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, 155 insertions, 0 deletions

diff --git a/io_uring/query.c b/io_uring/query.c
new file mode 100644
index 000000000..7bce2ec16
--- /dev/null
+++ b/io_uring/query.c
@@ -0,0 +1,155 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include "linux/io_uring/query.h"
+
+#include "query.h"
+#include "io_uring.h"
+#include "zcrx.h"
+
+union io_query_data {
+	struct io_uring_query_opcode opcodes;
+	struct io_uring_query_zcrx zcrx;
+	struct io_uring_query_zcrx_event zcrx_notif;
+	struct io_uring_query_scq scq;
+};
+
+#define IO_MAX_QUERY_SIZE		sizeof(union io_query_data)
+#define IO_MAX_QUERY_ENTRIES		1000
+
+static ssize_t io_query_ops(union io_query_data *data)
+{
+	struct io_uring_query_opcode *e = &data->opcodes;
+
+	e->nr_request_opcodes = IORING_OP_LAST;
+	e->nr_register_opcodes = IORING_REGISTER_LAST;
+	e->feature_flags = IORING_FEAT_FLAGS;
+	e->ring_setup_flags = IORING_SETUP_FLAGS;
+	e->enter_flags = IORING_ENTER_FLAGS;
+	e->sqe_flags = SQE_VALID_FLAGS;
+	e->nr_query_opcodes = __IO_URING_QUERY_MAX;
+	e->__pad = 0;
+	return sizeof(*e);
+}
+
+static ssize_t io_query_zcrx(union io_query_data *data)
+{
+	struct io_uring_query_zcrx *e = &data->zcrx;
+
+	e->register_flags = ZCRX_SUPPORTED_REG_FLAGS;
+	e->area_flags = IORING_ZCRX_AREA_DMABUF;
+	e->nr_ctrl_opcodes = __ZCRX_CTRL_LAST;
+	e->rq_hdr_size = sizeof(struct zcrx_rq_hdr);
+	e->rq_hdr_alignment = L1_CACHE_BYTES;
+	e->features = ZCRX_FEATURES;
+	e->__resv2 = 0;
+	return sizeof(*e);
+}
+
+static ssize_t io_query_zcrx_notif(union io_query_data *data)
+{
+	struct io_uring_query_zcrx_event *e = &data->zcrx_notif;
+
+	e->event_flags = ZCRX_EVENT_TYPE_MASK;
+	e->stats_size = sizeof(struct zcrx_stats);
+	e->stats_off_alignment = __alignof__(struct zcrx_stats);
+	e->__resv1 = 0;
+	memset(&e->__resv2, 0, sizeof(e->__resv2));
+	return sizeof(*e);
+}
+
+static ssize_t io_query_scq(union io_query_data *data)
+{
+	struct io_uring_query_scq *e = &data->scq;
+
+	e->hdr_size = sizeof(struct io_rings);
+	e->hdr_alignment = SMP_CACHE_BYTES;
+	return sizeof(*e);
+}
+
+static int io_handle_query_entry(union io_query_data *data, void __user *uhdr,
+				 u64 *next_entry)
+{
+	struct io_uring_query_hdr hdr;
+	size_t usize, res_size = 0;
+	ssize_t ret = -EINVAL;
+	void __user *udata;
+
+	if (copy_from_user(&hdr, uhdr, sizeof(hdr)))
+		return -EFAULT;
+	/* copy_struct_to_user() zeros up to usize bytes */
+	if (hdr.size > PAGE_SIZE)
+		return -E2BIG;
+	usize = hdr.size;
+	hdr.size = min(hdr.size, IO_MAX_QUERY_SIZE);
+	udata = u64_to_user_ptr(hdr.query_data);
+
+	if (hdr.query_op >= __IO_URING_QUERY_MAX) {
+		ret = -EOPNOTSUPP;
+		goto out;
+	}
+	if (!mem_is_zero(hdr.__resv, sizeof(hdr.__resv)) || hdr.result || !hdr.size)
+		goto out;
+	if (copy_from_user(data, udata, hdr.size))
+		return -EFAULT;
+
+	switch (hdr.query_op) {
+	case IO_URING_QUERY_OPCODES:
+		ret = io_query_ops(data);
+		break;
+	case IO_URING_QUERY_ZCRX:
+		ret = io_query_zcrx(data);
+		break;
+	case IO_URING_QUERY_ZCRX_EVENT:
+		ret = io_query_zcrx_notif(data);
+		break;
+	case IO_URING_QUERY_SCQ:
+		ret = io_query_scq(data);
+		break;
+	}
+
+	if (ret >= 0) {
+		if (WARN_ON_ONCE(ret > IO_MAX_QUERY_SIZE))
+			return -EFAULT;
+		res_size = ret;
+		ret = 0;
+	}
+out:
+	hdr.result = ret;
+	hdr.size = min_t(size_t, usize, res_size);
+
+	if (copy_struct_to_user(udata, usize, data, hdr.size, NULL))
+		return -EFAULT;
+	if (copy_to_user(uhdr, &hdr, sizeof(hdr)))
+		return -EFAULT;
+	*next_entry = hdr.next_entry;
+	return 0;
+}
+
+int io_query(void __user *arg, unsigned nr_args)
+{
+	union io_query_data entry_buffer;
+	void __user *uhdr = arg;
+	int ret, nr = 0;
+
+	memset(&entry_buffer, 0, sizeof(entry_buffer));
+
+	if (nr_args)
+		return -EINVAL;
+
+	while (uhdr) {
+		u64 next_hdr;
+
+		ret = io_handle_query_entry(&entry_buffer, uhdr, &next_hdr);
+		if (ret)
+			return ret;
+		uhdr = u64_to_user_ptr(next_hdr);
+
+		/* Have some limit to avoid a potential cycle */
+		if (++nr >= IO_MAX_QUERY_ENTRIES)
+			return -ERANGE;
+		if (fatal_signal_pending(current))
+			return -EINTR;
+		cond_resched();
+	}
+	return 0;
+}