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, 207 insertions, 0 deletions

diff --git a/drivers/rtc/rtc-goldfish.c b/drivers/rtc/rtc-goldfish.c
new file mode 100644
index 000000000..53ec7173c
--- /dev/null
+++ b/drivers/rtc/rtc-goldfish.c
@@ -0,0 +1,207 @@
+// SPDX-License-Identifier: GPL-2.0
+/* drivers/rtc/rtc-goldfish.c
+ *
+ * Copyright (C) 2007 Google, Inc.
+ * Copyright (C) 2017 Imagination Technologies Ltd.
+ */
+
+#include <linux/io.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/rtc.h>
+#include <linux/goldfish.h>
+#include <clocksource/timer-goldfish.h>
+
+struct goldfish_rtc {
+	void __iomem *base;
+	int irq;
+	struct rtc_device *rtc;
+};
+
+static int goldfish_rtc_read_alarm(struct device *dev,
+				   struct rtc_wkalrm *alrm)
+{
+	u64 rtc_alarm;
+	u64 rtc_alarm_low;
+	u64 rtc_alarm_high;
+	void __iomem *base;
+	struct goldfish_rtc *rtcdrv;
+
+	rtcdrv = dev_get_drvdata(dev);
+	base = rtcdrv->base;
+
+	rtc_alarm_low = gf_ioread32(base + TIMER_ALARM_LOW);
+	rtc_alarm_high = gf_ioread32(base + TIMER_ALARM_HIGH);
+	rtc_alarm = (rtc_alarm_high << 32) | rtc_alarm_low;
+
+	do_div(rtc_alarm, NSEC_PER_SEC);
+	memset(alrm, 0, sizeof(struct rtc_wkalrm));
+
+	rtc_time64_to_tm(rtc_alarm, &alrm->time);
+
+	if (gf_ioread32(base + TIMER_ALARM_STATUS))
+		alrm->enabled = 1;
+	else
+		alrm->enabled = 0;
+
+	return 0;
+}
+
+static int goldfish_rtc_set_alarm(struct device *dev,
+				  struct rtc_wkalrm *alrm)
+{
+	struct goldfish_rtc *rtcdrv;
+	u64 rtc_alarm64;
+	u64 rtc_status_reg;
+	void __iomem *base;
+
+	rtcdrv = dev_get_drvdata(dev);
+	base = rtcdrv->base;
+
+	if (alrm->enabled) {
+		rtc_alarm64 = rtc_tm_to_time64(&alrm->time) * NSEC_PER_SEC;
+		gf_iowrite32((rtc_alarm64 >> 32), base + TIMER_ALARM_HIGH);
+		gf_iowrite32(rtc_alarm64, base + TIMER_ALARM_LOW);
+		gf_iowrite32(1, base + TIMER_IRQ_ENABLED);
+	} else {
+		/*
+		 * if this function was called with enabled=0
+		 * then it could mean that the application is
+		 * trying to cancel an ongoing alarm
+		 */
+		rtc_status_reg = gf_ioread32(base + TIMER_ALARM_STATUS);
+		if (rtc_status_reg)
+			gf_iowrite32(1, base + TIMER_CLEAR_ALARM);
+	}
+
+	return 0;
+}
+
+static int goldfish_rtc_alarm_irq_enable(struct device *dev,
+					 unsigned int enabled)
+{
+	void __iomem *base;
+	struct goldfish_rtc *rtcdrv;
+
+	rtcdrv = dev_get_drvdata(dev);
+	base = rtcdrv->base;
+
+	if (enabled)
+		gf_iowrite32(1, base + TIMER_IRQ_ENABLED);
+	else
+		gf_iowrite32(0, base + TIMER_IRQ_ENABLED);
+
+	return 0;
+}
+
+static irqreturn_t goldfish_rtc_interrupt(int irq, void *dev_id)
+{
+	struct goldfish_rtc *rtcdrv = dev_id;
