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, 349 insertions, 0 deletions

diff --git a/drivers/mfd/max8998.c b/drivers/mfd/max8998.c
new file mode 100644
index 000000000..eb13bbaed
--- /dev/null
+++ b/drivers/mfd/max8998.c
@@ -0,0 +1,349 @@
+// SPDX-License-Identifier: GPL-2.0+
+//
+// max8998.c - mfd core driver for the Maxim 8998
+//
+//  Copyright (C) 2009-2010 Samsung Electronics
+//  Kyungmin Park <kyungmin.park@samsung.com>
+//  Marek Szyprowski <m.szyprowski@samsung.com>
+
+#include <linux/err.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/i2c.h>
+#include <linux/interrupt.h>
+#include <linux/of.h>
+#include <linux/of_irq.h>
+#include <linux/pm_runtime.h>
+#include <linux/mutex.h>
+#include <linux/mfd/core.h>
+#include <linux/mfd/max8998.h>
+#include <linux/mfd/max8998-private.h>
+
+#define RTC_I2C_ADDR		(0x0c >> 1)
+
+static const struct mfd_cell max8998_devs[] = {
+	{
+		.name = "max8998-pmic",
+	}, {
+		.name = "max8998-rtc",
+	}, {
+		.name = "max8998-battery",
+	},
+};
+
+static const struct mfd_cell lp3974_devs[] = {
+	{
+		.name = "lp3974-pmic",
+	}, {
+		.name = "lp3974-rtc",
+	},
+};
+
+int max8998_read_reg(struct i2c_client *i2c, u8 reg, u8 *dest)
+{
+	struct max8998_dev *max8998 = i2c_get_clientdata(i2c);
+	int ret;
+
+	mutex_lock(&max8998->iolock);
+	ret = i2c_smbus_read_byte_data(i2c, reg);
+	mutex_unlock(&max8998->iolock);
+	if (ret < 0)
+		return ret;
+
+	ret &= 0xff;
+	*dest = ret;
+	return 0;
+}
+EXPORT_SYMBOL(max8998_read_reg);
+
+int max8998_bulk_read(struct i2c_client *i2c, u8 reg, int count, u8 *buf)
+{
+	struct max8998_dev *max8998 = i2c_get_clientdata(i2c);
+	int ret;
+
+	mutex_lock(&max8998->iolock);
+	ret = i2c_smbus_read_i2c_block_data(i2c, reg, count, buf);
+	mutex_unlock(&max8998->iolock);
+	if (ret < 0)
+		return ret;
+
+	return 0;
+}
+EXPORT_SYMBOL(max8998_bulk_read);
+
+int max8998_write_reg(struct i2c_client *i2c, u8 reg, u8 value)
+{
+	struct max8998_dev *max8998 = i2c_get_clientdata(i2c);
+	int ret;
+
+	mutex_lock(&max8998->iolock);
+	ret = i2c_smbus_write_byte_data(i2c, reg, value);
+	mutex_unlock(&max8998->iolock);
+	return ret;
+}
+EXPORT_SYMBOL(max8998_write_reg);
+
+int max8998_bulk_write(struct i2c_client *i2c, u8 reg, int count, u8 *buf)
+{
+	struct max8998_dev *max8998 = i2c_get_clientdata(i2c);
+	int ret;
+
+	mutex_lock(&max8998->iolock);
+	ret = i2c_smbus_write_i2c_block_data(i2c, reg, count, buf);
+	mutex_unlock(&max8998->iolock);
+	if (ret < 0)
+		return ret;
+
+	return 0;
+}
+EXPORT_SYMBOL(max8998_bulk_write);
+
+int max8998_update_reg(struct i2c_client *i2c, u8 reg, u8 val, u8 mask)
+{
+	struct max8998_dev *max8998 = i2c_get_clientdata(i2c);
+	int ret;
+
+	mutex_lock(&max8998->iolock);
+	ret = i2c_smbus_read_byte_data(i2c, reg);
+	if (ret >= 0) {
+		u8 old_val = ret & 0xff;
+		u8 new_val = (val & mask) | (old_val & (~mask));
+		ret = i2c_smbus_write_byte_data(i2c, reg, new_val);
+	}
+	mutex_unlock(&max8998->iolock);
+	return ret;
+}
+EXPORT_SYMBOL(max8998_update_reg);
+
+#ifdef CONFIG_OF
+static const struct of_device_id max8998_dt_match[] = {
+	{ .compatible = "maxim,max8998", .data = (void *)TYPE_MAX8998 },
+	{ .compatible = "national,lp3974", .data = (void *)TYPE_LP3974 },
+	{ .compatible = "ti,lp3974", .data = (void *)TYPE_LP3974 },
+	{},
+};
+#endif
+
+/*
+ * Only the common platform data elements for max8998 are parsed here from the
+ * device tree. Other sub-modules of max8998 such as pmic, rtc and others have
+ * to parse their own platform data elements from device tree.
