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, 381 insertions, 0 deletions

diff --git a/drivers/rtc/rtc-pic32.c b/drivers/rtc/rtc-pic32.c
new file mode 100644
index 000000000..3c7a38a4a
--- /dev/null
+++ b/drivers/rtc/rtc-pic32.c
@@ -0,0 +1,381 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * PIC32 RTC driver
+ *
+ * Joshua Henderson <joshua.henderson@microchip.com>
+ * Copyright (C) 2016 Microchip Technology Inc.  All rights reserved.
+ *
+ */
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/io.h>
+#include <linux/slab.h>
+#include <linux/clk.h>
+#include <linux/rtc.h>
+#include <linux/bcd.h>
+#include <linux/platform_data/pic32.h>
+
+#define PIC32_RTCCON		0x00
+#define PIC32_RTCCON_ON		BIT(15)
+#define PIC32_RTCCON_SIDL	BIT(13)
+#define PIC32_RTCCON_RTCCLKSEL	(3 << 9)
+#define PIC32_RTCCON_RTCCLKON	BIT(6)
+#define PIC32_RTCCON_RTCWREN	BIT(3)
+#define PIC32_RTCCON_RTCSYNC	BIT(2)
+#define PIC32_RTCCON_HALFSEC	BIT(1)
+#define PIC32_RTCCON_RTCOE	BIT(0)
+
+#define PIC32_RTCALRM		0x10
+#define PIC32_RTCALRM_ALRMEN	BIT(15)
+#define PIC32_RTCALRM_CHIME	BIT(14)
+#define PIC32_RTCALRM_PIV	BIT(13)
+#define PIC32_RTCALRM_ALARMSYNC	BIT(12)
+#define PIC32_RTCALRM_AMASK	0x0F00
+#define PIC32_RTCALRM_ARPT	0xFF
+
+#define PIC32_RTCHOUR		0x23
+#define PIC32_RTCMIN		0x22
+#define PIC32_RTCSEC		0x21
+#define PIC32_RTCYEAR		0x33
+#define PIC32_RTCMON		0x32
+#define PIC32_RTCDAY		0x31
+
+#define PIC32_ALRMTIME		0x40
+#define PIC32_ALRMDATE		0x50
+
+#define PIC32_ALRMHOUR		0x43
+#define PIC32_ALRMMIN		0x42
+#define PIC32_ALRMSEC		0x41
+#define PIC32_ALRMYEAR		0x53
+#define PIC32_ALRMMON		0x52
+#define PIC32_ALRMDAY		0x51
+
+struct pic32_rtc_dev {
+	struct rtc_device	*rtc;
+	void __iomem		*reg_base;
+	struct clk		*clk;
+	spinlock_t		alarm_lock;
+	int			alarm_irq;
+	bool			alarm_clk_enabled;
+};
+
+static void pic32_rtc_alarm_clk_enable(struct pic32_rtc_dev *pdata,
+				       bool enable)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&pdata->alarm_lock, flags);
+	if (enable) {
+		if (!pdata->alarm_clk_enabled) {
+			clk_enable(pdata->clk);
+			pdata->alarm_clk_enabled = true;
+		}
+	} else {
+		if (pdata->alarm_clk_enabled) {
+			clk_disable(pdata->clk);
+			pdata->alarm_clk_enabled = false;
+		}
+	}
+	spin_unlock_irqrestore(&pdata->alarm_lock, flags);
+}
+
+static irqreturn_t pic32_rtc_alarmirq(int irq, void *id)
+{
+	struct pic32_rtc_dev *pdata = (struct pic32_rtc_dev *)id;
+
+	clk_enable(pdata->clk);
+	rtc_update_irq(pdata->rtc, 1, RTC_AF | RTC_IRQF);
+	clk_disable(pdata->clk);
+
+	pic32_rtc_alarm_clk_enable(pdata, false);
+
