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, 84 insertions, 0 deletions

diff --git a/Documentation/bpf/linux-notes.rst b/Documentation/bpf/linux-notes.rst
new file mode 100644
index 000000000..00d2693de
--- /dev/null
+++ b/Documentation/bpf/linux-notes.rst
@@ -0,0 +1,84 @@
+.. contents::
+.. sectnum::
+
+==========================
+Linux implementation notes
+==========================
+
+This document provides more details specific to the Linux kernel implementation of the eBPF instruction set.
+
+Byte swap instructions
+======================
+
+``BPF_FROM_LE`` and ``BPF_FROM_BE`` exist as aliases for ``BPF_TO_LE`` and ``BPF_TO_BE`` respectively.
+
+Jump instructions
+=================
+
+``BPF_CALL | BPF_X | BPF_JMP`` (0x8d), where the helper function
+integer would be read from a specified register, is not currently supported
+by the verifier.  Any programs with this instruction will fail to load
+until such support is added.
+
+Maps
+====
+
+Linux only supports the 'map_val(map)' operation on array maps with a single element.
+
+Linux uses an fd_array to store maps associated with a BPF program. Thus,
+map_by_idx(imm) uses the fd at that index in the array.
+
+Variables
+=========
+
+The following 64-bit immediate instruction specifies that a variable address,
+which corresponds to some integer stored in the 'imm' field, should be loaded:
+
+=========================  ======  ===  =========================================  ===========  ==============
+opcode construction        opcode  src  pseudocode                                 imm type     dst type
+=========================  ======  ===  =========================================  ===========  ==============
+BPF_IMM | BPF_DW | BPF_LD  0x18    0x3  dst = var_addr(imm)                        variable id  data pointer
+=========================  ======  ===  =========================================  ===========  ==============
+
+On Linux, this integer is a BTF ID.
+
+Legacy BPF Packet access instructions
+=====================================
+
+As mentioned in the `ISA standard documentation
+<instruction-set.html#legacy-bpf-packet-access-instructions>`_,
+Linux has special eBPF instructions for access to packet data that have been
+carried over from classic BPF to retain the performance of legacy socket
+filters running in the eBPF interpreter.
+
+The instructions come in two forms: ``BPF_ABS | <size> | BPF_LD`` and
+``BPF_IND | <size> | BPF_LD``.
+
+These instructions are used to access packet data and can only be used when
+the program context is a pointer to a networking packet.  ``BPF_ABS``
+accesses packet data at an absolute offset specified by the immediate data
+and ``BPF_IND`` access packet data at an offset that includes the value of
+a register in addition to the immediate data.
+
+These instructions have seven implicit operands:
+
+* Register R6 is an implicit input that must contain a pointer to a
+  struct sk_buff.
+* Register R0 is an implicit output which contains the data fetched from
+  the packet.
+* Registers R1-R5 are scratch registers that are clobbered by the
+  instruction.
+
+These instructions have an implicit program exit condition as well. If an
+eBPF program attempts access data beyond the packet boundary, the
+program execution will be aborted.
+
+``BPF_ABS | BPF_W | BPF_LD`` (0x20) means::
+
+  R0 = ntohl(*(u32 *) ((struct sk_buff *) R6->data + imm))
+
+where ``ntohl()`` converts a 32-bit value from network byte order to host byte order.
+
+``BPF_IND | BPF_W | BPF_LD`` (0x40) means::
+
+  R0 = ntohl(*(u32 *) ((struct sk_buff *) R6->data + src + imm))