+ *
+ * The max8998 platform data structure is instantiated here and the drivers for
+ * the sub-modules need not instantiate another instance while parsing their
+ * platform data.
+ */
+static struct max8998_platform_data *max8998_i2c_parse_dt_pdata(
+							struct device *dev)
+{
+	struct max8998_platform_data *pd;
+
+	pd = devm_kzalloc(dev, sizeof(*pd), GFP_KERNEL);
+	if (!pd)
+		return ERR_PTR(-ENOMEM);
+
+	pd->ono = irq_of_parse_and_map(dev->of_node, 1);
+
+	/*
+	 * ToDo: the 'wakeup' member in the platform data is more of a linux
+	 * specfic information. Hence, there is no binding for that yet and
+	 * not parsed here.
+	 */
+	return pd;
+}
+
+static int max8998_i2c_probe(struct i2c_client *i2c)
+{
+	struct max8998_platform_data *pdata = dev_get_platdata(&i2c->dev);
+	struct max8998_dev *max8998;
+	int ret = 0;
+
+	max8998 = devm_kzalloc(&i2c->dev, sizeof(struct max8998_dev),
+				GFP_KERNEL);
+	if (max8998 == NULL)
+		return -ENOMEM;
+
+	if (IS_ENABLED(CONFIG_OF) && i2c->dev.of_node) {
+		pdata = max8998_i2c_parse_dt_pdata(&i2c->dev);
+		if (IS_ERR(pdata))
+			return PTR_ERR(pdata);
+	}
+
+	i2c_set_clientdata(i2c, max8998);
+	max8998->dev = &i2c->dev;
+	max8998->i2c = i2c;
+	max8998->irq = i2c->irq;
+	max8998->type = (uintptr_t)i2c_get_match_data(i2c);
+	max8998->pdata = pdata;
+	if (pdata) {
+		max8998->ono = pdata->ono;
+		max8998->irq_base = pdata->irq_base;
+		max8998->wakeup = pdata->wakeup;
+	}
+	mutex_init(&max8998->iolock);
+
+	max8998->rtc = i2c_new_dummy_device(i2c->adapter, RTC_I2C_ADDR);
+	if (IS_ERR(max8998->rtc)) {
+		dev_err(&i2c->dev, "Failed to allocate I2C device for RTC\n");
+		return PTR_ERR(max8998->rtc);
+	}
+	i2c_set_clientdata(max8998->rtc, max8998);
+
+	max8998_irq_init(max8998);
+
+	pm_runtime_set_active(max8998->dev);
+
+	switch (max8998->type) {
+	case TYPE_LP3974:
+		ret = mfd_add_devices(max8998->dev, -1,
+				      lp3974_devs, ARRAY_SIZE(lp3974_devs),
+				      NULL, 0, NULL);
+		break;
+	case TYPE_MAX8998:
+		ret = mfd_add_devices(max8998->dev, -1,
+				      max8998_devs, ARRAY_SIZE(max8998_devs),
+				      NULL, 0, NULL);
+		break;
+	default:
+		ret = -EINVAL;
+	}
+
+	if (ret < 0)
+		goto err;
+
+	device_init_wakeup(max8998->dev, max8998->wakeup);
+
+	return ret;
+
+err:
+	mfd_remove_devices(max8998->dev);
+	max8998_irq_exit(max8998);
+	i2c_unregister_device(max8998->rtc);
+	return ret;
+}
+
+static const struct i2c_device_id max8998_i2c_id[] = {
+	{ "max8998", TYPE_MAX8998 },
+	{ "lp3974", TYPE_LP3974},
+	{ }
+};
+
+static int max8998_suspend(struct device *dev)
+{
+	struct i2c_client *i2c = to_i2c_client(dev);
+	struct max8998_dev *max8998 = i2c_get_clientdata(i2c);
+
+	if (device_may_wakeup(dev))
+		enable_irq_wake(max8998->irq);
+	return 0;
+}
+
+static int max8998_resume(struct device *dev)
+{
+	struct i2c_client *i2c = to_i2c_client(dev);
+	struct max8998_dev *max8998 = i2c_get_clientdata(i2c);
+
+	if (device_may_wakeup(dev))
+		disable_irq_wake(max8998->irq);
+	/*
+	 * In LP3974, if IRQ registers are not "read & clear"
+	 * when it's set during sleep, the interrupt becomes
+	 * disabled.
