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, 82 insertions, 0 deletions

diff --git a/Documentation/i2c/summary.rst b/Documentation/i2c/summary.rst
new file mode 100644
index 000000000..579a1c7df
--- /dev/null
+++ b/Documentation/i2c/summary.rst
@@ -0,0 +1,82 @@
+=============================
+Introduction to I2C and SMBus
+=============================
+
+I²C (pronounce: I squared C and written I2C in the kernel documentation) is
+a protocol developed by Philips. It is a two-wire protocol with variable
+speed (typically up to 400 kHz, high speed modes up to 5 MHz). It provides
+an inexpensive bus for connecting many types of devices with infrequent or
+low bandwidth communications needs. I2C is widely used with embedded
+systems. Some systems use variants that don't meet branding requirements,
+and so are not advertised as being I2C but come under different names,
+e.g. TWI (Two Wire Interface), IIC.
+
+The latest official I2C specification is the `"I²C-bus specification and user
+manual" (UM10204) <https://www.nxp.com/docs/en/user-guide/UM10204.pdf>`_
+published by NXP Semiconductors, version 7 as of this writing.
+
+SMBus (System Management Bus) is based on the I2C protocol, and is mostly
+a subset of I2C protocols and signaling. Many I2C devices will work on an
+SMBus, but some SMBus protocols add semantics beyond what is required to
+achieve I2C branding. Modern PC mainboards rely on SMBus. The most common
+devices connected through SMBus are RAM modules configured using I2C EEPROMs,
+and hardware monitoring chips.
+
+Because the SMBus is mostly a subset of the generalized I2C bus, we can
+use its protocols on many I2C systems. However, there are systems that don't
+meet both SMBus and I2C electrical constraints; and others which can't
+implement all the common SMBus protocol semantics or messages.
+
+
+Terminology
+===========
+
+The I2C bus connects one or more controller chips and one or more target chips.
+
+.. kernel-figure::  i2c_bus.svg
+   :alt:    Simple I2C bus with one controller and 3 targets
+
+   Simple I2C bus
+
+A **controller** chip is a node that starts communications with targets. In the
+Linux kernel implementation it is also called an "adapter" or "bus". Controller
+drivers are usually in the ``drivers/i2c/busses/`` subdirectory.
+
+An **algorithm** contains general code that can be used to implement a whole
+class of I2C controllers. Each specific controller driver either depends on an
+algorithm driver in the ``drivers/i2c/algos/`` subdirectory, or includes its
+own implementation.
+
+A **target** chip is a node that responds to communications when addressed by a
+controller. In the Linux kernel implementation it is also called a "client".
+While targets are usually separate external chips, Linux can also act as a
+target (needs hardware support) and respond to another controller on the bus.
+This is then called a **local target**. In contrast, an external chip is called
+a **remote target**.
+
+Target drivers are kept in a directory specific to the feature they provide,
+for example ``drivers/gpio/`` for GPIO expanders and ``drivers/media/i2c/`` for
+video-related chips.
+
+For the example configuration in the figure above, you will need one driver for
+the I2C controller, and drivers for your I2C targets. Usually one driver for
+each target.
+
+Synonyms
+--------
+
+As mentioned above, the Linux I2C implementation historically uses the terms
+"adapter" for controller and "client" for target. A number of data structures
+have these synonyms in their name. So, when discussing implementation details,
+you should be aware of these terms as well. The official wording is preferred,
+though.
+
+Outdated terminology
+--------------------
+
+In earlier I2C specifications, controller was named "master" and target was
+named "slave". These terms have been obsoleted with v7 of the specification and
+their use is also discouraged by the Linux Kernel Code of Conduct. You may
+still find them in references to documentation which has not been updated. The
+general attitude, however, is to use the inclusive terms: controller and
+target. Work to replace the old terminology in the Linux Kernel is on-going.