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, 104 insertions, 0 deletions

diff --git a/Documentation/input/userio.rst b/Documentation/input/userio.rst
new file mode 100644
index 000000000..7aaaa629b
--- /dev/null
+++ b/Documentation/input/userio.rst
@@ -0,0 +1,104 @@
+.. include:: <isonum.txt>
+
+===================
+The userio Protocol
+===================
+
+
+:Copyright: |copy| 2015 Lyude Paul <thatslyude@gmail.com>
+
+Sponsored by Red Hat
+
+
+Introduction
+=============
+
+This module is intended to try to make the lives of input driver developers
+easier by allowing them to test various serio devices (mainly the various
+touchpads found on laptops) without having to have the physical device in front
+of them. userio accomplishes this by allowing any privileged userspace program
+to directly interact with the kernel's serio driver and control a virtual serio
+port from there.
+
+Usage overview
+==============
+
+In order to interact with the userio kernel module, one simply opens the
+/dev/userio character device in their applications. Commands are sent to the
+kernel module by writing to the device, and any data received from the serio
+driver is read as-is from the /dev/userio device. All of the structures and
+macros you need to interact with the device are defined in <linux/userio.h> and
+<linux/serio.h>.
+
+Command Structure
+=================
+
+The struct used for sending commands to /dev/userio is as follows::
+
+	struct userio_cmd {
+		__u8 type;
+		__u8 data;
+	};
+
+``type`` describes the type of command that is being sent. This can be any one
+of the USERIO_CMD macros defined in <linux/userio.h>. ``data`` is the argument
+that goes along with the command. In the event that the command doesn't have an
+argument, this field can be left untouched and will be ignored by the kernel.
+Each command should be sent by writing the struct directly to the character
+device. In the event that the command you send is invalid, an error will be
+returned by the character device and a more descriptive error will be printed
+to the kernel log. Only one command can be sent at a time, any additional data
+written to the character device after the initial command will be ignored.
+
+To close the virtual serio port, just close /dev/userio.
+
+Commands
+========
+
+USERIO_CMD_REGISTER
+~~~~~~~~~~~~~~~~~~~
+
+Registers the port with the serio driver and begins transmitting data back and
+forth. Registration can only be performed once a port type is set with
+USERIO_CMD_SET_PORT_TYPE. Has no argument.
+
+USERIO_CMD_SET_PORT_TYPE
+~~~~~~~~~~~~~~~~~~~~~~~~
+
+Sets the type of port we're emulating, where ``data`` is the port type being
+set. Can be any of the serio type macros from <linux/serio.h>. For example:
+SERIO_8042 would set the port type to be a normal PS/2 port.
+
+USERIO_CMD_SET_PORT_PROTO
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Sets the protocol of port we're emulating, where ``data`` is the protocol being
+set. Can be any of the serio proto macros from <linux/serio.h>. For example:
+SERIO_IFORCE would set the port type to be an I-Force serial joystick.
+
+USERIO_CMD_SET_PORT_ID
+~~~~~~~~~~~~~~~~~~~~~~
+
+Sets the ``id`` value on the identification of port we're emulating, where
+``data`` is the value being set.
+
+USERIO_CMD_SET_PORT_EXTRA
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Sets the ``extra`` value on the identification of port we're emulating, where
+``data`` is the value being set.
+
+USERIO_CMD_SEND_INTERRUPT
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Sends an interrupt through the virtual serio port to the serio driver, where
+``data`` is the interrupt data being sent.
+
+Userspace tools
+===============
+
+The userio userspace tools are able to record PS/2 devices using some of the
+debugging information from i8042, and play back the devices on /dev/userio. The
+latest version of these tools can be found at:
+
+	https://github.com/Lyude/ps2emu