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, 60 insertions, 0 deletions

diff --git a/Documentation/filesystems/bfs.rst b/Documentation/filesystems/bfs.rst
new file mode 100644
index 000000000..ce14b9018
--- /dev/null
+++ b/Documentation/filesystems/bfs.rst
@@ -0,0 +1,60 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+========================
+BFS Filesystem for Linux
+========================
+
+The BFS filesystem is used by SCO UnixWare OS for the /stand slice, which
+usually contains the kernel image and a few other files required for the
+boot process.
+
+In order to access /stand partition under Linux you obviously need to
+know the partition number and the kernel must support UnixWare disk slices
+(CONFIG_UNIXWARE_DISKLABEL config option). However BFS support does not
+depend on having UnixWare disklabel support because one can also mount
+BFS filesystem via loopback::
+
+    # losetup /dev/loop0 stand.img
+    # mount -t bfs /dev/loop0 /mnt/stand
+
+where stand.img is a file containing the image of BFS filesystem.
+When you have finished using it and umounted you need to also deallocate
+/dev/loop0 device by::
+
+    # losetup -d /dev/loop0
+
+You can simplify mounting by just typing::
+
+    # mount -t bfs -o loop stand.img /mnt/stand
+
+this will allocate the first available loopback device (and load loop.o
+kernel module if necessary) automatically. If the loopback driver is not
+loaded automatically, make sure that you have compiled the module and
+that modprobe is functioning. Beware that umount will not deallocate
+/dev/loopN device if /etc/mtab file on your system is a symbolic link to
+/proc/mounts. You will need to do it manually using "-d" switch of
+losetup(8). Read losetup(8) manpage for more info.
+
+To create the BFS image under UnixWare you need to find out first which
+slice contains it. The command prtvtoc(1M) is your friend::
+
+    # prtvtoc /dev/rdsk/c0b0t0d0s0
+
+(assuming your root disk is on target=0, lun=0, bus=0, controller=0). Then you
+look for the slice with tag "STAND", which is usually slice 10. With this
+information you can use dd(1) to create the BFS image::
+
+    # umount /stand
+    # dd if=/dev/rdsk/c0b0t0d0sa of=stand.img bs=512
+
+Just in case, you can verify that you have done the right thing by checking
+the magic number::
+
+    # od -Ad -tx4 stand.img | more
+
+The first 4 bytes should be 0x1badface.
+
+If you have any patches, questions or suggestions regarding this BFS
+implementation please contact the author:
+
+Tigran Aivazian <aivazian.tigran@gmail.com>