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, 226 insertions, 0 deletions

diff --git a/scripts/package/kernel.spec b/scripts/package/kernel.spec
new file mode 100644
index 000000000..46e80970f
--- /dev/null
+++ b/scripts/package/kernel.spec
@@ -0,0 +1,226 @@
+# _arch is undefined if /usr/lib/rpm/platform/*/macros was not included.
+%{!?_arch: %define _arch dummy}
+%{!?make: %define make make}
+%define makeflags %{?_smp_mflags} ARCH=%{ARCH}
+
+Name: kernel
+Summary: The Linux Kernel
+Version: %(echo %{KERNELRELEASE} | sed -e 's/-/_/g')
+Release: %{pkg_release}%{?dist}
+License: GPL
+Group: System Environment/Kernel
+Vendor: The Linux Community
+URL: https://www.kernel.org
+Source0: linux.tar.gz
+Source1: config
+Source2: diff.patch
+Provides: kernel-%{KERNELRELEASE}
+BuildRequires: bc binutils bison dwarves
+BuildRequires: (elfutils-devel or libdw-devel)
+BuildRequires: (elfutils-libelf-devel or libelf-devel) flex
+BuildRequires: gcc make openssl openssl-devel perl python3 rsync
+
+%description
+The Linux Kernel, the operating system core itself
+
+%package headers
+Summary: Header files for the Linux kernel for use by glibc
+Group: Development/System
+Obsoletes: kernel-headers < %{version}
+Provides: kernel-headers = %{version}
+%description headers
+Kernel-headers includes the C header files that specify the interface
+between the Linux kernel and userspace libraries and programs.  The
+header files define structures and constants that are needed for
+building most standard programs and are also needed for rebuilding the
+glibc package.
+
+%if %{with_devel}
+%package devel
+Summary: Development package for building kernel modules to match the %{version} kernel
+Group: System Environment/Kernel
+AutoReqProv: no
+%description -n kernel-devel
+This package provides kernel headers and makefiles sufficient to build modules
+against the %{version} kernel package.
+%endif
+
+%if %{with_debuginfo_manual}
+%package debuginfo
+Summary: Debug information package for the Linux kernel
+Group: Development/Debug
+AutoReq: 0
+AutoProv: 1
+%description debuginfo
+This package provides debug information for the kernel image and modules from the
+%{version} package.
+%define install_mod_strip 1
+%endif
+
+%if %{with_debuginfo_rpm}
+# list of debuginfo-related options taken from distribution kernel.spec
+# files
+%undefine _include_minidebuginfo
+%undefine _find_debuginfo_dwz_opts
+%undefine _unique_build_ids
+%undefine _unique_debug_names
+%undefine _unique_debug_srcs
+%undefine _debugsource_packages
+%undefine _debuginfo_subpackages
+
+# Preserve .BTF and .BTF.base sections in kernel modules during debuginfo
+# stripping. find-debuginfo.sh uses eu-strip which removes non-allocated ELF
+# sections like .BTF by default. .BTF.base is required for BTF distillation
+# support; without it, module BTF validation fails.
+%global with_keep_section %(%{__find_debuginfo} --help 2>&1 | grep -c keep-section)
+%if %{with_keep_section}
+%global _find_debuginfo_opts -r --keep-section .BTF --keep-section .BTF.base
+%else
+%global _find_debuginfo_opts -r
+%endif
+
+%global _missing_build_ids_terminate_build 1
+%global _no_recompute_build_ids 1
+%{debug_package}
+
+# later, we make all modules executable so that find-debuginfo.sh strips
+# them up. but they don't actually need to be executable, so remove the
+# executable bit, taking care to do it _after_ find-debuginfo.sh has run
+%define __spec_install_post \
+	%{?__debug_package:%{__debug_install_post}} \
+	%{__arch_install_post} \
+	%{__os_install_post} \
+	find %{buildroot}/lib/modules/%{KERNELRELEASE} -name "*.ko" -type f \\\
+		| xargs --no-run-if-empty chmod u-x
+%else
+%define __spec_install_post /usr/lib/rpm/brp-compress || :
+%endif
+# some (but not all) versions of rpmbuild emit %%debug_package with
+# %%install. since we've already emitted it manually, that would cause
+# a package redefinition error. ensure that doesn't happen
+%define debug_package %{nil}
+
+%prep
+%setup -q -n linux
+cp %{SOURCE1} .config
+patch -p1 < %{SOURCE2}
+
+%build
+%{make} %{makeflags} KERNELRELEASE=%{KERNELRELEASE} KBUILD_BUILD_VERSION=%{release}
+
+%install
+mkdir -p %{buildroot}/lib/modules/%{KERNELRELEASE}
+cp $(%{make} %{makeflags} -s image_name) %{buildroot}/lib/modules/%{KERNELRELEASE}/vmlinuz
+# DEPMOD=true makes depmod no-op. We do not package depmod-generated files.