+	return IRQ_HANDLED;
+}
+
+static int pic32_rtc_setaie(struct device *dev, unsigned int enabled)
+{
+	struct pic32_rtc_dev *pdata = dev_get_drvdata(dev);
+	void __iomem *base = pdata->reg_base;
+
+	clk_enable(pdata->clk);
+
+	writel(PIC32_RTCALRM_ALRMEN,
+	       base + (enabled ? PIC32_SET(PIC32_RTCALRM) :
+		       PIC32_CLR(PIC32_RTCALRM)));
+
+	clk_disable(pdata->clk);
+
+	pic32_rtc_alarm_clk_enable(pdata, enabled);
+
+	return 0;
+}
+
+static int pic32_rtc_setfreq(struct device *dev, int freq)
+{
+	struct pic32_rtc_dev *pdata = dev_get_drvdata(dev);
+	void __iomem *base = pdata->reg_base;
+
+	clk_enable(pdata->clk);
+
+	writel(PIC32_RTCALRM_AMASK, base + PIC32_CLR(PIC32_RTCALRM));
+	writel(freq << 8, base + PIC32_SET(PIC32_RTCALRM));
+	writel(PIC32_RTCALRM_CHIME, base + PIC32_SET(PIC32_RTCALRM));
+
+	clk_disable(pdata->clk);
+
+	return 0;
+}
+
+static int pic32_rtc_gettime(struct device *dev, struct rtc_time *rtc_tm)
+{
+	struct pic32_rtc_dev *pdata = dev_get_drvdata(dev);
+	void __iomem *base = pdata->reg_base;
+	unsigned int tries = 0;
+
+	clk_enable(pdata->clk);
+
+	do {
+		rtc_tm->tm_hour = readb(base + PIC32_RTCHOUR);
+		rtc_tm->tm_min = readb(base + PIC32_RTCMIN);
+		rtc_tm->tm_mon  = readb(base + PIC32_RTCMON);
+		rtc_tm->tm_mday = readb(base + PIC32_RTCDAY);
+		rtc_tm->tm_year = readb(base + PIC32_RTCYEAR);
+		rtc_tm->tm_sec  = readb(base + PIC32_RTCSEC);
+
+		/*
+		 * The only way to work out whether the system was mid-update
+		 * when we read it is to check the second counter, and if it
+		 * is zero, then we re-try the entire read.
+		 */
+		tries += 1;
+	} while (rtc_tm->tm_sec == 0 && tries < 2);
+
+	rtc_tm->tm_sec = bcd2bin(rtc_tm->tm_sec);
+	rtc_tm->tm_min = bcd2bin(rtc_tm->tm_min);
+	rtc_tm->tm_hour = bcd2bin(rtc_tm->tm_hour);
+	rtc_tm->tm_mday = bcd2bin(rtc_tm->tm_mday);
+	rtc_tm->tm_mon = bcd2bin(rtc_tm->tm_mon) - 1;
+	rtc_tm->tm_year = bcd2bin(rtc_tm->tm_year);
+
+	rtc_tm->tm_year += 100;
+
+	dev_dbg(dev, "read time %ptR\n", rtc_tm);
+
+	clk_disable(pdata->clk);
+	return 0;
+}
+
+static int pic32_rtc_settime(struct device *dev, struct rtc_time *tm)
+{
+	struct pic32_rtc_dev *pdata = dev_get_drvdata(dev);
+	void __iomem *base = pdata->reg_base;
+
+	dev_dbg(dev, "set time %ptR\n", tm);
+
+	clk_enable(pdata->clk);
+	writeb(bin2bcd(tm->tm_sec),  base + PIC32_RTCSEC);
+	writeb(bin2bcd(tm->tm_min),  base + PIC32_RTCMIN);
+	writeb(bin2bcd(tm->tm_hour), base + PIC32_RTCHOUR);
+	writeb(bin2bcd(tm->tm_mday), base + PIC32_RTCDAY);
+	writeb(bin2bcd(tm->tm_mon + 1), base + PIC32_RTCMON);
+	writeb(bin2bcd(tm->tm_year - 100), base + PIC32_RTCYEAR);
