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, 308 insertions, 0 deletions

diff --git a/drivers/acpi/acpica/exdebug.c b/drivers/acpi/acpica/exdebug.c
new file mode 100644
index 000000000..a592bcc5d
--- /dev/null
+++ b/drivers/acpi/acpica/exdebug.c
@@ -0,0 +1,308 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/******************************************************************************
+ *
+ * Module Name: exdebug - Support for stores to the AML Debug Object
+ *
+ * Copyright (C) 2000 - 2026, Intel Corp.
+ *
+ *****************************************************************************/
+
+#include <acpi/acpi.h>
+#include "accommon.h"
+#include "acinterp.h"
+
+#define _COMPONENT          ACPI_EXECUTER
+ACPI_MODULE_NAME("exdebug")
+
+#ifndef ACPI_NO_ERROR_MESSAGES
+/*******************************************************************************
+ *
+ * FUNCTION:    acpi_ex_do_debug_object
+ *
+ * PARAMETERS:  source_desc         - Object to be output to "Debug Object"
+ *              level               - Indentation level (used for packages)
+ *              index               - Current package element, zero if not pkg
+ *
+ * RETURN:      None
+ *
+ * DESCRIPTION: Handles stores to the AML Debug Object. For example:
+ *              Store(INT1, Debug)
+ *
+ * This function is not compiled if ACPI_NO_ERROR_MESSAGES is set.
+ *
+ * This function is only enabled if acpi_gbl_enable_aml_debug_object is set, or
+ * if ACPI_LV_DEBUG_OBJECT is set in the acpi_dbg_level. Thus, in the normal
+ * operational case, stores to the debug object are ignored but can be easily
+ * enabled if necessary.
+ *
+ ******************************************************************************/
+void
+acpi_ex_do_debug_object(union acpi_operand_object *source_desc,
+			u32 level, u32 index)
+{
+	u32 i;
+	u32 timer;
+	union acpi_operand_object *object_desc;
+	u32 value;
+
+	ACPI_FUNCTION_TRACE_PTR(ex_do_debug_object, source_desc);
+
+	/* Output must be enabled via the debug_object global or the dbg_level */
+
+	if (!acpi_gbl_enable_aml_debug_object &&
+	    !(acpi_dbg_level & ACPI_LV_DEBUG_OBJECT)) {
+		return_VOID;
+	}
+
+	/* Newline -- don't emit the line header */
+
+	if (source_desc &&
+	    (ACPI_GET_DESCRIPTOR_TYPE(source_desc) == ACPI_DESC_TYPE_OPERAND) &&
+	    (source_desc->common.type == ACPI_TYPE_STRING)) {
+		if ((source_desc->string.length == 1) &&
+		    (*source_desc->string.pointer == '\n')) {
+			acpi_os_printf("\n");
+			return_VOID;
+		}
+	}
+
+	/*
+	 * Print line header as long as we are not in the middle of an
+	 * object display
+	 */
+	if (!((level > 0) && index == 0)) {
+		if (acpi_gbl_display_debug_timer) {
+			/*
+			 * We will emit the current timer value (in microseconds) with each
+			 * debug output. Only need the lower 26 bits. This allows for 67
+			 * million microseconds or 67 seconds before rollover.
+			 *
+			 * Convert 100 nanosecond units to microseconds
+			 */
+			timer = ((u32)acpi_os_get_timer() / 10);
+			timer &= 0x03FFFFFF;
+
+			acpi_os_printf("ACPI Debug: T=0x%8.8X %*s", timer,
+				       level, " ");
+		} else {
+			acpi_os_printf("ACPI Debug: %*s", level, " ");
+		}
+	}
+
+	/* Display the index for package output only */
+
+	if (index > 0) {
+		acpi_os_printf("(%.2u) ", index - 1);
+	}
+
+	if (!source_desc) {
+		acpi_os_printf("[Null Object]\n");
+		return_VOID;
+	}
+
+	if (ACPI_GET_DESCRIPTOR_TYPE(source_desc) == ACPI_DESC_TYPE_OPERAND) {
+
+		/* No object type prefix needed for integers and strings */
+
+		if ((source_desc->common.type != ACPI_TYPE_INTEGER) &&
+		    (source_desc->common.type != ACPI_TYPE_STRING)) {
+			acpi_os_printf("%s ",
+				       acpi_ut_get_object_type_name
+				       (source_desc));
+		}
+
+		if (!acpi_ut_valid_internal_object(source_desc)) {
+			acpi_os_printf("%p, Invalid Internal Object!\n",
+				       source_desc);
+			return_VOID;
+		}
+	} else if (ACPI_GET_DESCRIPTOR_TYPE(source_desc) ==
+		   ACPI_DESC_TYPE_NAMED) {
+		acpi_os_printf("%s (Node %p)\n",
+			       acpi_ut_get_type_name(((struct
+						       acpi_namespace_node *)
+						      source_desc)->type),
+			       source_desc);
+		return_VOID;
+	} else {
+		return_VOID;
+	}
+
+	/* source_desc is of type ACPI_DESC_TYPE_OPERAND */
+
+	switch (source_desc->common.type) {
+	case ACPI_TYPE_INTEGER:
+
+		/* Output correct integer width */
+
+		if (acpi_gbl_integer_byte_width == 4) {
+			acpi_os_printf("0x%8.8X\n",
+				       (u32)source_desc->integer.value);
+		} else {
+			acpi_os_printf("0x%8.8X%8.8X\n",
+				       ACPI_FORMAT_UINT64(source_desc->integer.
