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, 106 insertions, 0 deletions

diff --git a/scripts/Makefile.kasan b/scripts/Makefile.kasan
new file mode 100644
index 000000000..91504e812
--- /dev/null
+++ b/scripts/Makefile.kasan
@@ -0,0 +1,106 @@
+# SPDX-License-Identifier: GPL-2.0
+
+ifdef CONFIG_CC_HAS_KASAN_MEMINTRINSIC_PREFIX
+# Safe for compiler to generate meminstrinsic calls in uninstrumented files.
+CFLAGS_KASAN_NOSANITIZE :=
+else
+# Don't let compiler generate memintrinsic calls in uninstrumented files
+# because they are instrumented.
+CFLAGS_KASAN_NOSANITIZE := -fno-builtin
+endif
+
+KASAN_SHADOW_OFFSET ?= $(CONFIG_KASAN_SHADOW_OFFSET)
+
+cc-param = $(call cc-option, -mllvm -$(1), $(call cc-option, --param $(1)))
+rustc-param = $(call rustc-option, -Cllvm-args=-$(1),)
+
+check-args = $(foreach arg,$(2),$(call $(1),$(arg)))
+
+kasan_params :=
+
+ifdef CONFIG_KASAN_STACK
+	stack_enable := 1
+else
+	stack_enable := 0
+endif
+
+ifdef CONFIG_KASAN_GENERIC
+
+ifdef CONFIG_KASAN_INLINE
+	# When the number of memory accesses in a function is less than this
+	# call threshold number, the compiler will use inline instrumentation.
+	# 10000 is chosen offhand as a sufficiently large number to make all
+	# kernel functions to be instrumented inline.
+	call_threshold := 10000
+else
+	call_threshold := 0
+endif
+
+# First, enable -fsanitize=kernel-address together with providing the shadow
+# mapping offset, as for GCC, -fasan-shadow-offset fails without -fsanitize
+# (GCC accepts the shadow mapping offset via -fasan-shadow-offset instead of
+# a --param like the other KASAN parameters).
+# Instead of ifdef-checking the compiler, rely on cc-option.
+CFLAGS_KASAN := $(call cc-option, -fsanitize=kernel-address \
+		-fasan-shadow-offset=$(KASAN_SHADOW_OFFSET), \
+		$(call cc-option, -fsanitize=kernel-address \
+		-mllvm -asan-mapping-offset=$(KASAN_SHADOW_OFFSET)))
+
+# The minimum supported `rustc` version has a minimum supported LLVM
+# version late enough that we can assume support for -asan-mapping-offset.
+RUSTFLAGS_KASAN := -Zsanitizer=kernel-address \
+		   -Zsanitizer-recover=kernel-address \
+		   -Cllvm-args=-asan-mapping-offset=$(KASAN_SHADOW_OFFSET)
+
+# Now, add other parameters enabled similarly in GCC, Clang, and rustc.
+# As some of them are not supported by older compilers, these will be filtered
+# through `cc-param` or `rust-param` as applicable.
+kasan_params += asan-instrumentation-with-call-threshold=$(call_threshold) \
+		asan-stack=$(stack_enable) \
+		asan-instrument-allocas=1 \
+		asan-globals=1
+
+# Instrument memcpy/memset/memmove calls by using instrumented __asan_mem*()
+# instead. With compilers that don't support this option, compiler-inserted
+# memintrinsics won't be checked by KASAN on GENERIC_ENTRY architectures.
+kasan_params += asan-kernel-mem-intrinsic-prefix=1
+
+endif # CONFIG_KASAN_GENERIC
+
+ifdef CONFIG_KASAN_SW_TAGS
+
+CFLAGS_KASAN := -fsanitize=kernel-hwaddress
+
+RUSTFLAGS_KASAN := -Zsanitizer=kernel-hwaddress \
+		   -Zsanitizer-recover=kernel-hwaddress
+
+ifdef CONFIG_KASAN_INLINE
+	kasan_params += hwasan-mapping-offset=$(KASAN_SHADOW_OFFSET)
+else
+	kasan_params += hwasan-instrument-with-calls=1
+endif
+
+kasan_params += hwasan-instrument-stack=$(stack_enable) \
+		hwasan-use-short-granules=0 \
+		hwasan-inline-all-checks=0
+
+# Instrument memcpy/memset/memmove calls by using instrumented __(hw)asan_mem*().
+ifdef CONFIG_CC_HAS_KASAN_MEMINTRINSIC_PREFIX
+	ifdef CONFIG_CC_IS_GCC
+		kasan_params += asan-kernel-mem-intrinsic-prefix=1
+	else
+		kasan_params += hwasan-kernel-mem-intrinsic-prefix=1
+	endif
+endif # CONFIG_CC_HAS_KASAN_MEMINTRINSIC_PREFIX
+
+endif # CONFIG_KASAN_SW_TAGS
+
+# Add all as-supported KASAN LLVM parameters requested by the configuration.
+CFLAGS_KASAN += $(call check-args, cc-param, $(kasan_params))
+
+ifdef CONFIG_RUST
+	# Avoid calling `rustc-param` unless Rust is enabled.
+	RUSTFLAGS_KASAN += $(call check-args, rustc-param, $(kasan_params))
+endif # CONFIG_RUST
+
+export CFLAGS_KASAN CFLAGS_KASAN_NOSANITIZE RUSTFLAGS_KASAN