From a5f7a5bb3b7f28ba7e4fa246775b29a0e5537255 Mon Sep 17 00:00:00 2001 From: Jinke Han Date: Tue, 8 Sep 2026 15:37:42 +0800 Subject: x86/kprobes: Fix crash when probing CS CALL instructions 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) Signed-off-by: Jinke Han Signed-off-by: Ingo Molnar Reviewed-by: Masami Hiramatsu (Google) Acked-by: Yafang Shao Acked-by: Borislav Petkov Cc: Peter Zijlstra Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000 --- Documentation/rust/general-information.rst | 161 +++++++++++++++++++++++++++++ 1 file changed, 161 insertions(+) create mode 100644 Documentation/rust/general-information.rst (limited to 'Documentation/rust/general-information.rst') diff --git a/Documentation/rust/general-information.rst b/Documentation/rust/general-information.rst new file mode 100644 index 000000000..09234bed2 --- /dev/null +++ b/Documentation/rust/general-information.rst @@ -0,0 +1,161 @@ +.. SPDX-License-Identifier: GPL-2.0 + +General Information +=================== + +This document contains useful information to know when working with +the Rust support in the kernel. + + +``no_std`` +---------- + +The Rust support in the kernel can link only `core `_, +but not `std `_. Crates for use in the +kernel must opt into this behavior using the ``#![no_std]`` attribute. + + +.. _rust_code_documentation: + +Code documentation +------------------ + +Rust kernel code is documented using ``rustdoc``, its built-in documentation +generator. + +The generated HTML docs include integrated search, linked items (e.g. types, +functions, constants), source code, etc. They may be read at: + + https://rust.docs.kernel.org + +For linux-next, please see: + + https://rust.docs.kernel.org/next/ + +There are also tags for each main release, e.g.: + + https://rust.docs.kernel.org/6.10/ + +The docs can also be easily generated and read locally. This is quite fast +(same order as compiling the code itself) and no special tools or environment +are needed. This has the added advantage that they will be tailored to +the particular kernel configuration used. To generate them, use the ``rustdoc`` +target with the same invocation used for compilation, e.g.:: + + make LLVM=1 rustdoc + +To read the docs locally in your web browser, run e.g.:: + + xdg-open Documentation/output/rust/rustdoc/kernel/index.html + +To learn about how to write the documentation, please see coding-guidelines.rst. + + +Extra lints +----------- + +While ``rustc`` is a very helpful compiler, some extra lints and analyses are +available via ``clippy``, a Rust linter. To enable it, pass ``CLIPPY=1`` to +the same invocation used for compilation, e.g.:: + + make LLVM=1 CLIPPY=1 + +Please note that Clippy may change code generation, thus it should not be +enabled while building a production kernel. + + +Abstractions vs. bindings +------------------------- + +Abstractions are Rust code wrapping kernel functionality from the C side. + +In order to use functions and types from the C side, bindings are created. +Bindings are the declarations for Rust of those functions and types from +the C side. + +For instance, one may write a ``Mutex`` abstraction in Rust which wraps +a ``struct mutex`` from the C side and calls its functions through the bindings. + +Abstractions are not available for all the kernel internal APIs and concepts, +but it is intended that coverage is expanded as time goes on. "Leaf" modules +(e.g. drivers) should not use the C bindings directly. Instead, subsystems +should provide as-safe-as-possible abstractions as needed. + +.. code-block:: + + rust/bindings/ + (rust/helpers/) + + include/ -----+ <-+ + | | + drivers/ rust/kernel/ +----------+ <-+ | + fs/ | bindgen | | + .../ +-------------------+ +----------+ --+ | + | Abstractions | | | + +---------+ | +------+ +------+ | +----------+ | | + | my_foo | -----> | | foo | | bar | | -------> | Bindings | <-+ | + | driver | Safe | | sub- | | sub- | | Unsafe | | | + +---------+ | |system| |system| | | bindings | <-----+ + | | +------+ +------+ | | crate | | + | | kernel crate | +----------+ | + | +-------------------+ | + | | + +------------------# FORBIDDEN #--------------------------------+ + +The main idea is to encapsulate all direct interaction with the kernel's C APIs +into carefully reviewed and documented abstractions. Then users of these +abstractions cannot introduce undefined behavior (UB) as long as: + +#. The abstractions are correct ("sound"). +#. Any ``unsafe`` blocks respect the safety contract necessary to call the + operations inside the block. Similarly, any ``unsafe impl``\ s respect the + safety contract necessary to implement the trait. + +Bindings +~~~~~~~~ + +By including a C header from ``include/`` into +``rust/bindings/bindings_helper.h``, the ``bindgen`` tool will auto-generate the +bindings for the included subsystem. After building, see the ``*_generated.rs`` +output files in the ``rust/bindings/`` directory. + +For parts of the C header that ``bindgen`` does not auto generate, e.g. C +``inline`` functions or non-trivial macros, it is acceptable to add a small +wrapper function to ``rust/helpers/`` to make it available for the Rust side as +well. + +Abstractions +~~~~~~~~~~~~ + +Abstractions are the layer between the bindings and the in-kernel users. They +are located in ``rust/kernel/`` and their role is to encapsulate the unsafe +access to the bindings into an as-safe-as-possible API that they expose to their +users. Users of the abstractions include things like drivers or file systems +written in Rust. + +Besides the safety aspect, the abstractions are supposed to be "ergonomic", in +the sense that they turn the C interfaces into "idiomatic" Rust code. Basic +examples are to turn the C resource acquisition and release into Rust +constructors and destructors or C integer error codes into Rust's ``Result``\ s. + + +Conditional compilation +----------------------- + +Rust code has access to conditional compilation based on the kernel +configuration: + +.. code-block:: rust + + #[cfg(CONFIG_X)] // Enabled (`y` or `m`) + #[cfg(CONFIG_X="y")] // Enabled as a built-in (`y`) + #[cfg(CONFIG_X="m")] // Enabled as a module (`m`) + #[cfg(not(CONFIG_X))] // Disabled + +For other predicates that Rust's ``cfg`` does not support, e.g. expressions with +numerical comparisons, one may define a new Kconfig symbol: + +.. code-block:: kconfig + + config RUSTC_HAS_SPAN_FILE + def_bool RUSTC_VERSION >= 108800 -- cgit v1.3.1