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, 220 insertions, 0 deletions

diff --git a/drivers/rtc/rtc-ds1742.c b/drivers/rtc/rtc-ds1742.c
new file mode 100644
index 000000000..6ae8b9a29
--- /dev/null
+++ b/drivers/rtc/rtc-ds1742.c
@@ -0,0 +1,220 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * An rtc driver for the Dallas DS1742
+ *
+ * Copyright (C) 2006 Atsushi Nemoto <anemo@mba.ocn.ne.jp>
+ *
+ * Copyright (C) 2006 Torsten Ertbjerg Rasmussen <tr@newtec.dk>
+ *  - nvram size determined from resource
+ *  - this ds1742 driver now supports ds1743.
+ */
+
+#include <linux/bcd.h>
+#include <linux/kernel.h>
+#include <linux/gfp.h>
+#include <linux/delay.h>
+#include <linux/jiffies.h>
+#include <linux/rtc.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/io.h>
+#include <linux/module.h>
+
+#define RTC_SIZE		8
+
+#define RTC_CONTROL		0
+#define RTC_CENTURY		0
+#define RTC_SECONDS		1
+#define RTC_MINUTES		2
+#define RTC_HOURS		3
+#define RTC_DAY			4
+#define RTC_DATE		5
+#define RTC_MONTH		6
+#define RTC_YEAR		7
+
+#define RTC_CENTURY_MASK	0x3f
+#define RTC_SECONDS_MASK	0x7f
+#define RTC_DAY_MASK		0x07
+
+/* Bits in the Control/Century register */
+#define RTC_WRITE		0x80
+#define RTC_READ		0x40
+
+/* Bits in the Seconds register */
+#define RTC_STOP		0x80
+
+/* Bits in the Day register */
+#define RTC_BATT_FLAG		0x80
+
+struct rtc_plat_data {
+	void __iomem *ioaddr_nvram;
+	void __iomem *ioaddr_rtc;
+	unsigned long last_jiffies;
+};
+
+static int ds1742_rtc_set_time(struct device *dev, struct rtc_time *tm)
+{
+	struct rtc_plat_data *pdata = dev_get_drvdata(dev);
+	void __iomem *ioaddr = pdata->ioaddr_rtc;
+	u8 century;
+
+	century = bin2bcd((tm->tm_year + 1900) / 100);
+
+	writeb(RTC_WRITE, ioaddr + RTC_CONTROL);
+
+	writeb(bin2bcd(tm->tm_year % 100), ioaddr + RTC_YEAR);
+	writeb(bin2bcd(tm->tm_mon + 1), ioaddr + RTC_MONTH);
+	writeb(bin2bcd(tm->tm_wday) & RTC_DAY_MASK, ioaddr + RTC_DAY);
+	writeb(bin2bcd(tm->tm_mday), ioaddr + RTC_DATE);
+	writeb(bin2bcd(tm->tm_hour), ioaddr + RTC_HOURS);
+	writeb(bin2bcd(tm->tm_min), ioaddr + RTC_MINUTES);
+	writeb(bin2bcd(tm->tm_sec) & RTC_SECONDS_MASK, ioaddr + RTC_SECONDS);
+
+	/* RTC_CENTURY and RTC_CONTROL share same register */
+	writeb(RTC_WRITE | (century & RTC_CENTURY_MASK), ioaddr + RTC_CENTURY);
+	writeb(century & RTC_CENTURY_MASK, ioaddr + RTC_CONTROL);
+	return 0;
+}
+
+static int ds1742_rtc_read_time(struct device *dev, struct rtc_time *tm)
+{
+	struct rtc_plat_data *pdata = dev_get_drvdata(dev);
+	void __iomem *ioaddr = pdata->ioaddr_rtc;
+	unsigned int year, month, day, hour, minute, second, week;
+	unsigned int century;
+
+	/* give enough time to update RTC in case of continuous read */
+	if (pdata->last_jiffies == jiffies)
+		msleep(1);
+	pdata->last_jiffies = jiffies;
+	writeb(RTC_READ, ioaddr + RTC_CONTROL);
+	second = readb(ioaddr + RTC_SECONDS) & RTC_SECONDS_MASK;
+	minute = readb(ioaddr + RTC_MINUTES);
+	hour = readb(ioaddr + RTC_HOURS);
+	day = readb(ioaddr + RTC_DATE);
+	week = readb(ioaddr + RTC_DAY) & RTC_DAY_MASK;
+	month = readb(ioaddr + RTC_MONTH);
+	year = readb(ioaddr + RTC_YEAR);
+	century = readb(ioaddr + RTC_CENTURY) & RTC_CENTURY_MASK;