+	clk_disable(pdata->clk);
+
+	return 0;
+}
+
+static int pic32_rtc_getalarm(struct device *dev, struct rtc_wkalrm *alrm)
+{
+	struct pic32_rtc_dev *pdata = dev_get_drvdata(dev);
+	struct rtc_time *alm_tm = &alrm->time;
+	void __iomem *base = pdata->reg_base;
+	unsigned int alm_en;
+
+	clk_enable(pdata->clk);
+	alm_tm->tm_sec  = readb(base + PIC32_ALRMSEC);
+	alm_tm->tm_min  = readb(base + PIC32_ALRMMIN);
+	alm_tm->tm_hour = readb(base + PIC32_ALRMHOUR);
+	alm_tm->tm_mon  = readb(base + PIC32_ALRMMON);
+	alm_tm->tm_mday = readb(base + PIC32_ALRMDAY);
+	alm_tm->tm_year = readb(base + PIC32_ALRMYEAR);
+
+	alm_en = readb(base + PIC32_RTCALRM);
+
+	alrm->enabled = (alm_en & PIC32_RTCALRM_ALRMEN) ? 1 : 0;
+
+	dev_dbg(dev, "getalarm: %d, %ptR\n", alm_en, alm_tm);
+
+	alm_tm->tm_sec = bcd2bin(alm_tm->tm_sec);
+	alm_tm->tm_min = bcd2bin(alm_tm->tm_min);
+	alm_tm->tm_hour = bcd2bin(alm_tm->tm_hour);
+	alm_tm->tm_mday = bcd2bin(alm_tm->tm_mday);
+	alm_tm->tm_mon = bcd2bin(alm_tm->tm_mon) - 1;
+	alm_tm->tm_year = bcd2bin(alm_tm->tm_year);
+
+	clk_disable(pdata->clk);
+	return 0;
+}
+
+static int pic32_rtc_setalarm(struct device *dev, struct rtc_wkalrm *alrm)
+{
+	struct pic32_rtc_dev *pdata = dev_get_drvdata(dev);
+	struct rtc_time *tm = &alrm->time;
+	void __iomem *base = pdata->reg_base;
+
+	clk_enable(pdata->clk);
+	dev_dbg(dev, "setalarm: %d, %ptR\n", alrm->enabled, tm);
+
+	writel(0x00, base + PIC32_ALRMTIME);
+	writel(0x00, base + PIC32_ALRMDATE);
+
+	pic32_rtc_setaie(dev, alrm->enabled);
+
+	clk_disable(pdata->clk);
+	return 0;
+}
+
+static int pic32_rtc_proc(struct device *dev, struct seq_file *seq)
+{
+	struct pic32_rtc_dev *pdata = dev_get_drvdata(dev);
+	void __iomem *base = pdata->reg_base;
+	unsigned int repeat;
+
+	clk_enable(pdata->clk);
+
+	repeat = readw(base + PIC32_RTCALRM);
+	repeat &= PIC32_RTCALRM_ARPT;
+	seq_printf(seq, "periodic_IRQ\t: %s\n", repeat  ? "yes" : "no");
+
+	clk_disable(pdata->clk);
+	return 0;
+}
+
+static const struct rtc_class_ops pic32_rtcops = {
+	.read_time	  = pic32_rtc_gettime,
+	.set_time	  = pic32_rtc_settime,
+	.read_alarm	  = pic32_rtc_getalarm,
+	.set_alarm	  = pic32_rtc_setalarm,
+	.proc		  = pic32_rtc_proc,
+	.alarm_irq_enable = pic32_rtc_setaie,
+};
+
+static void pic32_rtc_enable(struct pic32_rtc_dev *pdata, int en)
+{
+	void __iomem *base = pdata->reg_base;
+
+	if (!base)
+		return;
+
+	clk_enable(pdata->clk);
+	if (!en) {
+		writel(PIC32_RTCCON_ON, base + PIC32_CLR(PIC32_RTCCON));
+	} else {
+		pic32_syskey_unlock();
+
+		writel(PIC32_RTCCON_RTCWREN, base + PIC32_SET(PIC32_RTCCON));
