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, 194 insertions, 0 deletions

diff --git a/drivers/rtc/rtc-wilco-ec.c b/drivers/rtc/rtc-wilco-ec.c
new file mode 100644
index 000000000..2a205a646
--- /dev/null
+++ b/drivers/rtc/rtc-wilco-ec.c
@@ -0,0 +1,194 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * RTC interface for Wilco Embedded Controller with R/W abilities
+ *
+ * Copyright 2018 Google LLC
+ *
+ * The corresponding platform device is typically registered in
+ * drivers/platform/chrome/wilco_ec/core.c
+ */
+
+#include <linux/bcd.h>
+#include <linux/err.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
+#include <linux/platform_data/wilco-ec.h>
+#include <linux/rtc.h>
+#include <linux/timekeeping.h>
+
+#define EC_COMMAND_CMOS			0x7c
+#define EC_CMOS_TOD_WRITE		0x02
+#define EC_CMOS_TOD_READ		0x08
+
+/* Message sent to the EC to request the current time. */
+struct ec_rtc_read_request {
+	u8 command;
+	u8 reserved;
+	u8 param;
+} __packed;
+static struct ec_rtc_read_request read_rq = {
+	.command = EC_COMMAND_CMOS,
+	.param = EC_CMOS_TOD_READ,
+};
+
+/**
+ * struct ec_rtc_read_response - Format of RTC returned by EC.
+ * @reserved: Unused byte
+ * @second: Second value (0..59)
+ * @minute: Minute value (0..59)
+ * @hour: Hour value (0..23)
+ * @day: Day value (1..31)
+ * @month: Month value (1..12)
+ * @year: Year value (full year % 100)
+ * @century: Century value (full year / 100)
+ *
+ * All values are presented in binary (not BCD).
+ */
+struct ec_rtc_read_response {
+	u8 reserved;
+	u8 second;
+	u8 minute;
+	u8 hour;
+	u8 day;
+	u8 month;
+	u8 year;
+	u8 century;
+} __packed;
+
+/**
+ * struct ec_rtc_write_request - Format of RTC sent to the EC.
+ * @command: Always EC_COMMAND_CMOS
+ * @reserved: Unused byte
+ * @param: Always EC_CMOS_TOD_WRITE
+ * @century: Century value (full year / 100)
+ * @year: Year value (full year % 100)
+ * @month: Month value (1..12)
+ * @day: Day value (1..31)
+ * @hour: Hour value (0..23)
+ * @minute: Minute value (0..59)
+ * @second: Second value (0..59)
+ * @weekday: Day of the week (0=Saturday)
+ *
+ * All values are presented in BCD.
+ */
+struct ec_rtc_write_request {
+	u8 command;
+	u8 reserved;
+	u8 param;
+	u8 century;
+	u8 year;
+	u8 month;
+	u8 day;
+	u8 hour;
+	u8 minute;
+	u8 second;
+	u8 weekday;
+} __packed;
+
+static int wilco_ec_rtc_read(struct device *dev, struct rtc_time *tm)
+{
+	struct wilco_ec_device *ec = dev_get_drvdata(dev->parent);
+	struct ec_rtc_read_response rtc;
+	struct wilco_ec_message msg;
+	int ret;
+
+	memset(&msg, 0, sizeof(msg));
+	msg.type = WILCO_EC_MSG_LEGACY;
+	msg.request_data = &read_rq;
+	msg.request_size = sizeof(read_rq);
+	msg.response_data = &rtc;
+	msg.response_size = sizeof(rtc);
+
+	ret = wilco_ec_mailbox(ec, &msg);
+	if (ret < 0)
+		return ret;
+
+	tm->tm_sec	= rtc.second;
+	tm->tm_min	= rtc.minute;
+	tm->tm_hour	= rtc.hour;
+	tm->tm_mday	= rtc.day;
+	tm->tm_mon	= rtc.month - 1;
+	tm->tm_year	= rtc.year + (rtc.century * 100) - 1900;
+	/* Ignore other tm fields, man rtc says userspace shouldn't use them. */
+
+	if (rtc_valid_tm(tm)) {
+		dev_err(dev, "Time from RTC is invalid: %ptRr\n", tm);
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int wilco_ec_rtc_write(struct device *dev, struct rtc_time *tm)
+{
+	struct wilco_ec_device *ec = dev_get_drvdata(dev->parent);
+	struct ec_rtc_write_request rtc;
+	struct wilco_ec_message msg;
+	int year = tm->tm_year + 1900;
+	/*
+	 * Convert from 0=Sunday to 0=Saturday for the EC
+	 * We DO need to set weekday because the EC controls battery charging
+	 * schedules that depend on the day of the week.
+	 */
+	int wday = tm->tm_wday == 6 ? 0 : tm->tm_wday + 1;
+	int ret;
+
+	rtc.command	= EC_COMMAND_CMOS;
+	rtc.param	= EC_CMOS_TOD_WRITE;
+	rtc.century	= bin2bcd(year / 100);
+	rtc.year	= bin2bcd(year % 100);
+	rtc.month	= bin2bcd(tm->tm_mon + 1);
+	rtc.day		= bin2bcd(tm->tm_mday);
+	rtc.hour	= bin2bcd(tm->tm_hour);
+	rtc.minute	= bin2bcd(tm->tm_min);
+	rtc.second	= bin2bcd(tm->tm_sec);
+	rtc.weekday	= bin2bcd(wday);
+
+	memset(&msg, 0, sizeof(msg));
+	msg.type = WILCO_EC_MSG_LEGACY;
+	msg.request_data = &rtc;
+	msg.request_size = sizeof(rtc);
+
+	ret = wilco_ec_mailbox(ec, &msg);
+	if (ret < 0)
+		return ret;
+
+	return 0;
+}
+
+static const struct rtc_class_ops wilco_ec_rtc_ops = {
+	.read_time = wilco_ec_rtc_read,
+	.set_time = wilco_ec_rtc_write,
+};
+
+static int wilco_ec_rtc_probe(struct platform_device *pdev)
+{
+	struct rtc_device *rtc;
+
+	rtc = devm_rtc_allocate_device(&pdev->dev);
+	if (IS_ERR(rtc))
+		return PTR_ERR(rtc);
+
+	rtc->ops = &wilco_ec_rtc_ops;
+	/* EC only supports this century */
+	rtc->range_min = RTC_TIMESTAMP_BEGIN_2000;
+	rtc->range_max = RTC_TIMESTAMP_END_2099;
+	rtc->owner = THIS_MODULE;
+
+	return devm_rtc_register_device(rtc);
+}
+
+static struct platform_driver wilco_ec_rtc_driver = {
+	.driver = {
+		.name = "rtc-wilco-ec",
+	},
+	.probe = wilco_ec_rtc_probe,
+};
+
+module_platform_driver(wilco_ec_rtc_driver);
+
+MODULE_ALIAS("platform:rtc-wilco-ec");
+MODULE_AUTHOR("Nick Crews <ncrews@chromium.org>");
+MODULE_LICENSE("GPL v2");
+MODULE_DESCRIPTION("Wilco EC RTC driver");