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, 218 insertions, 0 deletions

diff --git a/drivers/rtc/rtc-ftrtc010.c b/drivers/rtc/rtc-ftrtc010.c
new file mode 100644
index 000000000..98fa0f518
--- /dev/null
+++ b/drivers/rtc/rtc-ftrtc010.c
@@ -0,0 +1,218 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ *  Faraday Technology FTRTC010 driver
+ *
+ *  Copyright (C) 2009 Janos Laube <janos.dev@gmail.com>
+ *
+ * Original code for older kernel 2.6.15 are from Stormlinksemi
+ * first update from Janos Laube for > 2.6.29 kernels
+ *
+ * checkpatch fixes and usage of rtc-lib code
+ * Hans Ulli Kroll <ulli.kroll@googlemail.com>
+ */
+
+#include <linux/rtc.h>
+#include <linux/io.h>
+#include <linux/slab.h>
+#include <linux/platform_device.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/clk.h>
+
+#define DRV_NAME        "rtc-ftrtc010"
+
+MODULE_AUTHOR("Hans Ulli Kroll <ulli.kroll@googlemail.com>");
+MODULE_DESCRIPTION("RTC driver for Gemini SoC");
+MODULE_LICENSE("GPL");
+MODULE_ALIAS("platform:" DRV_NAME);
+
+struct ftrtc010_rtc {
+	void __iomem		*rtc_base;
+	int			rtc_irq;
+	struct clk		*pclk;
+	struct clk		*extclk;
+};
+
+enum ftrtc010_rtc_offsets {
+	FTRTC010_RTC_SECOND		= 0x00,
+	FTRTC010_RTC_MINUTE		= 0x04,
+	FTRTC010_RTC_HOUR		= 0x08,
+	FTRTC010_RTC_DAYS		= 0x0C,
+	FTRTC010_RTC_ALARM_SECOND	= 0x10,
+	FTRTC010_RTC_ALARM_MINUTE	= 0x14,
+	FTRTC010_RTC_ALARM_HOUR		= 0x18,
+	FTRTC010_RTC_RECORD		= 0x1C,
+	FTRTC010_RTC_CR			= 0x20,
+};
+
+static irqreturn_t ftrtc010_rtc_interrupt(int irq, void *dev)
+{
+	return IRQ_HANDLED;
+}
+
+/*
+ * Looks like the RTC in the Gemini SoC is (totaly) broken
+ * We can't read/write directly the time from RTC registers.
+ * We must do some "offset" calculation to get the real time
+ *
+ * This FIX works pretty fine and Stormlinksemi aka Cortina-Networks does
+ * the same thing, without the rtc-lib.c calls.
+ */
+
+static int ftrtc010_rtc_read_time(struct device *dev, struct rtc_time *tm)
+{
+	struct ftrtc010_rtc *rtc = dev_get_drvdata(dev);
+
+	u32 days, hour, min, sec, offset;
+	timeu64_t time;
+
+	sec  = readl(rtc->rtc_base + FTRTC010_RTC_SECOND);
+	min  = readl(rtc->rtc_base + FTRTC010_RTC_MINUTE);
+	hour = readl(rtc->rtc_base + FTRTC010_RTC_HOUR);
+	days = readl(rtc->rtc_base + FTRTC010_RTC_DAYS);
+	offset = readl(rtc->rtc_base + FTRTC010_RTC_RECORD);
+
+	time = offset + days * 86400 + hour * 3600 + min * 60 + sec;
+
+	rtc_time64_to_tm(time, tm);
+
+	return 0;
+}
+
+static int ftrtc010_rtc_set_time(struct device *dev, struct rtc_time *tm)
+{
+	struct ftrtc010_rtc *rtc = dev_get_drvdata(dev);
+	u32 sec, min, hour, day, offset;
+	timeu64_t time;
+
+	time = rtc_tm_to_time64(tm);
+
+	sec = readl(rtc->rtc_base + FTRTC010_RTC_SECOND);
+	min = readl(rtc->rtc_base + FTRTC010_RTC_MINUTE);
+	hour = readl(rtc->rtc_base + FTRTC010_RTC_HOUR);
+	day = readl(rtc->rtc_base + FTRTC010_RTC_DAYS);
+
+	offset = time - (day * 86400 + hour * 3600 + min * 60 + sec);
+
+	writel(offset, rtc->rtc_base + FTRTC010_RTC_RECORD);
+	writel(0x01, rtc->rtc_base + FTRTC010_RTC_CR);
