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, 273 insertions, 0 deletions

diff --git a/scripts/package/mkdebian b/scripts/package/mkdebian
new file mode 100755
index 000000000..d4b007b38
--- /dev/null
+++ b/scripts/package/mkdebian
@@ -0,0 +1,273 @@
+#!/bin/sh
+#
+# Copyright 2003 Wichert Akkerman <wichert@wiggy.net>
+#
+# Simple script to generate a debian/ directory for a Linux kernel.
+
+set -eu
+
+is_enabled() {
+	grep -q "^$1=y" include/config/auto.conf
+}
+
+if_enabled_echo() {
+	if is_enabled "$1"; then
+		echo -n "$2"
+	elif [ $# -ge 3 ]; then
+		echo -n "$3"
+	fi
+}
+
+set_debarch() {
+	if [ "${KBUILD_DEBARCH:+set}" ]; then
+		debarch="$KBUILD_DEBARCH"
+		return
+	fi
+
+	# Attempt to find the correct Debian architecture
+	case "$UTS_MACHINE" in
+	i386|alpha|m68k|riscv*)
+		debarch="$UTS_MACHINE" ;;
+	x86_64)
+		debarch=amd64 ;;
+	sparc*)
+		debarch=sparc$(if_enabled_echo CONFIG_64BIT 64) ;;
+	s390*)
+		debarch=s390x ;;
+	ppc*)
+		if is_enabled CONFIG_64BIT; then
+			debarch=ppc64$(if_enabled_echo CONFIG_CPU_LITTLE_ENDIAN el)
+		else
+			debarch=powerpc$(if_enabled_echo CONFIG_SPE spe)
+		fi
+		;;
+	parisc*)
+		debarch=hppa ;;
+	mips*)
+		if is_enabled CONFIG_CPU_LITTLE_ENDIAN; then
+			debarch=mips$(if_enabled_echo CONFIG_64BIT 64)$(if_enabled_echo CONFIG_CPU_MIPSR6 r6)el
+		elif is_enabled CONFIG_CPU_MIPSR6; then
+			debarch=mips$(if_enabled_echo CONFIG_64BIT 64)r6
+		else
+			debarch=mips
+		fi
+		;;
+	aarch64|arm64)
+		debarch=arm64 ;;
+	arm*)
+		if is_enabled CONFIG_AEABI; then
+			debarch=arm$(if_enabled_echo CONFIG_VFP hf el)
+		else
+			debarch=arm
+		fi
+		;;
+	openrisc)
+		debarch=or1k ;;
+	sh)
+		if is_enabled CONFIG_CPU_SH3; then
+			debarch=sh3$(if_enabled_echo CONFIG_CPU_BIG_ENDIAN eb)
+		elif is_enabled CONFIG_CPU_SH4; then
+			debarch=sh4$(if_enabled_echo CONFIG_CPU_BIG_ENDIAN eb)
+		fi
+		;;
+	um)
+		if is_enabled CONFIG_64BIT; then
+			debarch=amd64
+		else
+			debarch=i386
+		fi
+		;;
+	loongarch64)
+		debarch=loong64 ;;
+	esac
+	if [ -z "$debarch" ]; then
+		debarch=$(dpkg-architecture -qDEB_HOST_ARCH)
+		echo "" >&2
+		echo "** ** **  WARNING  ** ** **" >&2
+		echo "" >&2
+		echo "Your architecture doesn't have its equivalent" >&2
+		echo "Debian userspace architecture defined!" >&2
+		echo "Falling back to the current host architecture ($debarch)." >&2
+		echo "Please add support for $UTS_MACHINE to ${0} ..." >&2
+		echo "" >&2
+	fi
+}
+
+# Create debian/source/ if it is a source package build
+gen_source ()
+{
+	mkdir -p debian/source
+
+	echo "3.0 (quilt)" > debian/source/format
+
+	{
+		echo "diff-ignore"
+		echo "extend-diff-ignore = .*"
+	} > debian/source/local-options
+
+	# Add .config as a patch
+	mkdir -p debian/patches
+	{
+		echo "Subject: Add .config"
+		echo "Author: ${maintainer}"
+		echo
+		echo "--- /dev/null"
+		echo "+++ linux/.config"
+		diff -u /dev/null "${KCONFIG_CONFIG}" | tail -n +3
+	} > debian/patches/config.patch
+	echo config.patch > debian/patches/series
+
+	"${srctree}/scripts/package/gen-diff-patch" debian/patches/diff.patch
+	if [ -s debian/patches/diff.patch ]; then
+		sed -i "
+			1iSubject: Add local diff
+			1iAuthor: ${maintainer}
+			1i
+		" debian/patches/diff.patch
+
+		echo diff.patch >> debian/patches/series
+	else
+		rm -f debian/patches/diff.patch
+	fi
+}
+
+rm -rf debian
+mkdir debian
+
+user=${KBUILD_BUILD_USER:-$(id -nu)}
+name=${DEBFULLNAME:-${user}}
+if [ "${DEBEMAIL:+set}" ]; then
+	email=${DEBEMAIL}
+else
+	buildhost=${KBUILD_BUILD_HOST:-$(hostname -f 2>/dev/null || hostname)}
+	email="${user}@${buildhost}"
+fi
+maintainer="${name} <${email}>"
+
+while [ $# -gt 0 ]; do
+	case "$1" in
+	--need-source)
