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, 180 insertions, 0 deletions

diff --git a/rust/zerocopy-derive/lib.rs b/rust/zerocopy-derive/lib.rs
new file mode 100644
index 000000000..d387de368
--- /dev/null
+++ b/rust/zerocopy-derive/lib.rs
@@ -0,0 +1,180 @@
+// SPDX-License-Identifier: (BSD-2-Clause OR Apache-2.0) OR MIT
+//
+// Copyright 2019 The Fuchsia Authors
+//
+// Licensed under a BSD-style license <LICENSE-BSD>, Apache License, Version 2.0
+// <LICENSE-APACHE or https://www.apache.org/licenses/LICENSE-2.0>, or the MIT
+// license <LICENSE-MIT or https://opensource.org/licenses/MIT>, at your option.
+// This file may not be copied, modified, or distributed except according to
+// those terms.
+
+//! Derive macros for [zerocopy]'s traits.
+//!
+//! [zerocopy]: https://docs.rs/zerocopy
+
+// Sometimes we want to use lints which were added after our MSRV.
+// `unknown_lints` is `warn` by default and we deny warnings in CI, so without
+// this attribute, any unknown lint would cause a CI failure when testing with
+// our MSRV.
+#![allow(unknown_lints)]
+#![deny(renamed_and_removed_lints)]
+#![deny(
+    clippy::all,
+    clippy::missing_safety_doc,
+    clippy::multiple_unsafe_ops_per_block,
+    clippy::undocumented_unsafe_blocks
+)]
+// We defer to own discretion on type complexity.
+#![allow(clippy::type_complexity)]
+// Inlining format args isn't supported on our MSRV.
+#![allow(clippy::uninlined_format_args)]
+#![deny(
+    rustdoc::bare_urls,
+    rustdoc::broken_intra_doc_links,
+    rustdoc::invalid_codeblock_attributes,
+    rustdoc::invalid_html_tags,
+    rustdoc::invalid_rust_codeblocks,
+    rustdoc::missing_crate_level_docs,
+    rustdoc::private_intra_doc_links
+)]
+#![recursion_limit = "128"]
+
+macro_rules! ident {
+    (($fmt:literal $(, $arg:expr)*), $span:expr) => {
+        syn::Ident::new(&format!($fmt $(, crate::util::to_ident_str($arg))*), $span)
+    };
+}
+
+mod derive;
+#[cfg(test)]
+mod output_tests;
+mod repr;
+mod util;
+
+use syn::{DeriveInput, Error};
+
+use crate::util::*;
+
+// FIXME(https://github.com/rust-lang/rust/issues/54140): Some errors could be
+// made better if we could add multiple lines of error output like this:
+//
+// error: unsupported representation
+//   --> enum.rs:28:8
+//    |
+// 28 | #[repr(transparent)]
+//    |
+// help: required by the derive of FromBytes
+//
+// Instead, we have more verbose error messages like "unsupported representation
+// for deriving FromZeros, FromBytes, IntoBytes, or Unaligned on an enum"
+//
+// This will probably require Span::error
+// (https://doc.rust-lang.org/nightly/proc_macro/struct.Span.html#method.error),
+// which is currently unstable. Revisit this once it's stable.
+
+/// Defines a derive function named `$outer` which parses its input
+/// `TokenStream` as a `DeriveInput` and then invokes the `$inner` function.
+///
+/// Note that the separate `$outer` parameter is required - proc macro functions
+/// are currently required to live at the crate root, and so the caller must
+/// specify the name in order to avoid name collisions.
+macro_rules! derive {
+    ($trait:ident => $outer:ident => $inner:path) => {
+        #[proc_macro_derive($trait, attributes(zerocopy))]
+        pub fn $outer(ts: proc_macro::TokenStream) -> proc_macro::TokenStream {
+            let ast = syn::parse_macro_input!(ts as DeriveInput);
+            let ctx = match Ctx::try_from_derive_input(ast) {
+                Ok(ctx) => ctx,
+                Err(e) => return e.into_compile_error().into(),
+            };
+            let ts = $inner(&ctx, Trait::$trait).into_ts();
+            // We wrap in `const_block` as a backstop in case any derive fails
+            // to wrap its output in `const_block` (and thus fails to annotate)
+            // with the full set of `#[allow(...)]` attributes).
