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, 98 insertions, 0 deletions

diff --git a/Documentation/ABI/testing/sysfs-bus-acpi b/Documentation/ABI/testing/sysfs-bus-acpi
new file mode 100644
index 000000000..6f2b907a8
--- /dev/null
+++ b/Documentation/ABI/testing/sysfs-bus-acpi
@@ -0,0 +1,98 @@
+What:		/sys/bus/acpi/devices/.../path
+Date:		December 2006
+Contact:	Rafael J. Wysocki <rafael@kernel.org>
+Description:
+		This attribute indicates the full path of ACPI namespace
+		object associated with the device object.  For example,
+		\_SB_.PCI0.
+
+		This file is not present for device objects representing
+		fixed ACPI hardware features (like power and sleep
+		buttons).
+
+What:		/sys/bus/acpi/devices/.../modalias
+Date:		July 2007
+Contact:	Rafael J. Wysocki <rafael@kernel.org>
+Description:
+		This attribute indicates the PNP IDs of the device object.
+		That is acpi:HHHHHHHH:[CCCCCCC:].  Where each HHHHHHHH or
+		CCCCCCCC contains device object's PNPID (_HID or _CID).
+
+What:		/sys/bus/acpi/devices/.../hid
+Date:		April 2005
+Contact:	Rafael J. Wysocki <rafael@kernel.org>
+Description:
+		This attribute indicates the hardware ID (_HID) of the
+		device object.  For example, PNP0103.
+		This file is present for device objects having the _HID
+		control method.
+
+What:		/sys/bus/acpi/devices/.../description
+Date:		October 2012
+Contact:	Rafael J. Wysocki <rafael@kernel.org>
+Description:
+		This attribute contains the output of the device object's
+		_STR control method, if present.
+
+What:		/sys/bus/acpi/devices/.../adr
+Date:		October 2012
+Contact:	Rafael J. Wysocki <rafael@kernel.org>
+Description:
+		This attribute contains the output of the device object's
+		_ADR control method, which is present for ACPI device
+		objects representing devices having standard enumeration
+		algorithms, such as PCI.
+
+What:		/sys/bus/acpi/devices/.../uid
+Date:		October 2012
+Contact:	Rafael J. Wysocki <rafael@kernel.org>
+Description:
+		This attribute contains the output of the device object's
+		_UID control method, if present.
+
+What:		/sys/bus/acpi/devices/.../eject
+Date:		December 2006
+Contact:	Rafael J. Wysocki <rafael@kernel.org>
+Description:
+		Writing 1 to this attribute will trigger hot removal of
+		this device object.  This file exists for every device
+		object that has _EJ0 method.
+
+What:		/sys/bus/acpi/devices/.../status
+Date:		Jan, 2014
+Contact:	Rafael J. Wysocki <rafael@kernel.org>
+Description:
+		(RO) Returns the ACPI device status: enabled, disabled or
+		functioning or present, if the method _STA is present.
+
+		The return value is a decimal integer representing the device's
+		status bitmap:
+
+		===========  ==================================================
+		Bit [0]      Set if the device is present.
+		Bit [1]      Set if the device is enabled and decoding its
+		             resources.
+		Bit [2]      Set if the device should be shown in the UI.
+		Bit [3]      Set if the device is functioning properly (cleared
+			     if device failed its diagnostics).
+		Bit [4]      Set if the battery is present.
+		Bits [31:5]  Reserved (must be cleared)
+		===========  ==================================================
+
+		If bit [0] is clear, then bit 1 must also be clear (a device
+		that is not present cannot be enabled).
+
+		Bit 0 can be clear (not present) with bit [3] set (device is
+		functional).  This case is used to indicate a valid device for
+		which no device driver should be loaded.
+
+		More special cases are covered in the ACPI specification.
+
+What:		/sys/bus/acpi/devices/.../hrv
+Date:		Apr, 2016
+Contact:	Rafael J. Wysocki <rafael@kernel.org>
+Description:
+		(RO) Allows users to read the hardware version of non-PCI
+		hardware, if the _HRV control method is present.  It is mostly
+		useful for non-PCI devices because lspci can list the hardware
+		version for PCI devices.