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, 105 insertions, 0 deletions

diff --git a/Documentation/ABI/testing/sysfs-driver-wacom b/Documentation/ABI/testing/sysfs-driver-wacom
new file mode 100644
index 000000000..16acaa571
--- /dev/null
+++ b/Documentation/ABI/testing/sysfs-driver-wacom
@@ -0,0 +1,105 @@
+What:		/sys/bus/hid/devices/<bus>:<vid>:<pid>.<n>/speed
+Date:		April 2010
+KernelVersion:	2.6.35
+Contact:	linux-bluetooth@vger.kernel.org
+Description:
+		The /sys/bus/hid/devices/<bus>:<vid>:<pid>.<n>/speed file
+		controls reporting speed of Wacom bluetooth tablet. Reading
+		from this file returns 1 if tablet reports in high speed mode
+		or 0 otherwise. Writing to this file one of these values
+		switches reporting speed.
+
+What:		/sys/bus/hid/devices/<bus>:<vid>:<pid>.<n>/wacom_led/led
+Date:		August 2014
+Contact:	linux-input@vger.kernel.org
+Description:
+		Attribute group for control of the status LEDs and the OLEDs.
+		This attribute group is only available for Intuos 4 M, L,
+		and XL (with LEDs and OLEDs), Intuos 4 WL, Intuos 5 (LEDs only),
+		Intuos Pro (LEDs only) and Cintiq 21UX2 and Cintiq 24HD
+		(LEDs only). Therefore its presence implicitly signifies the
+		presence of said LEDs and OLEDs on the tablet device.
+
+What:		/sys/bus/hid/devices/<bus>:<vid>:<pid>.<n>/wacom_led/status0_luminance
+Date:		August 2014
+Contact:	linux-input@vger.kernel.org
+Description:
+		<obsoleted by the LED class API now exported by the driver>
+		Writing to this file sets the status LED luminance (1..127)
+		when the stylus does not touch the tablet surface, and no
+		button is pressed on the stylus. This luminance level is
+		normally lower than the level when a button is pressed.
+
+What:		/sys/bus/hid/devices/<bus>:<vid>:<pid>.<n>/wacom_led/status1_luminance
+Date:		August 2014
+Contact:	linux-input@vger.kernel.org
+Description:
+		<obsoleted by the LED class API now exported by the driver>
+		Writing to this file sets the status LED luminance (1..127)
+		when the stylus touches the tablet surface, or any button is
+		pressed on the stylus.
+
+What:		/sys/bus/hid/devices/<bus>:<vid>:<pid>.<n>/wacom_led/status_led0_select
+Date:		August 2014
+Contact:	linux-input@vger.kernel.org
+Description:
+		<obsoleted by the LED class API now exported by the driver>
+		Writing to this file sets which one of the four (for Intuos 4
+		and Intuos 5) or of the right four (for Cintiq 21UX2 and Cintiq
+		24HD) status LEDs is active (0..3). The other three LEDs on the
+		same side are always inactive.
+
+What:		/sys/bus/hid/devices/<bus>:<vid>:<pid>.<n>/wacom_led/status_led1_select
+Date:		August 2014
+Contact:	linux-input@vger.kernel.org
+Description:
+		<obsoleted by the LED class API now exported by the driver>
+		Writing to this file sets which one of the left four (for Cintiq 21UX2
+		and Cintiq 24HD) status LEDs is active (0..3). The other three LEDs on
+		the left are always inactive.
+
+What:		/sys/bus/hid/devices/<bus>:<vid>:<pid>.<n>/wacom_led/buttons_luminance
+Date:		August 2014
+Contact:	linux-input@vger.kernel.org
+Description:
+		Writing to this file sets the overall luminance level (0..15)
+		of all eight button OLED displays.
+
+What:		/sys/bus/hid/devices/<bus>:<vid>:<pid>.<n>/wacom_led/button<n>_rawimg
+Date:		August 2014
+Contact:	linux-input@vger.kernel.org
+Description:
+		When writing a 1024 byte raw image in Wacom Intuos 4
+		interleaving format to the file, the image shows up on Button N
+		of the device. The image is a 64x32 pixel 4-bit gray image. The
+		1024 byte binary is split up into 16x 64 byte chunks. Each 64
+		byte chunk encodes the image data for two consecutive lines on
+		the display. The low nibble of each byte contains the first
+		line, and the high nibble contains the second line.
+		When the Wacom Intuos 4 is connected over Bluetooth, the
+		image has to contain 256 bytes (64x32 px 1 bit colour).
+		The format is also scrambled, like in the USB mode, and it can
+		be summarized by converting::
+
+					    76543210 into GECA6420.
+					    HGFEDCBA      HFDB7531
+
+What:		/sys/bus/hid/devices/<bus>:<vid>:<pid>.<n>/wacom_remote/unpair_remote
+Date:		July 2015
+Contact:	linux-input@vger.kernel.org
+Description:
+		Writing the character sequence '*' followed by a newline to
+		this file will delete all of the current pairings on the
+		device. Other character sequences are reserved. This file is
+		write only.
+
+What:		/sys/bus/hid/devices/<bus>:<vid>:<pid>.<n>/wacom_remote/<serial_number>/remote_mode
+Date:		July 2015
+Contact:	linux-input@vger.kernel.org
+Description:
+		<obsoleted by the LED class API now exported by the driver>
+		Reading from this file reports the mode status of the
+		remote as indicated by the LED lights on the device. If no
+		reports have been received from the paired device, reading
+		from this file will report '-1'. The mode is read-only
+		and cannot be set through the driver.