+							  value));
+		}
+		break;
+
+	case ACPI_TYPE_BUFFER:
+
+		acpi_os_printf("[0x%.2X]\n", (u32)source_desc->buffer.length);
+		acpi_ut_dump_buffer(source_desc->buffer.pointer,
+				    (source_desc->buffer.length < 256) ?
+				    source_desc->buffer.length : 256,
+				    DB_BYTE_DISPLAY, 0);
+		break;
+
+	case ACPI_TYPE_STRING:
+
+		acpi_os_printf("\"%s\"\n", source_desc->string.pointer);
+		break;
+
+	case ACPI_TYPE_PACKAGE:
+
+		acpi_os_printf("(Contains 0x%.2X Elements):\n",
+			       source_desc->package.count);
+
+		/* Output the entire contents of the package */
+
+		for (i = 0; i < source_desc->package.count; i++) {
+			acpi_ex_do_debug_object(source_desc->package.
+						elements[i], level + 4, i + 1);
+		}
+		break;
+
+	case ACPI_TYPE_LOCAL_REFERENCE:
+
+		acpi_os_printf("[%s] ",
+			       acpi_ut_get_reference_name(source_desc));
+
+		/* Decode the reference */
+
+		switch (source_desc->reference.class) {
+		case ACPI_REFCLASS_INDEX:
+
+			acpi_os_printf("0x%X\n", source_desc->reference.value);
+			break;
+
+		case ACPI_REFCLASS_TABLE:
+
+			/* Case for ddb_handle */
+
+			acpi_os_printf("Table Index 0x%X\n",
+				       source_desc->reference.value);
+			return_VOID;
+
+		default:
+
+			break;
+		}
+
+		acpi_os_printf(" ");
+
+		/* Check for valid node first, then valid object */
+
+		if (source_desc->reference.node) {
+			if (ACPI_GET_DESCRIPTOR_TYPE
+			    (source_desc->reference.node) !=
+			    ACPI_DESC_TYPE_NAMED) {
+				acpi_os_printf
+				    (" %p - Not a valid namespace node\n",
+				     source_desc->reference.node);
+			} else {
+				acpi_os_printf("Node %p [%4.4s] ",
+					       source_desc->reference.node,
+					       (source_desc->reference.node)->
+					       name.ascii);
+
+				switch ((source_desc->reference.node)->type) {
+
+					/* These types have no attached object */
+
+				case ACPI_TYPE_DEVICE:
+					acpi_os_printf("Device\n");
+					break;
+
+				case ACPI_TYPE_THERMAL:
+					acpi_os_printf("Thermal Zone\n");
+					break;
+
+				default:
+
+					acpi_ex_do_debug_object((source_desc->
+								 reference.
+								 node)->object,
+								level + 4, 0);
+					break;
+				}
+			}
+		} else if (source_desc->reference.object) {
+			if (ACPI_GET_DESCRIPTOR_TYPE
+			    (source_desc->reference.object) ==
+			    ACPI_DESC_TYPE_NAMED) {
+
+				/* Reference object is a namespace node */
+
+				acpi_ex_do_debug_object(ACPI_CAST_PTR
+							(union
+							 acpi_operand_object,
+							 source_desc->reference.
+							 object), level + 4, 0);
+			} else {
+				object_desc = source_desc->reference.object;
+				value = source_desc->reference.value;
+
+				switch (object_desc->common.type) {
+				case ACPI_TYPE_BUFFER:
+
+					acpi_os_printf("Buffer[%u] = 0x%2.2X\n",
+						       value,
+						       *source_desc->reference.
+						       index_pointer);
+					break;
+
+				case ACPI_TYPE_STRING:
+
+					acpi_os_printf
+					    ("String[%u] = \"%c\" (0x%2.2X)\n",
+					     value,
+					     *source_desc->reference.
+					     index_pointer,
+					     *source_desc->reference.
+					     index_pointer);
+					break;
+
+				case ACPI_TYPE_PACKAGE:
+
+					acpi_os_printf("Package[%u] = ", value);
+					if (!(*source_desc->reference.where)) {
+						acpi_os_printf
+						    ("[Uninitialized Package Element]\n");
+					} else {
+						acpi_ex_do_debug_object
+						    (*source_desc->reference.
+						     where, level + 4, 0);
+					}
+					break;
+
+				default:
+
+					acpi_os_printf
+					    ("Unknown Reference object type %X\n",
+					     object_desc->common.type);
+					break;
+				}
+			}
+		}
+		break;
+
+	default:
+
+		acpi_os_printf("(Descriptor %p)\n", source_desc);
+		break;
+	}
+
+	ACPI_DEBUG_PRINT_RAW((ACPI_DB_EXEC, "\n"));
+	return_VOID;
+}
+#endif