+%{make} %{makeflags} INSTALL_MOD_PATH=%{buildroot} %{?install_mod_strip:INSTALL_MOD_STRIP=1} DEPMOD=true modules_install
+%{make} %{makeflags} INSTALL_HDR_PATH=%{buildroot}/usr headers_install
+cp System.map %{buildroot}/lib/modules/%{KERNELRELEASE}
+cp .config %{buildroot}/lib/modules/%{KERNELRELEASE}/config
+if %{make} %{makeflags} run-command KBUILD_RUN_COMMAND='test -d ${srctree}/arch/${SRCARCH}/boot/dts' 2>/dev/null; then
+	%{make} %{makeflags} INSTALL_DTBS_PATH=%{buildroot}/lib/modules/%{KERNELRELEASE}/dtb dtbs_install
+fi
+ln -fns /usr/src/kernels/%{KERNELRELEASE} %{buildroot}/lib/modules/%{KERNELRELEASE}/build
+%if %{with_devel}
+%{make} %{makeflags} run-command KBUILD_RUN_COMMAND='${srctree}/scripts/package/install-extmod-build %{buildroot}/usr/src/kernels/%{KERNELRELEASE}'
+%endif
+
+{
+	echo "/lib/modules/%{KERNELRELEASE}"
+
+	for x in alias alias.bin builtin.alias.bin builtin.bin dep dep.bin \
+				devname softdep symbols symbols.bin weakdep; do
+		echo "%ghost /lib/modules/%{KERNELRELEASE}/modules.${x}"
+	done
+
+	for x in System.map config vmlinuz; do
+		echo "%ghost /boot/${x}-%{KERNELRELEASE}"
+	done
+
+	if [ -d "%{buildroot}/lib/modules/%{KERNELRELEASE}/dtb" ];then
+		find "%{buildroot}/lib/modules/%{KERNELRELEASE}/dtb" -printf "%%%ghost /boot/dtb-%{KERNELRELEASE}/%%P\n"
+	fi
+
+	echo "%exclude /lib/modules/%{KERNELRELEASE}/build"
+} > %{buildroot}/kernel.list
+
+%if 0%{with_debuginfo_manual}%{with_debuginfo_rpm} > 0
+# copying vmlinux directly to the debug directory means it will not get
+# stripped (but its source paths will still be collected + fixed up)
+mkdir -p %{buildroot}/usr/lib/debug/lib/modules/%{KERNELRELEASE}
+cp vmlinux %{buildroot}/usr/lib/debug/lib/modules/%{KERNELRELEASE}
+%endif
+
+%if %{with_debuginfo_rpm}
+# make modules executable so that find-debuginfo.sh strips them. this
+# will be undone later in %%__spec_install_post
+find %{buildroot}/lib/modules/%{KERNELRELEASE} -name "*.ko" -type f \
+	| xargs --no-run-if-empty chmod u+x
+%endif
+
+%if %{with_debuginfo_manual}
+echo /usr/lib/debug/lib/modules/%{KERNELRELEASE}/vmlinux > %{buildroot}/debuginfo.list
+while read -r mod; do
+	mod="${mod%.o}.ko"
+	dbg="%{buildroot}/usr/lib/debug/lib/modules/%{KERNELRELEASE}/kernel/${mod}"
+	buildid=$("${READELF:-readelf}" -n "${mod}" | sed -n 's@^.*Build ID: \(..\)\(.*\)@\1/\2@p')
+	link="%{buildroot}/usr/lib/debug/.build-id/${buildid}.debug"
+
+	mkdir -p "${dbg%/*}" "${link%/*}"
+	"${OBJCOPY:-objcopy}" --only-keep-debug "${mod}" "${dbg}"
+	ln -sf --relative "${dbg}" "${link}"
+
+	echo "${dbg#%{buildroot}}" >> %{buildroot}/debuginfo.list
+	echo "${link#%{buildroot}}" >> %{buildroot}/debuginfo.list
+done < modules.order
+%endif
+
+%clean
+rm -rf %{buildroot}
+%if %{with_debuginfo_rpm}
+rm -f debugfiles.list debuglinks.list debugsourcefiles.list debugsources.list \
+	elfbins.list
+%endif
+
+%post
+if [ -x /usr/bin/kernel-install ]; then
+	/usr/bin/kernel-install add %{KERNELRELEASE} /lib/modules/%{KERNELRELEASE}/vmlinuz
+fi
+for file in vmlinuz System.map config; do
+	if ! cmp --silent "/lib/modules/%{KERNELRELEASE}/${file}" "/boot/${file}-%{KERNELRELEASE}"; then
+		cp "/lib/modules/%{KERNELRELEASE}/${file}" "/boot/${file}-%{KERNELRELEASE}"
+	fi
+done
+if [ -d "/lib/modules/%{KERNELRELEASE}/dtb" ] && \
+   ! diff -rq "/lib/modules/%{KERNELRELEASE}/dtb" "/boot/dtb-%{KERNELRELEASE}" >/dev/null 2>&1; then
+	rm -rf "/boot/dtb-%{KERNELRELEASE}"
+	cp -r "/lib/modules/%{KERNELRELEASE}/dtb" "/boot/dtb-%{KERNELRELEASE}"
+fi
+if [ ! -e "/lib/modules/%{KERNELRELEASE}/modules.dep" ]; then
+	/usr/sbin/depmod %{KERNELRELEASE}
+fi
+
+%preun
+if [ -x /usr/bin/kernel-install ]; then
+kernel-install remove %{KERNELRELEASE}
+fi
+
+%files -f %{buildroot}/kernel.list
+%defattr (-, root, root)
+%exclude /kernel.list
+
+%files headers
+%defattr (-, root, root)
+/usr/include
+
+%if %{with_devel}
+%files devel
+%defattr (-, root, root)
+/usr/src/kernels/%{KERNELRELEASE}
+/lib/modules/%{KERNELRELEASE}/build
+%endif
+
+%if %{with_debuginfo_manual}
+%files -f %{buildroot}/debuginfo.list debuginfo
+%defattr (-, root, root)
+%exclude /debuginfo.list
+%endif