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/arch/powerpc/boot/gamecube-head.S b/arch/powerpc/boot/gamecube-head.S
new file mode 100644
index 000000000..ccf5f1045
--- /dev/null
+++ b/arch/powerpc/boot/gamecube-head.S
@@ -0,0 +1,106 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * arch/powerpc/boot/gamecube-head.S
+ *
+ * Nintendo GameCube bootwrapper entry.
+ * Copyright (C) 2004-2009 The GameCube Linux Team
+ * Copyright (C) 2008,2009 Albert Herranz
+ */
+
+#include "ppc_asm.h"
+
+/*
+ * The entry code does no assumptions regarding:
+ * - if the data and instruction caches are enabled or not
+ * - if the MMU is enabled or not
+ *
+ * We enable the caches if not already enabled, enable the MMU with an
+ * identity mapping scheme and jump to the start code.
+ */
+
+	.text
+
+	.globl _zimage_start
+_zimage_start:
+
+	/* turn the MMU off */
+	mfmsr	9
+	rlwinm	9, 9, 0, ~((1<<4)|(1<<5)) /* MSR_DR|MSR_IR */
+	bcl	20, 31, 1f
+1:
+	mflr	8
+	clrlwi	8, 8, 3		/* convert to a real address */
+	addi	8, 8, _mmu_off - 1b
+	mtsrr0	8
+	mtsrr1	9
+	rfi
+_mmu_off:
+	/* MMU disabled */
+
+	/* setup BATs */
+	isync
+	li      8, 0
+	mtspr	0x210, 8	/* IBAT0U */
+	mtspr	0x212, 8	/* IBAT1U */
+	mtspr	0x214, 8	/* IBAT2U */
+	mtspr	0x216, 8	/* IBAT3U */
+	mtspr	0x218, 8	/* DBAT0U */
+	mtspr	0x21a, 8	/* DBAT1U */
+	mtspr	0x21c, 8	/* DBAT2U */
+	mtspr	0x21e, 8	/* DBAT3U */
+
+	li	8, 0x01ff	/* first 16MiB */
+	li	9, 0x0002	/* rw */
+	mtspr	0x211, 9	/* IBAT0L */
+	mtspr	0x210, 8	/* IBAT0U */
+	mtspr	0x219, 9	/* DBAT0L */
+	mtspr	0x218, 8	/* DBAT0U */
+
+	lis	8, 0x0c00	/* I/O mem */
+	ori	8, 8, 0x3ff	/* 32MiB */
+	lis	9, 0x0c00
+	ori	9, 9, 0x002a	/* uncached, guarded, rw */
+	mtspr	0x21b, 9	/* DBAT1L */
+	mtspr	0x21a, 8	/* DBAT1U */
+
+	lis	8, 0x0100	/* next 8MiB */
+	ori	8, 8, 0x00ff	/* 8MiB */
+	lis	9, 0x0100
+	ori	9, 9, 0x0002	/* rw */
+	mtspr	0x215, 9	/* IBAT2L */
+	mtspr	0x214, 8	/* IBAT2U */
+	mtspr	0x21d, 9	/* DBAT2L */
+	mtspr	0x21c, 8	/* DBAT2U */
+
+	/* enable and invalidate the caches if not already enabled */
+	mfspr	8, 0x3f0	/* HID0 */
+	andi.	0, 8, (1<<15)		/* HID0_ICE */
+	bne	1f
+	ori	8, 8, (1<<15)|(1<<11)	/* HID0_ICE|HID0_ICFI*/
+1:
+	andi.	0, 8, (1<<14)		/* HID0_DCE */
+	bne	1f
+	ori	8, 8, (1<<14)|(1<<10)	/* HID0_DCE|HID0_DCFI*/
+1:
+	mtspr	0x3f0, 8	/* HID0 */
+	isync
+
+	/* initialize arguments */
+	li	3, 0
+	li	4, 0
+	li	5, 0
+
+	/* turn the MMU on */
+	bcl	20, 31, 1f
+1:
+	mflr	8
+	addi	8, 8, _mmu_on - 1b
+	mfmsr	9
+	ori	9, 9, (1<<4)|(1<<5) /* MSR_DR|MSR_IR */
+	mtsrr0	8
+	mtsrr1	9
+	sync
+	rfi
+_mmu_on:
+	b _zimage_start_lib
+