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, 134 insertions, 0 deletions

diff --git a/arch/xtensa/lib/mulsi3.S b/arch/xtensa/lib/mulsi3.S
new file mode 100644
index 000000000..c6b4fd46b
--- /dev/null
+++ b/arch/xtensa/lib/mulsi3.S
@@ -0,0 +1,134 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later WITH GCC-exception-2.0 */
+#include <linux/linkage.h>
+#include <asm/asmmacro.h>
+#include <asm/core.h>
+
+	.macro	do_addx2 dst, as, at, tmp
+#if XCHAL_HAVE_ADDX
+	addx2	\dst, \as, \at
+#else
+	slli	\tmp, \as, 1
+	add	\dst, \tmp, \at
+#endif
+	.endm
+
+	.macro	do_addx4 dst, as, at, tmp
+#if XCHAL_HAVE_ADDX
+	addx4	\dst, \as, \at
+#else
+	slli	\tmp, \as, 2
+	add	\dst, \tmp, \at
+#endif
+	.endm
+
+	.macro	do_addx8 dst, as, at, tmp
+#if XCHAL_HAVE_ADDX
+	addx8	\dst, \as, \at
+#else
+	slli	\tmp, \as, 3
+	add	\dst, \tmp, \at
+#endif
+	.endm
+
+ENTRY(__mulsi3)
+
+	abi_entry_default
+
+#if XCHAL_HAVE_MUL32
+	mull	a2, a2, a3
+
+#elif XCHAL_HAVE_MUL16
+	or	a4, a2, a3
+	srai	a4, a4, 16
+	bnez	a4, .LMUL16
+	mul16u	a2, a2, a3
+	abi_ret_default
+.LMUL16:
+	srai	a4, a2, 16
+	srai	a5, a3, 16
+	mul16u	a7, a4, a3
+	mul16u	a6, a5, a2
+	mul16u	a4, a2, a3
+	add	a7, a7, a6
+	slli	a7, a7, 16
+	add	a2, a7, a4
+
+#elif XCHAL_HAVE_MAC16
+	mul.aa.hl a2, a3
+	mula.aa.lh a2, a3
+	rsr	a5, ACCLO
+	umul.aa.ll a2, a3
+	rsr	a4, ACCLO
+	slli	a5, a5, 16
+	add	a2, a4, a5
+
+#else /* !MUL32 && !MUL16 && !MAC16 */
+
+	/* Multiply one bit at a time, but unroll the loop 4x to better
+	   exploit the addx instructions and avoid overhead.
+	   Peel the first iteration to save a cycle on init.  */
+
+	/* Avoid negative numbers.  */
+	xor	a5, a2, a3	/* Top bit is 1 if one input is negative.  */
+	do_abs	a3, a3, a6
+	do_abs	a2, a2, a6
+
+	/* Swap so the second argument is smaller.  */
+	sub	a7, a2, a3
+	mov	a4, a3
+	movgez	a4, a2, a7	/* a4 = max (a2, a3) */
+	movltz	a3, a2, a7	/* a3 = min (a2, a3) */
+
+	movi	a2, 0
+	extui	a6, a3, 0, 1
+	movnez	a2, a4, a6
+
+	do_addx2 a7, a4, a2, a7
+	extui	a6, a3, 1, 1
+	movnez	a2, a7, a6
+
+	do_addx4 a7, a4, a2, a7
+	extui	a6, a3, 2, 1
+	movnez	a2, a7, a6
+
+	do_addx8 a7, a4, a2, a7
+	extui	a6, a3, 3, 1
+	movnez	a2, a7, a6
+
+	bgeui	a3, 16, .Lmult_main_loop
+	neg	a3, a2
+	movltz	a2, a3, a5
+	abi_ret_default
+
+	.align	4
+.Lmult_main_loop:
+	srli	a3, a3, 4
+	slli	a4, a4, 4
+
+	add	a7, a4, a2
+	extui	a6, a3, 0, 1
+	movnez	a2, a7, a6
+
+	do_addx2 a7, a4, a2, a7
+	extui	a6, a3, 1, 1
+	movnez	a2, a7, a6
+
+	do_addx4 a7, a4, a2, a7
+	extui	a6, a3, 2, 1
+	movnez	a2, a7, a6
+
+	do_addx8 a7, a4, a2, a7
+	extui	a6, a3, 3, 1
+	movnez	a2, a7, a6
+
+	bgeui	a3, 16, .Lmult_main_loop
+
+	neg	a3, a2
+	movltz	a2, a3, a5
+
+#endif /* !MUL32 && !MUL16 && !MAC16 */
+
+	abi_ret_default
+
+ENDPROC(__mulsi3)
+EXPORT_SYMBOL(__mulsi3)