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/filesystems/ecryptfs.rst b/Documentation/filesystems/ecryptfs.rst
new file mode 100644
index 000000000..1f2edef4c
--- /dev/null
+++ b/Documentation/filesystems/ecryptfs.rst
@@ -0,0 +1,86 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+======================================================
+eCryptfs: A stacked cryptographic filesystem for Linux
+======================================================
+
+eCryptfs is free software. Please see the file COPYING for details.
+For documentation, please see the files in the doc/ subdirectory.  For
+building and installation instructions please see the INSTALL file.
+
+:Maintainer: Phillip Hellewell
+:Lead developer: Michael A. Halcrow <mhalcrow@us.ibm.com>
+:Developers: Michael C. Thompson
+             Kent Yoder
+:Web Site: http://ecryptfs.sf.net
+
+This software is currently undergoing development. Make sure to
+maintain a backup copy of any data you write into eCryptfs.
+
+eCryptfs requires the userspace tools downloadable from the
+SourceForge site:
+
+http://sourceforge.net/projects/ecryptfs/
+
+Userspace requirements include:
+
+- David Howells' userspace keyring headers and libraries (version
+  1.0 or higher), obtainable from
+  http://people.redhat.com/~dhowells/keyutils/
+- Libgcrypt
+
+
+.. note::
+
+   In the beta/experimental releases of eCryptfs, when you upgrade
+   eCryptfs, you should copy the files to an unencrypted location and
+   then copy the files back into the new eCryptfs mount to migrate the
+   files.
+
+
+Mount-wide Passphrase
+=====================
+
+Create a new directory into which eCryptfs will write its encrypted
+files (i.e., /root/crypt).  Then, create the mount point directory
+(i.e., /mnt/crypt).  Now it's time to mount eCryptfs::
+
+    mount -t ecryptfs /root/crypt /mnt/crypt
+
+You should be prompted for a passphrase and a salt (the salt may be
+blank).
+
+Try writing a new file::
+
+    echo "Hello, World" > /mnt/crypt/hello.txt
+
+The operation will complete.  Notice that there is a new file in
+/root/crypt that is at least 12288 bytes in size (depending on your
+host page size).  This is the encrypted underlying file for what you
+just wrote.  To test reading, from start to finish, you need to clear
+the user session keyring:
+
+keyctl clear @u
+
+Then umount /mnt/crypt and mount again per the instructions given
+above.
+
+::
+
+    cat /mnt/crypt/hello.txt
+
+
+Notes
+=====
+
+eCryptfs version 0.1 should only be mounted on (1) empty directories
+or (2) directories containing files only created by eCryptfs. If you
+mount a directory that has pre-existing files not created by eCryptfs,
+then behavior is undefined. Do not run eCryptfs in higher verbosity
+levels unless you are doing so for the sole purpose of debugging or
+development, since secret values will be written out to the system log
+in that case.
+
+
+Mike Halcrow
+mhalcrow@us.ibm.com