sparc/boot/Makefile b/arch/sparc/boot/Makefile
new file mode 100644
index 000000000..339c42d35
--- /dev/null
+++ b/arch/sparc/boot/Makefile
@@ -0,0 +1,72 @@
+# SPDX-License-Identifier: GPL-2.0
+# Makefile for the Sparc boot stuff.
+#
+# Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
+# Copyright (C) 1997,1998 Jakub Jelinek (jj@ultra.linux.cz)
+
+ROOT_IMG	:= /usr/src/root.img
+ELFTOAOUT	:= elftoaout
+
+hostprogs	:= piggyback
+targets		:= tftpboot.img image zImage vmlinux.aout
+clean-files	:= System.map
+
+quiet_cmd_elftoaout	= ELFTOAOUT $@
+      cmd_elftoaout	= $(ELFTOAOUT) $(obj)/image -o $@
+quiet_cmd_piggy		= PIGGY   $@
+      cmd_piggy		= $(obj)/piggyback $(BITS) $@ System.map $(ROOT_IMG)
+quiet_cmd_strip		= STRIP   $@
+      cmd_strip		= $(STRIP) -R .comment -R .note -K sun4u_init -K _end -K _start $< -o $@
+
+ifeq ($(CONFIG_SPARC64),y)
+
+# Actual linking
+
+$(obj)/zImage: $(obj)/image FORCE
+	$(call if_changed,gzip)
+	@$(kecho) 'Kernel: $@ is ready' '(#'$(or $(KBUILD_BUILD_VERSION),`cat .version`)')'
+
+$(obj)/vmlinux.aout: vmlinux FORCE
+	$(call if_changed,elftoaout)
+	@$(kecho) 'Kernel: $@ is ready' '(#'$(or $(KBUILD_BUILD_VERSION),`cat .version`)')'
+else
+
+$(obj)/zImage: $(obj)/image FORCE
+	$(call if_changed,strip)
+	@$(kecho) 'Kernel: $@ is ready' '(#'$(or $(KBUILD_BUILD_VERSION),`cat .version`)')'
+
+# The following lines make a readable image for U-Boot.
+#  uImage   - Binary file read by U-boot
+#  uImage.o - object file of uImage for loading with a
+#             flash programmer understanding ELF.
+
+OBJCOPYFLAGS_image.bin := -S -O binary -R .note -R .comment
+$(obj)/image.bin: $(obj)/image FORCE
+	$(call if_changed,objcopy)
+
+$(obj)/image.gz: $(obj)/image.bin FORCE
+	$(call if_changed,gzip)
+
+UIMAGE_LOADADDR = $(CONFIG_UBOOT_LOAD_ADDR)
+UIMAGE_ENTRYADDR = $(CONFIG_UBOOT_ENTRY_ADDR)
+UIMAGE_COMPRESSION = gzip
+
+quiet_cmd_uimage.o = UIMAGE.O $@
+      cmd_uimage.o = $(LD) -Tdata $(CONFIG_UBOOT_FLASH_ADDR) \
+                     -r -b binary $@ -o $@.o
+
+targets += uImage
+$(obj)/uImage: $(obj)/image.gz FORCE
+	$(call if_changed,uimage)
+	$(call if_changed,uimage.o)
+	@$(kecho) 'Kernel: $@ is ready' '(#'$(or $(KBUILD_BUILD_VERSION),`cat .version`)')'
+
+endif
+
+$(obj)/image: vmlinux FORCE
+	$(call if_changed,strip)
+	@$(kecho) 'Kernel: $@ is ready' '(#'$(or $(KBUILD_BUILD_VERSION),`cat .version`)')'
+
+$(obj)/tftpboot.img: $(obj)/image $(obj)/piggyback System.map $(ROOT_IMG) FORCE
+	$(call if_changed,elftoaout)
+	$(call if_changed,piggy)
diff --git a/arch/sparc/boot/install.sh b/arch/sparc/boot/install.sh
new file mode 100755
index 000000000..68de67c56
--- /dev/null
+++ b/arch/sparc/boot/install.sh
@@ -0,0 +1,30 @@
+#!/bin/sh
+#
+# This file is subject to the terms and conditions of the GNU General Public
+# License.  See the file "COPYING" in the main directory of this archive
+# for more details.