+            let ts = const_block([Some(ts)]);
+            #[cfg(test)]
+            crate::util::testutil::check_hygiene(ts.clone());
+            ts.into()
+        }
+    };
+}
+
+trait IntoTokenStream {
+    fn into_ts(self) -> proc_macro2::TokenStream;
+}
+
+impl IntoTokenStream for proc_macro2::TokenStream {
+    fn into_ts(self) -> proc_macro2::TokenStream {
+        self
+    }
+}
+
+impl IntoTokenStream for Result<proc_macro2::TokenStream, Error> {
+    fn into_ts(self) -> proc_macro2::TokenStream {
+        match self {
+            Ok(ts) => ts,
+            Err(err) => err.to_compile_error(),
+        }
+    }
+}
+
+derive!(KnownLayout => derive_known_layout => crate::derive::known_layout::derive);
+derive!(Immutable => derive_immutable => crate::derive::derive_immutable);
+derive!(TryFromBytes => derive_try_from_bytes => crate::derive::try_from_bytes::derive_try_from_bytes);
+derive!(FromZeros => derive_from_zeros => crate::derive::from_bytes::derive_from_zeros);
+derive!(FromBytes => derive_from_bytes => crate::derive::from_bytes::derive_from_bytes);
+derive!(IntoBytes => derive_into_bytes => crate::derive::into_bytes::derive_into_bytes);
+derive!(Unaligned => derive_unaligned => crate::derive::unaligned::derive_unaligned);
+derive!(ByteHash => derive_hash => crate::derive::derive_hash);
+derive!(ByteEq => derive_eq => crate::derive::derive_eq);
+derive!(SplitAt => derive_split_at => crate::derive::derive_split_at);
+
+#[cfg_attr(not(zerocopy_unstable_linux), doc(hidden))]
+#[proc_macro_derive(most_traits, attributes(zerocopy))]
+pub fn most_traits(ts: proc_macro::TokenStream) -> proc_macro::TokenStream {
+    let ast = syn::parse_macro_input!(ts as DeriveInput);
+    let ctx = match Ctx::try_from_derive_input(ast) {
+        Ok(ctx) => ctx,
+        Err(e) => return e.into_compile_error().into(),
+    }
+    .skip_on_error();
+
+    // top-level traits for which to attempt a derive
+    let derives: [(fn(&Ctx, Trait) -> _, _); 6] = [
+        (crate::derive::known_layout::derive, Trait::KnownLayout),
+        (crate::derive::derive_immutable, Trait::Immutable),
+        (crate::derive::from_bytes::derive_from_bytes, Trait::FromBytes),
+        (crate::derive::into_bytes::derive_into_bytes, Trait::IntoBytes),
+        (crate::derive::derive_split_at, Trait::SplitAt),
+        (crate::derive::unaligned::derive_unaligned, Trait::Unaligned),
+    ];
+
+    let mut tokens = proc_macro2::TokenStream::new();
+    for (derive, t) in derives {
+        tokens.extend(derive(&ctx, t))
+    }
+
+    // We wrap in `const_block` as a backstop in case any derive fails
+    // to wrap its output in `const_block` (and thus fails to annotate)
+    // with the full set of `#[allow(...)]` attributes).
+    let ts = const_block([Some(tokens)]);
+    #[cfg(test)]
+    crate::util::testutil::check_hygiene(ts.clone());
+    ts.into()
+}
+
+/// Deprecated: prefer [`FromZeros`] instead.
+#[deprecated(since = "0.8.0", note = "`FromZeroes` was renamed to `FromZeros`")]
+#[doc(hidden)]
+#[proc_macro_derive(FromZeroes)]
+pub fn derive_from_zeroes(ts: proc_macro::TokenStream) -> proc_macro::TokenStream {
+    derive_from_zeros(ts)
+}
+
+/// Deprecated: prefer [`IntoBytes`] instead.
+#[deprecated(since = "0.8.0", note = "`AsBytes` was renamed to `IntoBytes`")]
+#[doc(hidden)]
+#[proc_macro_derive(AsBytes)]
+pub fn derive_as_bytes(ts: proc_macro::TokenStream) -> proc_macro::TokenStream {
+    derive_into_bytes(ts)
+}