+	writeb(0, ioaddr + RTC_CONTROL);
+	tm->tm_sec = bcd2bin(second);
+	tm->tm_min = bcd2bin(minute);
+	tm->tm_hour = bcd2bin(hour);
+	tm->tm_mday = bcd2bin(day);
+	tm->tm_wday = bcd2bin(week);
+	tm->tm_mon = bcd2bin(month) - 1;
+	/* year is 1900 + tm->tm_year */
+	tm->tm_year = bcd2bin(year) + bcd2bin(century) * 100 - 1900;
+
+	return 0;
+}
+
+static const struct rtc_class_ops ds1742_rtc_ops = {
+	.read_time	= ds1742_rtc_read_time,
+	.set_time	= ds1742_rtc_set_time,
+};
+
+static int ds1742_nvram_read(void *priv, unsigned int pos, void *val,
+			     size_t bytes)
+{
+	struct rtc_plat_data *pdata = priv;
+	void __iomem *ioaddr = pdata->ioaddr_nvram;
+	u8 *buf = val;
+
+	for (; bytes; bytes--)
+		*buf++ = readb(ioaddr + pos++);
+	return 0;
+}
+
+static int ds1742_nvram_write(void *priv, unsigned int pos, void *val,
+			      size_t bytes)
+{
+	struct rtc_plat_data *pdata = priv;
+	void __iomem *ioaddr = pdata->ioaddr_nvram;
+	u8 *buf = val;
+
+	for (; bytes; bytes--)
+		writeb(*buf++, ioaddr + pos++);
+	return 0;
+}
+
+static int ds1742_rtc_probe(struct platform_device *pdev)
+{
+	struct rtc_device *rtc;
+	struct resource *res;
+	unsigned int cen, sec;
+	struct rtc_plat_data *pdata;
+	void __iomem *ioaddr;
+	int ret = 0;
+	struct nvmem_config nvmem_cfg = {
+		.name = "ds1742_nvram",
+		.reg_read = ds1742_nvram_read,
+		.reg_write = ds1742_nvram_write,
+	};
+
+
+	pdata = devm_kzalloc(&pdev->dev, sizeof(*pdata), GFP_KERNEL);
+	if (!pdata)
+		return -ENOMEM;
+
+	ioaddr = devm_platform_get_and_ioremap_resource(pdev, 0, &res);
+	if (IS_ERR(ioaddr))
+		return PTR_ERR(ioaddr);
+
+	pdata->ioaddr_nvram = ioaddr;
+	pdata->ioaddr_rtc = ioaddr + resource_size(res) - RTC_SIZE;
+
+	nvmem_cfg.size = resource_size(res) - RTC_SIZE;
+	nvmem_cfg.priv = pdata;
+
+	/* turn RTC on if it was not on */
+	ioaddr = pdata->ioaddr_rtc;
+	sec = readb(ioaddr + RTC_SECONDS);
+	if (sec & RTC_STOP) {
+		sec &= RTC_SECONDS_MASK;
+		cen = readb(ioaddr + RTC_CENTURY) & RTC_CENTURY_MASK;
+		writeb(RTC_WRITE, ioaddr + RTC_CONTROL);
+		writeb(sec, ioaddr + RTC_SECONDS);
+		writeb(cen & RTC_CENTURY_MASK, ioaddr + RTC_CONTROL);
+	}
+	if (!(readb(ioaddr + RTC_DAY) & RTC_BATT_FLAG))
+		dev_warn(&pdev->dev, "voltage-low detected.\n");
+
+	pdata->last_jiffies = jiffies;
+	platform_set_drvdata(pdev, pdata);
+
+	rtc = devm_rtc_allocate_device(&pdev->dev);
+	if (IS_ERR(rtc))
+		return PTR_ERR(rtc);
+
+	rtc->ops = &ds1742_rtc_ops;
+
+	ret = devm_rtc_register_device(rtc);
+	if (ret)
+		return ret;
+
+	devm_rtc_nvmem_register(rtc, &nvmem_cfg);
+
+	return 0;
+}
+
+static const struct of_device_id __maybe_unused ds1742_rtc_of_match[] = {
+	{ .compatible = "maxim,ds1742", },
+	{ }
+};
+MODULE_DEVICE_TABLE(of, ds1742_rtc_of_match);
+
+static struct platform_driver ds1742_rtc_driver = {
+	.probe		= ds1742_rtc_probe,
+	.driver		= {
+		.name	= "rtc-ds1742",
+		.of_match_table = of_match_ptr(ds1742_rtc_of_match),
+	},
+};
+
+module_platform_driver(ds1742_rtc_driver);
+
+MODULE_AUTHOR("Atsushi Nemoto <anemo@mba.ocn.ne.jp>");
+MODULE_DESCRIPTION("Dallas DS1742 RTC driver");
+MODULE_LICENSE("GPL");
+MODULE_ALIAS("platform:rtc-ds1742");