+		gen_source
+		shift
+		;;
+	*)
+		break
+		;;
+	esac
+done
+
+# Some variables and settings used throughout the script
+if [ "${KDEB_PKGVERSION:+set}" ]; then
+	packageversion=$KDEB_PKGVERSION
+else
+	upstream_version=$("${srctree}/scripts/setlocalversion" --no-local "${srctree}" | sed 's/-\(rc[1-9]\)/~\1/')
+	debian_revision=$("${srctree}/scripts/build-version")
+	packageversion=${upstream_version}-${debian_revision}
+fi
+sourcename=${KDEB_SOURCENAME:-linux-upstream}
+
+if [ "$ARCH" = "um" ] ; then
+	packagename=user-mode-linux
+else
+	packagename=linux-image
+fi
+
+debarch=
+set_debarch
+
+# Try to determine distribution
+if [ "${KDEB_CHANGELOG_DIST:+set}" ]; then
+        distribution=$KDEB_CHANGELOG_DIST
+# In some cases lsb_release returns the codename as n/a, which breaks dpkg-parsechangelog
+elif distribution=$(lsb_release -cs 2>/dev/null) && [ -n "$distribution" ] && [ "$distribution" != "n/a" ]; then
+        : # nothing to do in this case
+else
+        distribution="unstable"
+        echo >&2 "Using default distribution of 'unstable' in the changelog"
+        echo >&2 "Install lsb-release or set \$KDEB_CHANGELOG_DIST explicitly"
+fi
+
+echo $debarch > debian/arch
+
+host_gnu=$(dpkg-architecture -a "${debarch}" -q DEB_HOST_GNU_TYPE | sed 's/_/-/g')
+
+# Generate a simple changelog template
+cat <<EOF > debian/changelog
+$sourcename ($packageversion) $distribution; urgency=low
+
+  * Custom built Linux kernel.
+
+ -- $maintainer  $(date -R)
+EOF
+
+# Generate a control file
+cat <<EOF > debian/control
+Source: $sourcename
+Section: kernel
+Priority: optional
+Maintainer: $maintainer
+Rules-Requires-Root: no
+Build-Depends: debhelper-compat (= 12)
+Build-Depends-Arch: bc, bison, flex,
+ gcc-${host_gnu} <!pkg.${sourcename}.nokernelheaders>,
+ kmod, libdw-dev:native, libelf-dev:native,
+ libssl-dev:native, libssl-dev <!pkg.${sourcename}.nokernelheaders>,
+ python3:native, rsync
+Homepage: https://www.kernel.org/
+
+Package: $packagename-${KERNELRELEASE}
+Architecture: $debarch
+Description: Linux kernel, version ${KERNELRELEASE}
+ This package contains the Linux kernel, modules and corresponding other
+ files, version: ${KERNELRELEASE}.
+EOF
+
+if [ "${SRCARCH}" != um ]; then
+cat <<EOF >> debian/control
+
+Package: linux-libc-dev
+Section: devel
+Provides: linux-kernel-headers
+Architecture: $debarch
+Description: Linux support headers for userspace development
+ This package provides userspaces headers from the Linux kernel.  These headers
+ are used by the installed headers for GNU glibc and other system libraries.
+Multi-Arch: same
+EOF
+
+if is_enabled CONFIG_MODULES; then
+cat <<EOF >> debian/control
+
+Package: linux-headers-${KERNELRELEASE}
+Architecture: $debarch
+Build-Profiles: <!pkg.${sourcename}.nokernelheaders>
+Description: Linux kernel headers for ${KERNELRELEASE} on $debarch
+ This package provides kernel header files for ${KERNELRELEASE} on $debarch
+ .
+ This is useful for people who need to build external modules
+EOF
+fi
+fi
+
+if is_enabled CONFIG_DEBUG_INFO; then
+cat <<EOF >> debian/control
+
+Package: linux-image-${KERNELRELEASE}-dbg
+Section: debug
+Architecture: $debarch
+Build-Profiles: <!pkg.${sourcename}.nokerneldbg>
+Description: Linux kernel debugging symbols for ${KERNELRELEASE}
+ This package will come in handy if you need to debug the kernel. It provides
+ all the necessary debug symbols for the kernel and its modules.
+EOF
+fi
+
+cat <<EOF > debian/rules.vars
+ARCH := ${ARCH}
+KERNELRELEASE := ${KERNELRELEASE}
+EOF
+
+cp "${srctree}/scripts/package/debian/copyright" debian/
+cp "${srctree}/scripts/package/debian/rules" debian/
+
+exit 0