+
+	return 0;
+}
+
+static const struct rtc_class_ops ftrtc010_rtc_ops = {
+	.read_time     = ftrtc010_rtc_read_time,
+	.set_time      = ftrtc010_rtc_set_time,
+};
+
+static int ftrtc010_rtc_probe(struct platform_device *pdev)
+{
+	u32 days, hour, min, sec;
+	struct ftrtc010_rtc *rtc;
+	struct device *dev = &pdev->dev;
+	struct resource *res;
+	struct rtc_device *rtc_dev;
+	int ret;
+
+	rtc = devm_kzalloc(&pdev->dev, sizeof(*rtc), GFP_KERNEL);
+	if (unlikely(!rtc))
+		return -ENOMEM;
+	platform_set_drvdata(pdev, rtc);
+
+	rtc->pclk = devm_clk_get(dev, "PCLK");
+	if (IS_ERR(rtc->pclk)) {
+		dev_err(dev, "could not get PCLK\n");
+	} else {
+		ret = clk_prepare_enable(rtc->pclk);
+		if (ret) {
+			dev_err(dev, "failed to enable PCLK\n");
+			return ret;
+		}
+	}
+	rtc->extclk = devm_clk_get(dev, "EXTCLK");
+	if (IS_ERR(rtc->extclk)) {
+		dev_err(dev, "could not get EXTCLK\n");
+	} else {
+		ret = clk_prepare_enable(rtc->extclk);
+		if (ret) {
+			dev_err(dev, "failed to enable EXTCLK\n");
+			goto err_disable_pclk;
+		}
+	}
+
+	rtc->rtc_irq = platform_get_irq(pdev, 0);
+	if (rtc->rtc_irq < 0) {
+		ret = rtc->rtc_irq;
+		goto err_disable_extclk;
+	}
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (!res) {
+		ret = -ENODEV;
+		goto err_disable_extclk;
+	}
+
+	rtc->rtc_base = devm_ioremap(dev, res->start,
+				     resource_size(res));
+	if (!rtc->rtc_base) {
+		ret = -ENOMEM;
+		goto err_disable_extclk;
+	}
+
+	rtc_dev = devm_rtc_allocate_device(dev);
+	if (IS_ERR(rtc_dev)) {
+		ret = PTR_ERR(rtc_dev);
+		goto err_disable_extclk;
+	}
+
+	rtc_dev->ops = &ftrtc010_rtc_ops;
+
+	sec  = readl(rtc->rtc_base + FTRTC010_RTC_SECOND);
+	min  = readl(rtc->rtc_base + FTRTC010_RTC_MINUTE);
+	hour = readl(rtc->rtc_base + FTRTC010_RTC_HOUR);
+	days = readl(rtc->rtc_base + FTRTC010_RTC_DAYS);
+
+	rtc_dev->range_min = (u64)days * 86400 + hour * 3600 + min * 60 + sec;
+	rtc_dev->range_max = U32_MAX + rtc_dev->range_min;
+
+	ret = devm_request_irq(dev, rtc->rtc_irq, ftrtc010_rtc_interrupt,
+			       IRQF_SHARED, pdev->name, dev);
+	if (unlikely(ret))
+		goto err_disable_extclk;
+
+	return devm_rtc_register_device(rtc_dev);
+
+err_disable_extclk:
+	clk_disable_unprepare(rtc->extclk);
+err_disable_pclk:
+	clk_disable_unprepare(rtc->pclk);
+	return ret;
+}
+
+static void ftrtc010_rtc_remove(struct platform_device *pdev)
+{
+	struct ftrtc010_rtc *rtc = platform_get_drvdata(pdev);
+
+	if (!IS_ERR(rtc->extclk))
+		clk_disable_unprepare(rtc->extclk);
+	if (!IS_ERR(rtc->pclk))
+		clk_disable_unprepare(rtc->pclk);
+}
+
+static const struct of_device_id ftrtc010_rtc_dt_match[] = {
+	{ .compatible = "cortina,gemini-rtc" },
+	{ .compatible = "faraday,ftrtc010" },
+	{ }
+};
+MODULE_DEVICE_TABLE(of, ftrtc010_rtc_dt_match);
+
+static struct platform_driver ftrtc010_rtc_driver = {
+	.driver		= {
+		.name	= DRV_NAME,
+		.of_match_table = ftrtc010_rtc_dt_match,
+	},
+	.probe		= ftrtc010_rtc_probe,
+	.remove		= ftrtc010_rtc_remove,
+};
+
+module_platform_driver_probe(ftrtc010_rtc_driver, ftrtc010_rtc_probe);