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, 73 insertions, 0 deletions

diff --git a/rust/bindgen_parameters b/rust/bindgen_parameters
new file mode 100644
index 000000000..8402b0c93
--- /dev/null
+++ b/rust/bindgen_parameters
@@ -0,0 +1,73 @@
+# SPDX-License-Identifier: GPL-2.0
+
+# We want to map these types to `isize`/`usize` manually, instead of
+# define them as `int`/`long` depending on platform bitwidth.
+--blocklist-type __kernel_s?size_t
+--blocklist-type __kernel_ptrdiff_t
+
+--opaque-type xregs_state
+--opaque-type desc_struct
+--opaque-type arch_lbr_state
+--opaque-type local_apic
+
+# Packed type cannot transitively contain a `#[repr(align)]` type.
+--opaque-type alt_instr
+--opaque-type x86_msi_data
+--opaque-type x86_msi_addr_lo
+# s390-only: same packed/align issue as above (E0588).
+--opaque-type lowcore
+--opaque-type tod_clock
+--opaque-type tpi_info
+--opaque-type uv_cb.*
+--opaque-type uv_secret.*
+--opaque-type zpci_fib
+
+# If SMP is disabled, `arch_spinlock_t` is defined as a ZST which triggers a Rust
+# warning. We don't need to peek into it anyway.
+--opaque-type spinlock
+
+# enums that appear in indirect function calls should specify a cfi type
+--newtype-enum lru_status
+--with-attribute-custom-enum=lru_status='#[cfi_encoding="10lru_status"]'
+
+# `seccomp`'s comment gets understood as a doctest
+--no-doc-comments
+
+# These functions use the `__preserve_most` calling convention, which neither bindgen
+# nor Rust currently understand, and which Clang currently declares to be unstable.
+--blocklist-function __list_.*_report
+
+# These constants are sometimes not recognized by bindgen depending on config.
+# We use const helpers to aid bindgen, to avoid conflicts when constants are
+# recognized, block generation of the non-helper constants.
+--blocklist-item ARCH_SLAB_MINALIGN
+--blocklist-item ARCH_KMALLOC_MINALIGN
+--blocklist-item VM_MERGEABLE
+--blocklist-item VM_READ
+--blocklist-item VM_WRITE
+--blocklist-item VM_EXEC
+--blocklist-item VM_SHARED
+--blocklist-item VM_MAYREAD
+--blocklist-item VM_MAYWRITE
+--blocklist-item VM_MAYEXEC
+--blocklist-item VM_MAYEXEC
+--blocklist-item VM_PFNMAP
+--blocklist-item VM_IO
+--blocklist-item VM_DONTCOPY
+--blocklist-item VM_DONTEXPAND
+--blocklist-item VM_LOCKONFAULT
+--blocklist-item VM_ACCOUNT
+--blocklist-item VM_NORESERVE
+--blocklist-item VM_HUGETLB
+--blocklist-item VM_SYNC
+--blocklist-item VM_ARCH_1
+--blocklist-item VM_WIPEONFORK
+--blocklist-item VM_DONTDUMP
+--blocklist-item VM_SOFTDIRTY
+--blocklist-item VM_MIXEDMAP
+--blocklist-item VM_HUGEPAGE
+--blocklist-item VM_NOHUGEPAGE
+
+# Structs should implement `Zeroable` when all of their fields do.
+--with-derive-custom-struct .*=MaybeZeroable
+--with-derive-custom-union .*=MaybeZeroable