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, 92 insertions, 0 deletions

diff --git a/Documentation/admin-guide/media/omap3isp.rst b/Documentation/admin-guide/media/omap3isp.rst
new file mode 100644
index 000000000..f32e7375a
--- /dev/null
+++ b/Documentation/admin-guide/media/omap3isp.rst
@@ -0,0 +1,92 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+.. include:: <isonum.txt>
+
+OMAP 3 Image Signal Processor (ISP) driver
+==========================================
+
+Copyright |copy| 2010 Nokia Corporation
+
+Copyright |copy| 2009 Texas Instruments, Inc.
+
+Contacts: Laurent Pinchart <laurent.pinchart@ideasonboard.com>,
+Sakari Ailus <sakari.ailus@iki.fi>, David Cohen <dacohen@gmail.com>
+
+
+Introduction
+------------
+
+This file documents the Texas Instruments OMAP 3 Image Signal Processor (ISP)
+driver located under drivers/media/platform/ti/omap3isp. The original driver was
+written by Texas Instruments but since that it has been rewritten (twice) at
+Nokia.
+
+The driver has been successfully used on the following versions of OMAP 3:
+
+- 3430
+- 3530
+- 3630
+
+The driver implements V4L2, Media controller and v4l2_subdev interfaces.
+Sensor, lens and flash drivers using the v4l2_subdev interface in the kernel
+are supported.
+
+
+Split to subdevs
+----------------
+
+The OMAP 3 ISP is split into V4L2 subdevs, each of the blocks inside the ISP
+having one subdev to represent it. Each of the subdevs provide a V4L2 subdev
+interface to userspace.
+
+- OMAP3 ISP CCP2
+- OMAP3 ISP CSI2a
+- OMAP3 ISP CCDC
+- OMAP3 ISP preview
+- OMAP3 ISP resizer
+- OMAP3 ISP AEWB
+- OMAP3 ISP AF
+- OMAP3 ISP histogram
+
+Each possible link in the ISP is modelled by a link in the Media controller
+interface. For an example program see [#]_.
+
+
+Controlling the OMAP 3 ISP
+--------------------------
+
+In general, the settings given to the OMAP 3 ISP take effect at the beginning
+of the following frame. This is done when the module becomes idle during the
+vertical blanking period on the sensor. In memory-to-memory operation the pipe
+is run one frame at a time. Applying the settings is done between the frames.
+
+All the blocks in the ISP, excluding the CSI-2 and possibly the CCP2 receiver,
+insist on receiving complete frames. Sensors must thus never send the ISP
+partial frames.
+
+Autoidle does have issues with some ISP blocks on the 3430, at least.
+Autoidle is only enabled on 3630 when the omap3isp module parameter autoidle
+is non-zero.
+
+Technical reference manuals (TRMs) and other documentation
+----------------------------------------------------------
+
+OMAP 3430 TRM:
+<URL:http://focus.ti.com/pdfs/wtbu/OMAP34xx_ES3.1.x_PUBLIC_TRM_vZM.zip>
+Referenced 2011-03-05.
+
+OMAP 35xx TRM:
+<URL:http://www.ti.com/litv/pdf/spruf98o> Referenced 2011-03-05.
+
+OMAP 3630 TRM:
+<URL:http://focus.ti.com/pdfs/wtbu/OMAP36xx_ES1.x_PUBLIC_TRM_vQ.zip>
+Referenced 2011-03-05.
+
+DM 3730 TRM:
+<URL:http://www.ti.com/litv/pdf/sprugn4h> Referenced 2011-03-06.
+
+
+References
+----------
+
+.. [#] http://git.ideasonboard.org/?p=media-ctl.git;a=summary