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, 56 insertions, 0 deletions

diff --git a/Documentation/driver-api/pldmfw/driver-ops.rst b/Documentation/driver-api/pldmfw/driver-ops.rst
new file mode 100644
index 000000000..f0654783d
--- /dev/null
+++ b/Documentation/driver-api/pldmfw/driver-ops.rst
@@ -0,0 +1,56 @@
+.. SPDX-License-Identifier: GPL-2.0-only
+
+=========================
+Driver-specific callbacks
+=========================
+
+The ``pldmfw`` module relies on the device driver for implementing device
+specific behavior using the following operations.
+
+``.match_record``
+-----------------
+
+The ``.match_record`` operation is used to determine whether a given PLDM
+record matches the device being updated. This requires comparing the record
+descriptors in the record with information from the device. Many record
+descriptors are defined by the PLDM standard, but it is also allowed for
+devices to implement their own descriptors.
+
+The ``.match_record`` operation should return true if a given record matches
+the device.
+
+``.send_package_data``
+----------------------
+
+The ``.send_package_data`` operation is used to send the device-specific
+package data in a record to the device firmware. If the matching record
+provides package data, ``pldmfw`` will call the ``.send_package_data``
+function with a pointer to the package data and with the package data
+length. The device driver should send this data to firmware.
+
+``.send_component_table``
+-------------------------
+
+The ``.send_component_table`` operation is used to forward component
+information to the device. It is called once for each applicable component,
+that is, for each component indicated by the matching record. The
+device driver should send the component information to the device firmware,
+and wait for a response. The provided transfer flag indicates whether this
+is the first, last, or a middle component, and is expected to be forwarded
+to firmware as part of the component table information. The driver should an
+error in the case when the firmware indicates that the component cannot be
+updated, or return zero if the component can be updated.
+
+``.flash_component``
+--------------------
+
+The ``.flash_component`` operation is used to inform the device driver to
+flash a given component. The driver must perform any steps necessary to send
+the component data to the device.
+
+``.finalize_update``
+--------------------
+
+The ``.finalize_update`` operation is used by the ``pldmfw`` library in
+order to allow the device driver to perform any remaining device specific
+logic needed to finish the update.