+#
+# Copyright (C) 1995 by Linus Torvalds
+#
+# Adapted from code in arch/i386/boot/Makefile by H. Peter Anvin
+#
+# "make install" script for SPARC architecture
+#
+# Arguments:
+#   $1 - kernel version
+#   $2 - kernel image file
+#   $3 - kernel map file
+#   $4 - default install path (blank if root directory)
+
+set -e
+
+if [ -f $4/vmlinuz ]; then
+	mv $4/vmlinuz $4/vmlinuz.old
+fi
+
+if [ -f $4/System.map ]; then
+	mv $4/System.map $4/System.old
+fi
+
+cat $2 > $4/vmlinuz
+cp $3 $4/System.map
diff --git a/arch/sparc/boot/piggyback.c b/arch/sparc/boot/piggyback.c
new file mode 100644
index 000000000..6d74064ad
--- /dev/null
+++ b/arch/sparc/boot/piggyback.c
@@ -0,0 +1,265 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+   Simple utility to make a single-image install kernel with initial ramdisk
+   for Sparc tftpbooting without need to set up nfs.
+
+   Copyright (C) 1996,1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+   Pete Zaitcev <zaitcev@yahoo.com> endian fixes for cross-compiles, 2000.
+   Copyright (C) 2011 Sam Ravnborg <sam@ravnborg.org>
+
+ */
+
+#include <dirent.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <ctype.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <stdio.h>
+
+#include <sys/types.h>
+#include <sys/stat.h>
+
+/*
+ * Note: run this on an a.out kernel (use elftoaout for it),
+ * as PROM looks for a.out image only.
+ */
+
+#define AOUT_TEXT_OFFSET   32
+
+static int is64bit = 0;
+
+/* align to power-of-two size */
+static int align(int n)
+{
+	if (is64bit)
+		return (n + 0x1fff) & ~0x1fff;
+	else
+		return (n + 0xfff) & ~0xfff;
+}
+
+/* read two bytes as big endian */
+static unsigned short ld2(char *p)
+{
+	return (p[0] << 8) | p[1];
+}
+
+/* save 4 bytes as big endian */
+static void st4(char *p, unsigned int x)
+{
+	p[0] = x >> 24;
+	p[1] = x >> 16;
+	p[2] = x >> 8;
+	p[3] = x;
+}
+
+static void die(const char *str)
+{
+	perror(str);
+	exit(1);
+}
+
+static void usage(void)
+{
+	/* fs_img.gz is an image of initial ramdisk. */
+	fprintf(stderr, "Usage: piggyback bits vmlinux.aout System.map fs_img.gz\n");
+	fprintf(stderr, "\tKernel image will be modified in place.\n");
+	exit(1);
+}
+
+static int start_line(const char *line)
+{
+	if (strcmp(line + 10, " _start\n") == 0)
+		return 1;
+	else if (strcmp(line + 18, " _start\n") == 0)
+		return 1;
+	return 0;
+}
+
+static int end_line(const char *line)
+{
+	if (strcmp(line + 10, " _end\n") == 0)
+		return 1;
+	else if (strcmp (line + 18, " _end\n") == 0)
+		return 1;
+	return 0;
+}
+
+/*
+ * Find address for start and end in System.map.
+ * The file looks like this:
+ * f0004000 ... _start
+ * f0379f79 ... _end
+ * 1234567890123456
+ * ^coloumn 1
+ * There is support for 64 bit addresses too.
+ *
+ * Return 0 if either start or end is not found
+ */
+static int get_start_end(const char *filename, unsigned int *start,
+                                               unsigned int *end)
+{
+	FILE *map;
+	char buffer[1024];
+
+	*start = 0;
+	*end = 0;
+	map = fopen(filename, "r");
+	if (!map)
+		die(filename);
+	while (fgets(buffer, 1024, map)) {
+		if (start_line(buffer))
+			*start = strtoul(buffer, NULL, 16);
+		else if (end_line(buffer))
+			*end = strtoul(buffer, NULL, 16);
+	}
+	fclose (map);
+
+	if (*start == 0 || *end == 0)
+		return 0;
+
+	return 1;
+}
+
+#define LOOKBACK (128 * 4)
+#define BUFSIZE 1024
+/*
+ * Find the HdrS entry from head_32/head_64.
+ * We check if it is at the beginning of the file (sparc64 case)
+ * and if not we search for it.