+	void __iomem *base = rtcdrv->base;
+
+	gf_iowrite32(1, base + TIMER_CLEAR_INTERRUPT);
+
+	rtc_update_irq(rtcdrv->rtc, 1, RTC_IRQF | RTC_AF);
+
+	return IRQ_HANDLED;
+}
+
+static int goldfish_rtc_read_time(struct device *dev, struct rtc_time *tm)
+{
+	struct goldfish_rtc *rtcdrv;
+	void __iomem *base;
+	u64 time_high;
+	u64 time_low;
+	u64 time;
+
+	rtcdrv = dev_get_drvdata(dev);
+	base = rtcdrv->base;
+
+	time_low = gf_ioread32(base + TIMER_TIME_LOW);
+	time_high = gf_ioread32(base + TIMER_TIME_HIGH);
+	time = (time_high << 32) | time_low;
+
+	do_div(time, NSEC_PER_SEC);
+
+	rtc_time64_to_tm(time, tm);
+
+	return 0;
+}
+
+static int goldfish_rtc_set_time(struct device *dev, struct rtc_time *tm)
+{
+	struct goldfish_rtc *rtcdrv;
+	void __iomem *base;
+	u64 now64;
+
+	rtcdrv = dev_get_drvdata(dev);
+	base = rtcdrv->base;
+
+	now64 = rtc_tm_to_time64(tm) * NSEC_PER_SEC;
+	gf_iowrite32((now64 >> 32), base + TIMER_TIME_HIGH);
+	gf_iowrite32(now64, base + TIMER_TIME_LOW);
+
+	return 0;
+}
+
+static const struct rtc_class_ops goldfish_rtc_ops = {
+	.read_time	= goldfish_rtc_read_time,
+	.set_time	= goldfish_rtc_set_time,
+	.read_alarm	= goldfish_rtc_read_alarm,
+	.set_alarm	= goldfish_rtc_set_alarm,
+	.alarm_irq_enable = goldfish_rtc_alarm_irq_enable
+};
+
+static int goldfish_rtc_probe(struct platform_device *pdev)
+{
+	struct goldfish_rtc *rtcdrv;
+	int err;
+
+	rtcdrv = devm_kzalloc(&pdev->dev, sizeof(*rtcdrv), GFP_KERNEL);
+	if (!rtcdrv)
+		return -ENOMEM;
+
+	platform_set_drvdata(pdev, rtcdrv);
+	rtcdrv->base = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(rtcdrv->base))
+		return PTR_ERR(rtcdrv->base);
+
+	rtcdrv->irq = platform_get_irq(pdev, 0);
+	if (rtcdrv->irq < 0)
+		return -ENODEV;
+
+	rtcdrv->rtc = devm_rtc_allocate_device(&pdev->dev);
+	if (IS_ERR(rtcdrv->rtc))
+		return PTR_ERR(rtcdrv->rtc);
+
+	rtcdrv->rtc->ops = &goldfish_rtc_ops;
+	rtcdrv->rtc->range_max = U64_MAX / NSEC_PER_SEC;
+
+	err = devm_request_irq(&pdev->dev, rtcdrv->irq,
+			       goldfish_rtc_interrupt,
+			       0, pdev->name, rtcdrv);
+	if (err)
+		return err;
+
+	return devm_rtc_register_device(rtcdrv->rtc);
+}
+
+static const struct of_device_id goldfish_rtc_of_match[] = {
+	{ .compatible = "google,goldfish-rtc", },
+	{},
+};
+MODULE_DEVICE_TABLE(of, goldfish_rtc_of_match);
+
+static struct platform_driver goldfish_rtc = {
+	.probe = goldfish_rtc_probe,
+	.driver = {
+		.name = "goldfish_rtc",
+		.of_match_table = goldfish_rtc_of_match,
+	}
+};
+
+module_platform_driver(goldfish_rtc);
+
+MODULE_DESCRIPTION("Android Goldfish Real Time Clock driver");
+MODULE_LICENSE("GPL v2");