+	 */
+	return max8998_irq_resume(i2c_get_clientdata(i2c));
+}
+
+struct max8998_reg_dump {
+	u8	addr;
+	u8	val;
+};
+#define SAVE_ITEM(x)	{ .addr = (x), .val = 0x0, }
+static struct max8998_reg_dump max8998_dump[] = {
+	SAVE_ITEM(MAX8998_REG_IRQM1),
+	SAVE_ITEM(MAX8998_REG_IRQM2),
+	SAVE_ITEM(MAX8998_REG_IRQM3),
+	SAVE_ITEM(MAX8998_REG_IRQM4),
+	SAVE_ITEM(MAX8998_REG_STATUSM1),
+	SAVE_ITEM(MAX8998_REG_STATUSM2),
+	SAVE_ITEM(MAX8998_REG_CHGR1),
+	SAVE_ITEM(MAX8998_REG_CHGR2),
+	SAVE_ITEM(MAX8998_REG_LDO_ACTIVE_DISCHARGE1),
+	SAVE_ITEM(MAX8998_REG_LDO_ACTIVE_DISCHARGE1),
+	SAVE_ITEM(MAX8998_REG_BUCK_ACTIVE_DISCHARGE3),
+	SAVE_ITEM(MAX8998_REG_ONOFF1),
+	SAVE_ITEM(MAX8998_REG_ONOFF2),
+	SAVE_ITEM(MAX8998_REG_ONOFF3),
+	SAVE_ITEM(MAX8998_REG_ONOFF4),
+	SAVE_ITEM(MAX8998_REG_BUCK1_VOLTAGE1),
+	SAVE_ITEM(MAX8998_REG_BUCK1_VOLTAGE2),
+	SAVE_ITEM(MAX8998_REG_BUCK1_VOLTAGE3),
+	SAVE_ITEM(MAX8998_REG_BUCK1_VOLTAGE4),
+	SAVE_ITEM(MAX8998_REG_BUCK2_VOLTAGE1),
+	SAVE_ITEM(MAX8998_REG_BUCK2_VOLTAGE2),
+	SAVE_ITEM(MAX8998_REG_LDO2_LDO3),
+	SAVE_ITEM(MAX8998_REG_LDO4),
+	SAVE_ITEM(MAX8998_REG_LDO5),
+	SAVE_ITEM(MAX8998_REG_LDO6),
+	SAVE_ITEM(MAX8998_REG_LDO7),
+	SAVE_ITEM(MAX8998_REG_LDO8_LDO9),
+	SAVE_ITEM(MAX8998_REG_LDO10_LDO11),
+	SAVE_ITEM(MAX8998_REG_LDO12),
+	SAVE_ITEM(MAX8998_REG_LDO13),
+	SAVE_ITEM(MAX8998_REG_LDO14),
+	SAVE_ITEM(MAX8998_REG_LDO15),
+	SAVE_ITEM(MAX8998_REG_LDO16),
+	SAVE_ITEM(MAX8998_REG_LDO17),
+	SAVE_ITEM(MAX8998_REG_BKCHR),
+	SAVE_ITEM(MAX8998_REG_LBCNFG1),
+	SAVE_ITEM(MAX8998_REG_LBCNFG2),
+};
+/* Save registers before hibernation */
+static int max8998_freeze(struct device *dev)
+{
+	struct i2c_client *i2c = to_i2c_client(dev);
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(max8998_dump); i++)
+		max8998_read_reg(i2c, max8998_dump[i].addr,
+				&max8998_dump[i].val);
+
+	return 0;
+}
+
+/* Restore registers after hibernation */
+static int max8998_restore(struct device *dev)
+{
+	struct i2c_client *i2c = to_i2c_client(dev);
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(max8998_dump); i++)
+		max8998_write_reg(i2c, max8998_dump[i].addr,
+				max8998_dump[i].val);
+
+	return 0;
+}
+
+static const struct dev_pm_ops max8998_pm = {
+	.suspend = max8998_suspend,
+	.resume = max8998_resume,
+	.freeze = max8998_freeze,
+	.restore = max8998_restore,
+};
+
+static struct i2c_driver max8998_i2c_driver = {
+	.driver = {
+		   .name = "max8998",
+		   .pm = &max8998_pm,
+		   .suppress_bind_attrs = true,
+		   .of_match_table = of_match_ptr(max8998_dt_match),
+	},
+	.probe = max8998_i2c_probe,
+	.id_table = max8998_i2c_id,
+};
+
+static int __init max8998_i2c_init(void)
+{
+	return i2c_add_driver(&max8998_i2c_driver);
+}
+/* init early so consumer devices can complete system boot */
+subsys_initcall(max8998_i2c_init);