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, 59 insertions, 0 deletions

diff --git a/Documentation/arch/riscv/patch-acceptance.rst b/Documentation/arch/riscv/patch-acceptance.rst
new file mode 100644
index 000000000..634aa222b
--- /dev/null
+++ b/Documentation/arch/riscv/patch-acceptance.rst
@@ -0,0 +1,59 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+arch/riscv maintenance guidelines for developers
+================================================
+
+Overview
+--------
+The RISC-V instruction set architecture is developed in the open:
+in-progress drafts are available for all to review and to experiment
+with implementations.  New module or extension drafts can change
+during the development process - sometimes in ways that are
+incompatible with previous drafts.  This flexibility can present a
+challenge for RISC-V Linux maintenance.  Linux maintainers disapprove
+of churn, and the Linux development process prefers well-reviewed and
+tested code over experimental code.  We wish to extend these same
+principles to the RISC-V-related code that will be accepted for
+inclusion in the kernel.
+
+Patchwork
+---------
+
+RISC-V has a patchwork instance, where the status of patches can be checked:
+
+  https://patchwork.kernel.org/project/linux-riscv/list/
+
+If your patch does not appear in the default view, the RISC-V maintainers have
+likely either requested changes, or expect it to be applied to another tree.
+
+Automation runs against this patchwork instance, building/testing patches as
+they arrive. The automation applies patches against the current HEAD of the
+RISC-V `for-next` and `fixes` branches, depending on whether the patch has been
+detected as a fix. Failing those, it will use the RISC-V `master` branch.
+The exact commit to which a series has been applied will be noted on patchwork.
+Patches for which any of the checks fail are unlikely to be applied and in most
+cases will need to be resubmitted.
+
+Submit Checklist Addendum
+-------------------------
+We'll only accept patches for new modules or extensions if the
+specifications for those modules or extensions are listed as being
+unlikely to be incompatibly changed in the future.  For
+specifications from the RISC-V foundation this means "Frozen" or
+"Ratified", for the UEFI forum specifications this means a published
+ECR.  (Developers may, of course, maintain their own Linux kernel trees
+that contain code for any draft extensions that they wish.)
+
+Additionally, the RISC-V specification allows implementers to create
+their own custom extensions.  These custom extensions aren't required
+to go through any review or ratification process by the RISC-V
+Foundation.  To avoid the maintenance complexity and potential
+performance impact of adding kernel code for implementor-specific
+RISC-V extensions, we'll only consider patches for extensions that either:
+
+- Have been officially frozen or ratified by the RISC-V Foundation, or
+- Have been implemented in hardware that is widely available, per standard
+  Linux practice.
+
+(Implementers, may, of course, maintain their own Linux kernel trees containing
+code for any custom extensions that they wish.)