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, 166 insertions, 0 deletions

diff --git a/drivers/input/serio/q40kbd.c b/drivers/input/serio/q40kbd.c
new file mode 100644
index 000000000..2f553efbe
--- /dev/null
+++ b/drivers/input/serio/q40kbd.c
@@ -0,0 +1,166 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ *  Copyright (c) 2000-2001 Vojtech Pavlik
+ *
+ *  Based on the work of:
+ *	Richard Zidlicky <Richard.Zidlicky@stud.informatik.uni-erlangen.de>
+ */
+
+/*
+ * Q40 PS/2 keyboard controller driver for Linux/m68k
+ */
+
+#include <linux/module.h>
+#include <linux/serio.h>
+#include <linux/interrupt.h>
+#include <linux/err.h>
+#include <linux/bitops.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+
+#include <asm/io.h>
+#include <linux/uaccess.h>
+#include <asm/q40_master.h>
+#include <asm/irq.h>
+#include <asm/q40ints.h>
+
+#define DRV_NAME	"q40kbd"
+
+MODULE_AUTHOR("Vojtech Pavlik <vojtech@ucw.cz>");
+MODULE_DESCRIPTION("Q40 PS/2 keyboard controller driver");
+MODULE_LICENSE("GPL");
+MODULE_ALIAS("platform:" DRV_NAME);
+
+struct q40kbd {
+	struct serio *port;
+	spinlock_t lock;
+};
+
+static irqreturn_t q40kbd_interrupt(int irq, void *dev_id)
+{
+	struct q40kbd *q40kbd = dev_id;
+
+	guard(spinlock_irqsave)(&q40kbd->lock);
+
+	if (Q40_IRQ_KEYB_MASK & master_inb(INTERRUPT_REG))
+		serio_interrupt(q40kbd->port, master_inb(KEYCODE_REG), 0);
+
+	master_outb(-1, KEYBOARD_UNLOCK_REG);
+
+	return IRQ_HANDLED;
+}
+
+/*
+ * q40kbd_flush() flushes all data that may be in the keyboard buffers
+ */
+
+static void q40kbd_flush(struct q40kbd *q40kbd)
+{
+	int maxread = 100;
+
+	guard(spinlock_irqsave)(&q40kbd->lock);
+
+	while (maxread-- && (Q40_IRQ_KEYB_MASK & master_inb(INTERRUPT_REG)))
+		master_inb(KEYCODE_REG);
+}
+
+static void q40kbd_stop(void)
+{
+	master_outb(0, KEY_IRQ_ENABLE_REG);
+	master_outb(-1, KEYBOARD_UNLOCK_REG);
+}
+
+/*
+ * q40kbd_open() is called when a port is open by the higher layer.
+ * It allocates the interrupt and enables in in the chip.
+ */
+
+static int q40kbd_open(struct serio *port)
+{
+	struct q40kbd *q40kbd = port->port_data;
+
+	q40kbd_flush(q40kbd);
+
+	/* off we go */
+	master_outb(-1, KEYBOARD_UNLOCK_REG);
+	master_outb(1, KEY_IRQ_ENABLE_REG);
+
+	return 0;
+}
+
+static void q40kbd_close(struct serio *port)
+{
+	struct q40kbd *q40kbd = port->port_data;
+
+	q40kbd_stop();
+	q40kbd_flush(q40kbd);
+}
+
+static int q40kbd_probe(struct platform_device *pdev)
+{
+	struct q40kbd *q40kbd;
+	struct serio *port;
+	int error;
+
+	q40kbd = kzalloc_obj(*q40kbd);
+	port = kzalloc_obj(*port);
+	if (!q40kbd || !port) {
+		error = -ENOMEM;
+		goto err_free_mem;
+	}
+
+	q40kbd->port = port;
+	spin_lock_init(&q40kbd->lock);
+
+	port->id.type = SERIO_8042;
+	port->open = q40kbd_open;
+	port->close = q40kbd_close;
+	port->port_data = q40kbd;
+	port->dev.parent = &pdev->dev;
+	strscpy(port->name, "Q40 Kbd Port", sizeof(port->name));
+	strscpy(port->phys, "Q40", sizeof(port->phys));
+
+	q40kbd_stop();
+
+	error = request_irq(Q40_IRQ_KEYBOARD, q40kbd_interrupt, 0,
+			    DRV_NAME, q40kbd);
+	if (error) {
+		dev_err(&pdev->dev, "Can't get irq %d.\n", Q40_IRQ_KEYBOARD);
+		goto err_free_mem;
+	}
+
+	serio_register_port(q40kbd->port);
+
+	platform_set_drvdata(pdev, q40kbd);
+	printk(KERN_INFO "serio: Q40 kbd registered\n");
+
+	return 0;
+
+err_free_mem:
+	kfree(port);
+	kfree(q40kbd);
+	return error;
+}
+
+static void q40kbd_remove(struct platform_device *pdev)
+{
+	struct q40kbd *q40kbd = platform_get_drvdata(pdev);
+
+	/*
+	 * q40kbd_close() will be called as part of unregistering
+	 * and will ensure that IRQ is turned off, so it is safe
+	 * to unregister port first and free IRQ later.
+	 */
+	serio_unregister_port(q40kbd->port);
+	free_irq(Q40_IRQ_KEYBOARD, q40kbd);
+	kfree(q40kbd);
+}
+
+static struct platform_driver q40kbd_driver = {
+	.driver		= {
+		.name	= "q40kbd",
+	},
+	.remove		= q40kbd_remove,
+};
+
+module_platform_driver_probe(q40kbd_driver, q40kbd_probe);