+ * When we search do so in steps of 4 as HdrS is on a 4-byte aligned
+ * address (it is on same alignment as sparc instructions)
+ * Return the offset to the HdrS entry (as off_t)
+ */
+static off_t get_hdrs_offset(int kernelfd, const char *filename)
+{
+	char buffer[BUFSIZE];
+	off_t offset;
+	int i;
+
+	if (lseek(kernelfd, 0, SEEK_SET) < 0)
+		die("lseek");
+	if (read(kernelfd, buffer, BUFSIZE) != BUFSIZE)
+		die(filename);
+
+	if (buffer[40] == 'H' && buffer[41] == 'd' &&
+	    buffer[42] == 'r' && buffer[43] == 'S') {
+		return 40;
+	} else {
+		/*  Find the gokernel label */
+		/* Decode offset from branch instruction */
+		offset = ld2(buffer + AOUT_TEXT_OFFSET + 2) << 2;
+		/* Go back 512 bytes so we do not miss HdrS */
+		offset -= LOOKBACK;
+		/* skip a.out header */
+		offset += AOUT_TEXT_OFFSET;
+		if (offset < 0) {
+			errno = -EINVAL;
+			die("Calculated a negative offset, probably elftoaout generated an invalid image. Did you use a recent elftoaout ?");
+		}
+		if (lseek(kernelfd, offset, SEEK_SET) < 0)
+			die("lseek");
+		if (read(kernelfd, buffer, BUFSIZE) != BUFSIZE)
+			die(filename);
+
+		for (i = 0; i < LOOKBACK; i += 4) {
+			if (buffer[i + 0] == 'H' && buffer[i + 1] == 'd' &&
+			    buffer[i + 2] == 'r' && buffer[i + 3] == 'S') {
+				return offset + i;
+			}
+		}
+	}
+	fprintf (stderr, "Couldn't find headers signature in %s\n", filename);
+	exit(1);
+}
+
+int main(int argc,char **argv)
+{
+	static char aout_magic[] = { 0x01, 0x03, 0x01, 0x07 };
+	char buffer[1024];
+	unsigned int i, start, end;
+	off_t offset;
+	struct stat s;
+	int image, tail;
+
+	if (argc != 5)
+		usage();
+	if (strcmp(argv[1], "64") == 0)
+		is64bit = 1;
+	if (stat (argv[4], &s) < 0)
+		die(argv[4]);
+
+	if (!get_start_end(argv[3], &start, &end)) {
+		fprintf(stderr, "Could not determine start and end from %s\n",
+		        argv[3]);
+		exit(1);
+	}
+	if ((image = open(argv[2], O_RDWR)) < 0)
+		die(argv[2]);
+	if (read(image, buffer, 512) != 512)
+		die(argv[2]);
+	if (memcmp(buffer, aout_magic, 4) != 0) {
+		fprintf (stderr, "Not a.out. Don't blame me.\n");
+		exit(1);
+	}
+	/*
+	 * We need to fill in values for
+	 * sparc_ramdisk_image + sparc_ramdisk_size
+	 * To locate these symbols search for the "HdrS" text which appear
+	 * in the image a little before the gokernel symbol.
+	 * See definition of these in init_32.S
+	 */
+
+	offset = get_hdrs_offset(image, argv[2]);
+	/* skip HdrS + LINUX_VERSION_CODE + HdrS version */
+	offset += 10;
+
+	if (lseek(image, offset, 0) < 0)
+		die("lseek");
+
+	/*
+	 * root_flags = 0
+	 * root_dev = 1 (RAMDISK_MAJOR)
+	 * ram_flags = 0
+	 * sparc_ramdisk_image = "PAGE aligned address after _end")
+	 * sparc_ramdisk_size = size of image
+	 */
+	st4(buffer, 0);
+	st4(buffer + 4, 0x01000000);
+	st4(buffer + 8, align(end + 32));
+	st4(buffer + 12, s.st_size);
+
+	if (write(image, buffer + 2, 14) != 14)
+		die(argv[2]);
+
+	/* For sparc64 update a_text and clear a_data + a_bss */
+	if (is64bit)
+	{
+		if (lseek(image, 4, 0) < 0)
+			die("lseek");
+		/* a_text */
+		st4(buffer, align(end + 32 + 8191) - (start & ~0x3fffffUL) +
+		            s.st_size);
+		/* a_data */
+		st4(buffer + 4, 0);
+		/* a_bss */
+		st4(buffer + 8, 0);
+		if (write(image, buffer, 12) != 12)
+			die(argv[2]);
+	}
+
+	/* seek page aligned boundary in the image file and add boot image */
+	if (lseek(image, AOUT_TEXT_OFFSET - start + align(end + 32), 0) < 0)
+		die("lseek");
+	if ((tail = open(argv[4], O_RDONLY)) < 0)
+		die(argv[4]);
+	while ((i = read(tail, buffer, 1024)) > 0)
+		if (write(image, buffer, i) != i)
+			die(argv[2]);
+	if (close(image) < 0)
+		die("close");
+	if (close(tail) < 0)
+		die("close");
+	return 0;
+}