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, 76 insertions, 0 deletions

diff --git a/fs/stack.c b/fs/stack.c
new file mode 100644
index 000000000..d8c782e06
--- /dev/null
+++ b/fs/stack.c
@@ -0,0 +1,76 @@
+// SPDX-License-Identifier: GPL-2.0-only
+#include <linux/export.h>
+#include <linux/fs.h>
+#include <linux/fs_stack.h>
+
+/* does _NOT_ require i_rwsem to be held.
+ *
+ * This function cannot be inlined since i_size_{read,write} is rather
+ * heavy-weight on 32-bit systems
+ */
+void fsstack_copy_inode_size(struct inode *dst, struct inode *src)
+{
+	loff_t i_size;
+	blkcnt_t i_blocks;
+
+	/*
+	 * i_size_read() includes its own seqlocking and protection from
+	 * preemption (see include/linux/fs.h): we need nothing extra for
+	 * that here, and prefer to avoid nesting locks than attempt to keep
+	 * i_size and i_blocks in sync together.
+	 */
+	i_size = i_size_read(src);
+
+	/*
+	 * But on 32-bit, we ought to make an effort to keep the two halves of
+	 * i_blocks in sync despite SMP or PREEMPTION - though stat's
+	 * generic_fillattr() doesn't bother, and we won't be applying quotas
+	 * (where i_blocks does become important) at the upper level.
+	 *
+	 * We don't actually know what locking is used at the lower level;
+	 * but if it's a filesystem that supports quotas, it will be using
+	 * i_lock as in inode_add_bytes().
+	 */
+	if (sizeof(i_blocks) > sizeof(long))
+		spin_lock(&src->i_lock);
+	i_blocks = src->i_blocks;
+	if (sizeof(i_blocks) > sizeof(long))
+		spin_unlock(&src->i_lock);
+
+	/*
+	 * If CONFIG_SMP or CONFIG_PREEMPTION on 32-bit, it's vital for
+	 * fsstack_copy_inode_size() to hold some lock around
+	 * i_size_write(), otherwise i_size_read() may spin forever (see
+	 * include/linux/fs.h).  We don't necessarily hold i_rwsem when this
+	 * is called, so take i_lock for that case.
+	 *
+	 * And if on 32-bit, continue our effort to keep the two halves of
+	 * i_blocks in sync despite SMP or PREEMPTION: use i_lock for that case
+	 * too, and do both at once by combining the tests.
+	 *
+	 * There is none of this locking overhead in the 64-bit case.
+	 */
+	if (sizeof(i_size) > sizeof(long) || sizeof(i_blocks) > sizeof(long))
+		spin_lock(&dst->i_lock);
+	i_size_write(dst, i_size);
+	dst->i_blocks = i_blocks;
+	if (sizeof(i_size) > sizeof(long) || sizeof(i_blocks) > sizeof(long))
+		spin_unlock(&dst->i_lock);
+}
+EXPORT_SYMBOL_GPL(fsstack_copy_inode_size);
+
+/* copy all attributes */
+void fsstack_copy_attr_all(struct inode *dest, const struct inode *src)
+{
+	dest->i_mode = src->i_mode;
+	dest->i_uid = src->i_uid;
+	dest->i_gid = src->i_gid;
+	dest->i_rdev = src->i_rdev;
+	inode_set_atime_to_ts(dest, inode_get_atime(src));
+	inode_set_mtime_to_ts(dest, inode_get_mtime(src));
+	inode_set_ctime_to_ts(dest, inode_get_ctime(src));
+	dest->i_blkbits = src->i_blkbits;
+	dest->i_flags = src->i_flags;
+	set_nlink(dest, src->i_nlink);
+}
+EXPORT_SYMBOL_GPL(fsstack_copy_attr_all);