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, 184 insertions, 0 deletions

diff --git a/scripts/atomic/gen-atomic-instrumented.sh b/scripts/atomic/gen-atomic-instrumented.sh
new file mode 100755
index 000000000..9c1d53f81
--- /dev/null
+++ b/scripts/atomic/gen-atomic-instrumented.sh
@@ -0,0 +1,184 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0
+
+ATOMICDIR=$(dirname $0)
+
+. ${ATOMICDIR}/atomic-tbl.sh
+
+#gen_param_check(meta, arg)
+gen_param_check()
+{
+	local meta="$1"; shift
+	local arg="$1"; shift
+	local type="${arg%%:*}"
+	local name="$(gen_param_name "${arg}")"
+	local rw="atomic_write"
+
+	case "${type#c}" in
+	i) return;;
+	esac
+
+	if [ ${type#c} != ${type} ]; then
+		# We don't write to constant parameters.
+		rw="atomic_read"
+	elif [ "${type}" = "p" ] ; then
+		# The "old" argument in try_cmpxchg() gets accessed non-atomically
+		rw="read_write"
+	elif [ "${meta}" != "s" ]; then
+		# An atomic RMW: if this parameter is not a constant, and this atomic is
+		# not just a 's'tore, this parameter is both read from and written to.
+		rw="atomic_read_write"
+	fi
+
+	printf "\tinstrument_${rw}(${name}, sizeof(*${name}));\n"
+}
+
+#gen_params_checks(meta, arg...)
+gen_params_checks()
+{
+	local meta="$1"; shift
+	local order="$1"; shift
+
+	if [ "${order}" = "_release" ]; then
+		printf "\tkcsan_release();\n"
+	elif [ -z "${order}" ] && ! meta_in "$meta" "slv"; then
+		# RMW with return value is fully ordered
+		printf "\tkcsan_mb();\n"
+	fi
+
+	while [ "$#" -gt 0 ]; do
+		gen_param_check "$meta" "$1"
+		shift;
+	done
+}
+
+#gen_proto_order_variant(meta, pfx, name, sfx, order, atomic, int, arg...)
+gen_proto_order_variant()
+{
+	local meta="$1"; shift
+	local pfx="$1"; shift
+	local name="$1"; shift
+	local sfx="$1"; shift
+	local order="$1"; shift
+	local atomic="$1"; shift
+	local int="$1"; shift
+
+	local atomicname="${atomic}_${pfx}${name}${sfx}${order}"
+
+	local ret="$(gen_ret_type "${meta}" "${int}")"
+	local params="$(gen_params "${int}" "${atomic}" "$@")"
+	local checks="$(gen_params_checks "${meta}" "${order}" "$@")"
+	local args="$(gen_args "$@")"
+	local retstmt="$(gen_ret_stmt "${meta}")"
+
+	gen_kerneldoc "" "${meta}" "${pfx}" "${name}" "${sfx}" "${order}" "${atomic}" "${int}" "$@"
+
+cat <<EOF
+static __always_inline ${ret}
+${atomicname}(${params})
+{
+${checks}
+	${retstmt}raw_${atomicname}(${args});
+}
+EOF
+
+	printf "\n"
+}
+
+gen_xchg()
+{
+	local xchg="$1"; shift
+	local order="$1"; shift
+
+	kcsan_barrier=""
+	if [ "${xchg%_local}" = "${xchg}" ]; then
+		case "$order" in
+		_release)	kcsan_barrier="kcsan_release()" ;;
+		"")			kcsan_barrier="kcsan_mb()" ;;
+		esac
+	fi
+
+	if [ "${xchg%${xchg#try_cmpxchg}}" = "try_cmpxchg" ] ; then
+
+cat <<EOF
+#define ${xchg}${order}(ptr, oldp, ...) \\
+({ \\
+	typeof(ptr) __ai_ptr = (ptr); \\
+	typeof(oldp) __ai_oldp = (oldp); \\
+EOF
+[ -n "$kcsan_barrier" ] && printf "\t${kcsan_barrier}; \\\\\n"
+cat <<EOF
+	instrument_atomic_read_write(__ai_ptr, sizeof(*__ai_ptr)); \\
+	instrument_read_write(__ai_oldp, sizeof(*__ai_oldp)); \\
+	raw_${xchg}${order}(__ai_ptr, __ai_oldp, __VA_ARGS__); \\
+})
+EOF
+
+	else
+
+cat <<EOF
+#define ${xchg}${order}(ptr, ...) \\
+({ \\
+	typeof(ptr) __ai_ptr = (ptr); \\
+EOF
+[ -n "$kcsan_barrier" ] && printf "\t${kcsan_barrier}; \\\\\n"
+cat <<EOF
+	instrument_atomic_read_write(__ai_ptr, sizeof(*__ai_ptr)); \\
+	raw_${xchg}${order}(__ai_ptr, __VA_ARGS__); \\
+})
+EOF
+
+	fi
+}
+
+cat << EOF
+// SPDX-License-Identifier: GPL-2.0
+
+// Generated by $0
+// DO NOT MODIFY THIS FILE DIRECTLY
+
+/*
+ * This file provoides atomic operations with explicit instrumentation (e.g.
+ * KASAN, KCSAN), which should be used unless it is necessary to avoid
+ * instrumentation. Where it is necessary to aovid instrumenation, the
+ * raw_atomic*() operations should be used.
+ */
+#ifndef _LINUX_ATOMIC_INSTRUMENTED_H
+#define _LINUX_ATOMIC_INSTRUMENTED_H
+
+#include <linux/build_bug.h>
+#include <linux/compiler.h>
+#include <linux/instrumented.h>
+
+EOF
+
+grep '^[a-z]' "$1" | while read name meta args; do
+	gen_proto "${meta}" "${name}" "atomic" "int" ${args}
+done
+
+grep '^[a-z]' "$1" | while read name meta args; do
+	gen_proto "${meta}" "${name}" "atomic64" "s64" ${args}
+done
+
+grep '^[a-z]' "$1" | while read name meta args; do
+	gen_proto "${meta}" "${name}" "atomic_long" "long" ${args}
+done
+
+
+for xchg in "xchg" "cmpxchg" "cmpxchg64" "cmpxchg128" "try_cmpxchg" "try_cmpxchg64" "try_cmpxchg128"; do
+	for order in "" "_acquire" "_release" "_relaxed"; do
+		gen_xchg "${xchg}" "${order}"
+		printf "\n"
+	done
+done
+
+for xchg in "cmpxchg_local" "cmpxchg64_local" "cmpxchg128_local" "sync_cmpxchg" \
+	    "try_cmpxchg_local" "try_cmpxchg64_local" "try_cmpxchg128_local" "sync_try_cmpxchg"; do
+	gen_xchg "${xchg}" ""
+	printf "\n"
+done
+
+cat <<EOF
+
+#endif /* _LINUX_ATOMIC_INSTRUMENTED_H */
+EOF