+		writel(3 << 9, base + PIC32_CLR(PIC32_RTCCON));
+
+		if (!(readl(base + PIC32_RTCCON) & PIC32_RTCCON_ON))
+			writel(PIC32_RTCCON_ON, base + PIC32_SET(PIC32_RTCCON));
+	}
+	clk_disable(pdata->clk);
+}
+
+static void pic32_rtc_remove(struct platform_device *pdev)
+{
+	struct pic32_rtc_dev *pdata = platform_get_drvdata(pdev);
+
+	pic32_rtc_setaie(&pdev->dev, 0);
+	clk_unprepare(pdata->clk);
+	pdata->clk = NULL;
+}
+
+static int pic32_rtc_probe(struct platform_device *pdev)
+{
+	struct pic32_rtc_dev *pdata;
+	int ret;
+
+	pdata = devm_kzalloc(&pdev->dev, sizeof(*pdata), GFP_KERNEL);
+	if (!pdata)
+		return -ENOMEM;
+
+	platform_set_drvdata(pdev, pdata);
+
+	pdata->alarm_irq = platform_get_irq(pdev, 0);
+	if (pdata->alarm_irq < 0)
+		return pdata->alarm_irq;
+
+	pdata->reg_base = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(pdata->reg_base))
+		return PTR_ERR(pdata->reg_base);
+
+	pdata->clk = devm_clk_get(&pdev->dev, NULL);
+	if (IS_ERR(pdata->clk)) {
+		dev_err(&pdev->dev, "failed to find rtc clock source\n");
+		ret = PTR_ERR(pdata->clk);
+		pdata->clk = NULL;
+		return ret;
+	}
+
+	spin_lock_init(&pdata->alarm_lock);
+
+	pdata->rtc = devm_rtc_allocate_device(&pdev->dev);
+	if (IS_ERR(pdata->rtc))
+		return PTR_ERR(pdata->rtc);
+
+	clk_prepare_enable(pdata->clk);
+
+	pic32_rtc_enable(pdata, 1);
+
+	device_init_wakeup(&pdev->dev, true);
+
+	pdata->rtc->ops = &pic32_rtcops;
+	pdata->rtc->range_min = RTC_TIMESTAMP_BEGIN_2000;
+	pdata->rtc->range_max = RTC_TIMESTAMP_END_2099;
+
+	ret = devm_rtc_register_device(pdata->rtc);
+	if (ret)
+		goto err_nortc;
+
+	pic32_rtc_setfreq(&pdev->dev, 1);
+	ret = devm_request_irq(&pdev->dev, pdata->alarm_irq,
+			       pic32_rtc_alarmirq, 0,
+			       dev_name(&pdev->dev), pdata);
+	if (ret) {
+		dev_err(&pdev->dev,
+			"IRQ %d error %d\n", pdata->alarm_irq, ret);
+		goto err_nortc;
+	}
+
+	clk_disable(pdata->clk);
+
+	return 0;
+
+err_nortc:
+	pic32_rtc_enable(pdata, 0);
+	clk_disable_unprepare(pdata->clk);
+
+	return ret;
+}
+
+static const struct of_device_id pic32_rtc_dt_ids[] = {
+	{ .compatible = "microchip,pic32mzda-rtc" },
+	{ /* sentinel */ }
+};
+MODULE_DEVICE_TABLE(of, pic32_rtc_dt_ids);
+
+static struct platform_driver pic32_rtc_driver = {
+	.probe		= pic32_rtc_probe,
+	.remove		= pic32_rtc_remove,
+	.driver		= {
+		.name	= "pic32-rtc",
+		.of_match_table	= of_match_ptr(pic32_rtc_dt_ids),
+	},
+};
+module_platform_driver(pic32_rtc_driver);
+
+MODULE_DESCRIPTION("Microchip PIC32 RTC Driver");
+MODULE_AUTHOR("Joshua Henderson <joshua.henderson@microchip.com>");
+MODULE_LICENSE("GPL");