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, 86 insertions, 0 deletions

diff --git a/Documentation/userspace-api/futex2.rst b/Documentation/userspace-api/futex2.rst
new file mode 100644
index 000000000..9693f47a7
--- /dev/null
+++ b/Documentation/userspace-api/futex2.rst
@@ -0,0 +1,86 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+======
+futex2
+======
+
+:Author: André Almeida <andrealmeid@collabora.com>
+
+futex, or fast user mutex, is a set of syscalls to allow userspace to create
+performant synchronization mechanisms, such as mutexes, semaphores and
+conditional variables in userspace. C standard libraries, like glibc, uses it
+as a means to implement more high level interfaces like pthreads.
+
+futex2 is a followup version of the initial futex syscall, designed to overcome
+limitations of the original interface.
+
+User API
+========
+
+``futex_waitv()``
+-----------------
+
+Wait on an array of futexes, wake on any::
+
+  futex_waitv(struct futex_waitv *waiters, unsigned int nr_futexes,
+              unsigned int flags, struct timespec *timeout, clockid_t clockid)
+
+  struct futex_waitv {
+        __u64 val;
+        __u64 uaddr;
+        __u32 flags;
+        __u32 __reserved;
+  };
+
+Userspace sets an array of struct futex_waitv (up to a max of 128 entries),
+using ``uaddr`` for the address to wait for, ``val`` for the expected value
+and ``flags`` to specify the type (e.g. private) and size of futex.
+``__reserved`` needs to be 0, but it can be used for future extension. The
+pointer for the first item of the array is passed as ``waiters``. An invalid
+address for ``waiters`` or for any ``uaddr`` returns ``-EFAULT``.
+
+If userspace has 32-bit pointers, it should do a explicit cast to make sure
+the upper bits are zeroed. ``uintptr_t`` does the tricky and it works for
+both 32/64-bit pointers.
+
+``nr_futexes`` specifies the size of the array. Numbers out of [1, 128]
+interval will make the syscall return ``-EINVAL``.
+
+The ``flags`` argument of the syscall needs to be 0, but it can be used for
+future extension.
+
+For each entry in ``waiters`` array, the current value at ``uaddr`` is compared
+to ``val``. If it's different, the syscall undo all the work done so far and
+return ``-EAGAIN``. If all tests and verifications succeeds, syscall waits until
+one of the following happens:
+
+- The timeout expires, returning ``-ETIMEOUT``.
+- A signal was sent to the sleeping task, returning ``-ERESTARTSYS``.
+- Some futex at the list was woken, returning the index of some waked futex.
+
+An example of how to use the interface can be found at ``tools/testing/selftests/futex/functional/futex_waitv.c``.
+
+Timeout
+-------
+
+``struct timespec *timeout`` argument is an optional argument that points to an
+absolute timeout. You need to specify the type of clock being used at
+``clockid`` argument. ``CLOCK_MONOTONIC`` and ``CLOCK_REALTIME`` are supported.
+This syscall accepts only 64bit timespec structs.
+
+Types of futex
+--------------
+
+A futex can be either private or shared. Private is used for processes that
+shares the same memory space and the virtual address of the futex will be the
+same for all processes. This allows for optimizations in the kernel. To use
+private futexes, it's necessary to specify ``FUTEX_PRIVATE_FLAG`` in the futex
+flag. For processes that doesn't share the same memory space and therefore can
+have different virtual addresses for the same futex (using, for instance, a
+file-backed shared memory) requires different internal mechanisms to be get
+properly enqueued. This is the default behavior, and it works with both private
+and shared futexes.
+
+Futexes can be of different sizes: 8, 16, 32 or 64 bits. Currently, the only
+supported one is 32 bit sized futex, and it need to be specified using
+``FUTEX_32`` flag.