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, 196 insertions, 0 deletions

diff --git a/drivers/input/misc/ibm-panel.c b/drivers/input/misc/ibm-panel.c
new file mode 100644
index 000000000..aa48f62d7
--- /dev/null
+++ b/drivers/input/misc/ibm-panel.c
@@ -0,0 +1,196 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) IBM Corporation 2020
+ */
+
+#include <linux/i2c.h>
+#include <linux/init.h>
+#include <linux/input.h>
+#include <linux/kernel.h>
+#include <linux/limits.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/spinlock.h>
+
+#define DEVICE_NAME		"ibm-panel"
+#define PANEL_KEYCODES_COUNT	3
+
+struct ibm_panel {
+	u8 idx;
+	u8 command[11];
+	u32 keycodes[PANEL_KEYCODES_COUNT];
+	spinlock_t lock;	/* protects writes to idx and command */
+	struct input_dev *input;
+};
+
+static u8 ibm_panel_calculate_checksum(struct ibm_panel *panel)
+{
+	u8 chksum;
+	u16 sum = 0;
+	unsigned int i;
+
+	for (i = 0; i < sizeof(panel->command) - 1; ++i) {
+		sum += panel->command[i];
+		if (sum & 0xff00) {
+			sum &= 0xff;
+			sum++;
+		}
+	}
+
+	chksum = sum & 0xff;
+	chksum = ~chksum;
+	chksum++;
+
+	return chksum;
+}
+
+static void ibm_panel_process_command(struct ibm_panel *panel)
+{
+	u8 button;
+	u8 chksum;
+
+	if (panel->command[0] != 0xff && panel->command[1] != 0xf0) {
+		dev_dbg(&panel->input->dev, "command invalid: %02x %02x\n",
+			panel->command[0], panel->command[1]);
+		return;
+	}
+
+	chksum = ibm_panel_calculate_checksum(panel);
+	if (chksum != panel->command[sizeof(panel->command) - 1]) {
+		dev_dbg(&panel->input->dev,
+			"command failed checksum: %u != %u\n", chksum,
+			panel->command[sizeof(panel->command) - 1]);
+		return;
+	}
+
+	button = panel->command[2] & 0xf;
+	if (button < PANEL_KEYCODES_COUNT) {
+		input_report_key(panel->input, panel->keycodes[button],
+				 !(panel->command[2] & 0x80));
+		input_sync(panel->input);
+	} else {
+		dev_dbg(&panel->input->dev, "unknown button %u\n",
+			button);
+	}
+}
+
+static int ibm_panel_i2c_slave_cb(struct i2c_client *client,
+				  enum i2c_slave_event event, u8 *val)
+{
+	struct ibm_panel *panel = i2c_get_clientdata(client);
+
+	dev_dbg(&panel->input->dev, "event: %u data: %02x\n", event, *val);
+
+	guard(spinlock_irqsave)(&panel->lock);
+
+	switch (event) {
+	case I2C_SLAVE_STOP:
+		if (panel->idx == sizeof(panel->command))
+			ibm_panel_process_command(panel);
+		else
+			dev_dbg(&panel->input->dev,
+				"command incorrect size %u\n", panel->idx);
+		fallthrough;
+	case I2C_SLAVE_WRITE_REQUESTED:
+		panel->idx = 0;
+		break;
+	case I2C_SLAVE_WRITE_RECEIVED:
+		if (panel->idx < sizeof(panel->command))
+			panel->command[panel->idx++] = *val;
+		else
+			/*
+			 * The command is too long and therefore invalid, so set the index
+			 * to it's largest possible value. When a STOP is finally received,
+			 * the command will be rejected upon processing.
+			 */
+			panel->idx = U8_MAX;
+		break;
+	case I2C_SLAVE_READ_REQUESTED:
+	case I2C_SLAVE_READ_PROCESSED:
+		*val = 0xff;
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int ibm_panel_probe(struct i2c_client *client)
+{
+	struct ibm_panel *panel;
+	int i;
+	int error;
+
+	panel = devm_kzalloc(&client->dev, sizeof(*panel), GFP_KERNEL);
+	if (!panel)
+		return -ENOMEM;
+
+	spin_lock_init(&panel->lock);
+
+	panel->input = devm_input_allocate_device(&client->dev);
+	if (!panel->input)
+		return -ENOMEM;
+
+	panel->input->name = client->name;
+	panel->input->id.bustype = BUS_I2C;
+
+	error = device_property_read_u32_array(&client->dev,
+					       "linux,keycodes",
+					       panel->keycodes,
+					       PANEL_KEYCODES_COUNT);
+	if (error) {
+		/*
+		 * Use gamepad buttons as defaults for compatibility with
+		 * existing applications.
+		 */
+		panel->keycodes[0] = BTN_NORTH;
+		panel->keycodes[1] = BTN_SOUTH;
+		panel->keycodes[2] = BTN_SELECT;
+	}
+
+	for (i = 0; i < PANEL_KEYCODES_COUNT; ++i)
+		input_set_capability(panel->input, EV_KEY, panel->keycodes[i]);
+
+	error = input_register_device(panel->input);
+	if (error) {
+		dev_err(&client->dev,
+			"Failed to register input device: %d\n", error);
+		return error;
+	}
+
+	i2c_set_clientdata(client, panel);
+	error = i2c_slave_register(client, ibm_panel_i2c_slave_cb);
+	if (error) {
+		dev_err(&client->dev,
+			"Failed to register as i2c slave: %d\n", error);
+		return error;
+	}
+
+	return 0;
+}
+
+static void ibm_panel_remove(struct i2c_client *client)
+{
+	i2c_slave_unregister(client);
+}
+
+static const struct of_device_id ibm_panel_match[] = {
+	{ .compatible = "ibm,op-panel" },
+	{ }
+};
+MODULE_DEVICE_TABLE(of, ibm_panel_match);
+
+static struct i2c_driver ibm_panel_driver = {
+	.driver = {
+		.name = DEVICE_NAME,
+		.of_match_table = ibm_panel_match,
+	},
+	.probe = ibm_panel_probe,
+	.remove = ibm_panel_remove,
+};
+module_i2c_driver(ibm_panel_driver);
+
+MODULE_AUTHOR("Eddie James <eajames@linux.ibm.com>");
+MODULE_DESCRIPTION("IBM Operation Panel Driver");
+MODULE_LICENSE("GPL");