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, 127 insertions, 0 deletions

diff --git a/Documentation/admin-guide/laptops/alienware-wmi.rst b/Documentation/admin-guide/laptops/alienware-wmi.rst
new file mode 100644
index 000000000..e532c60db
--- /dev/null
+++ b/Documentation/admin-guide/laptops/alienware-wmi.rst
@@ -0,0 +1,127 @@
+.. SPDX-License-Identifier: GPL-2.0-or-later
+
+====================
+Alienware WMI Driver
+====================
+
+Kurt Borja <kuurtb@gmail.com>
+
+This is a driver for the "WMAX" WMI device, which is found in most Dell gaming
+laptops and controls various special features.
+
+Before the launch of M-Series laptops (~2018), the "WMAX" device controlled
+basic RGB lighting, deep sleep mode, HDMI mode and amplifier status.
+
+Later, this device was completely repurpused. Now it mostly deals with thermal
+profiles, sensor monitoring and overclocking. This interface is named "AWCC" and
+is known to be used by the AWCC OEM application to control these features.
+
+The alienware-wmi driver controls both interfaces.
+
+AWCC Interface
+==============
+
+WMI device documentation: Documentation/wmi/devices/alienware-wmi.rst
+
+Supported devices
+-----------------
+
+- Alienware M-Series laptops
+- Alienware X-Series laptops
+- Alienware Aurora Desktops
+- Dell G-Series laptops
+
+If you believe your device supports the AWCC interface and you don't have any of
+the features described in this document, try the following alienware-wmi module
+parameters:
+
+- ``force_platform_profile=1``: Forces probing for platform profile support
+- ``force_hwmon=1``: Forces probing for HWMON support
+
+If the module loads successfully with these parameters, consider submitting a
+patch adding your model to the ``awcc_dmi_table`` located in
+``drivers/platform/x86/dell/alienware-wmi-wmax.c`` or contacting the maintainer
+for further guidance.
+
+Status
+------
+
+The following features are currently supported:
+
+- :ref:`Platform Profile <platform-profile>`:
+
+  - Thermal profile control
+
+  - G-Mode toggling
+
+- :ref:`HWMON <hwmon>`:
+
+  - Sensor monitoring
+
+  - Manual fan control
+
+.. _platform-profile:
+
+Platform Profile
+----------------
+
+The AWCC interface exposes various firmware defined thermal profiles. These are
+exposed to user-space through the Platform Profile class interface. Refer to
+:ref:`sysfs-class-platform-profile <abi_file_testing_sysfs_class_platform_profile>`
+for more information.
+
+The name of the platform-profile class device exported by this driver is
+"alienware-wmi" and it's path can be found with:
+
+::
+
+ grep -l "alienware-wmi" /sys/class/platform-profile/platform-profile-*/name | sed 's|/[^/]*$||'
+
+If the device supports G-Mode, it is also toggled when selecting the
+``performance`` profile.
+
+.. note::
+   You may set the ``force_gmode`` module parameter to always try to toggle this
+   feature, without checking if your model supports it.
+
+.. _hwmon:
+
+HWMON
+-----
+
+The AWCC interface also supports sensor monitoring and manual fan control. Both
+of these features are exposed to user-space through the HWMON interface.
+
+The name of the hwmon class device exported by this driver is "alienware_wmi"
+and it's path can be found with:
+
+::
+
+ grep -l "alienware_wmi" /sys/class/hwmon/hwmon*/name | sed 's|/[^/]*$||'
+
+Sensor monitoring is done through the standard HWMON interface. Refer to
+:ref:`sysfs-class-hwmon <abi_file_testing_sysfs_class_hwmon>` for more
+information.
+
+Manual fan control on the other hand, is not exposed directly by the AWCC
+interface. Instead it let's us control a fan `boost` value. This `boost` value
+has the following approximate behavior over the fan pwm:
+
+::
+
+ pwm = pwm_base + (fan_boost / 255) * (pwm_max - pwm_base)
+
+Due to the above behavior, the fan `boost` control is exposed to user-space
+through the following, custom hwmon sysfs attribute:
+
+=============================== ======= =======================================
+Name				Perm	Description
+=============================== ======= =======================================
+fan[1-4]_boost			RW	Fan boost value.
+
+					Integer value between 0 and 255
+=============================== ======= =======================================
+
+.. note::
+   In some devices, manual fan control only works reliably if the ``custom``
+   platform profile is selected.