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, 187 insertions, 0 deletions

diff --git a/drivers/mfd/rohm-bd9576.c b/drivers/mfd/rohm-bd9576.c
new file mode 100644
index 000000000..17323ae39
--- /dev/null
+++ b/drivers/mfd/rohm-bd9576.c
@@ -0,0 +1,187 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2021 ROHM Semiconductors
+ *
+ * ROHM BD9576MUF and BD9573MUF PMIC driver
+ */
+
+#include <linux/i2c.h>
+#include <linux/interrupt.h>
+#include <linux/ioport.h>
+#include <linux/irq.h>
+#include <linux/mfd/core.h>
+#include <linux/mfd/rohm-bd957x.h>
+#include <linux/mfd/rohm-generic.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/regmap.h>
+#include <linux/types.h>
+
+enum {
+	BD957X_REGULATOR_CELL,
+	BD957X_WDT_CELL,
+};
+
+/*
+ * Due to the BD9576MUF nasty IRQ behaviour we don't always populate IRQs.
+ * These will be added to regulator resources only if IRQ information for the
+ * PMIC is populated in device-tree.
+ */
+static const struct resource bd9576_regulator_irqs[] = {
+	DEFINE_RES_IRQ_NAMED(BD9576_INT_THERM, "bd9576-temp"),
+	DEFINE_RES_IRQ_NAMED(BD9576_INT_OVD, "bd9576-ovd"),
+	DEFINE_RES_IRQ_NAMED(BD9576_INT_UVD, "bd9576-uvd"),
+};
+
+static struct mfd_cell bd9573_mfd_cells[] = {
+	[BD957X_REGULATOR_CELL]	= { .name = "bd9573-regulator", },
+	[BD957X_WDT_CELL]	= { .name = "bd9576-wdt", },
+};
+
+static struct mfd_cell bd9576_mfd_cells[] = {
+	[BD957X_REGULATOR_CELL]	= { .name = "bd9576-regulator", },
+	[BD957X_WDT_CELL]	= { .name = "bd9576-wdt", },
+};
+
+static const struct regmap_range volatile_ranges[] = {
+	regmap_reg_range(BD957X_REG_SMRB_ASSERT, BD957X_REG_SMRB_ASSERT),
+	regmap_reg_range(BD957X_REG_PMIC_INTERNAL_STAT,
+			 BD957X_REG_PMIC_INTERNAL_STAT),
+	regmap_reg_range(BD957X_REG_INT_THERM_STAT, BD957X_REG_INT_THERM_STAT),
+	regmap_reg_range(BD957X_REG_INT_OVP_STAT, BD957X_REG_INT_SYS_STAT),
+	regmap_reg_range(BD957X_REG_INT_MAIN_STAT, BD957X_REG_INT_MAIN_STAT),
+};
+
+static const struct regmap_access_table volatile_regs = {
+	.yes_ranges = &volatile_ranges[0],
+	.n_yes_ranges = ARRAY_SIZE(volatile_ranges),
+};
+
+static const struct regmap_config bd957x_regmap = {
+	.reg_bits = 8,
+	.val_bits = 8,
+	.volatile_table = &volatile_regs,
+	.max_register = BD957X_MAX_REGISTER,
+	.cache_type = REGCACHE_MAPLE,
+};
+
+static const struct regmap_irq bd9576_irqs[] = {
+	REGMAP_IRQ_REG(BD9576_INT_THERM, 0, BD957X_MASK_INT_MAIN_THERM),
+	REGMAP_IRQ_REG(BD9576_INT_OVP, 0, BD957X_MASK_INT_MAIN_OVP),
+	REGMAP_IRQ_REG(BD9576_INT_SCP, 0, BD957X_MASK_INT_MAIN_SCP),
+	REGMAP_IRQ_REG(BD9576_INT_OCP, 0, BD957X_MASK_INT_MAIN_OCP),
+	REGMAP_IRQ_REG(BD9576_INT_OVD, 0, BD957X_MASK_INT_MAIN_OVD),
+	REGMAP_IRQ_REG(BD9576_INT_UVD, 0, BD957X_MASK_INT_MAIN_UVD),
+	REGMAP_IRQ_REG(BD9576_INT_UVP, 0, BD957X_MASK_INT_MAIN_UVP),
+	REGMAP_IRQ_REG(BD9576_INT_SYS, 0, BD957X_MASK_INT_MAIN_SYS),
+};
+
+static const struct regmap_irq_chip bd9576_irq_chip = {
+	.name = "bd9576_irq",
+	.irqs = &bd9576_irqs[0],
+	.num_irqs = ARRAY_SIZE(bd9576_irqs),
+	.status_base = BD957X_REG_INT_MAIN_STAT,
+	.mask_base = BD957X_REG_INT_MAIN_MASK,
+	.ack_base = BD957X_REG_INT_MAIN_STAT,
+	.init_ack_masked = true,
+	.num_regs = 1,
+	.irq_reg_stride = 1,
+};
+
+static int bd957x_i2c_probe(struct i2c_client *i2c)
+{
+	int ret;
+	struct regmap *regmap;
+	struct mfd_cell *cells;
+	int num_cells;
+	unsigned long chip_type;
+	struct irq_domain *domain;
+	bool usable_irqs;
+
+	chip_type = (unsigned long)of_device_get_match_data(&i2c->dev);
+
+	switch (chip_type) {
+	case ROHM_CHIP_TYPE_BD9576:
+		cells = bd9576_mfd_cells;
+		num_cells = ARRAY_SIZE(bd9576_mfd_cells);
+		usable_irqs = !!i2c->irq;
+		break;
+	case ROHM_CHIP_TYPE_BD9573:
+		cells = bd9573_mfd_cells;
+		num_cells = ARRAY_SIZE(bd9573_mfd_cells);
+		/*
+		 * BD9573 only supports fatal IRQs which we can not handle
+		 * because SoC is going to lose the power.
+		 */
+		usable_irqs = false;
+		break;
+	default:
+		dev_err(&i2c->dev, "Unknown device type");
+		return -EINVAL;
+	}
+
+	regmap = devm_regmap_init_i2c(i2c, &bd957x_regmap);
+	if (IS_ERR(regmap))
+		return dev_err_probe(&i2c->dev, PTR_ERR(regmap),
+				     "Failed to initialize Regmap\n");
+
+	/*
+	 * BD9576 behaves badly. It kepts IRQ line asserted for the whole
+	 * duration of detected HW condition (like over temperature). So we
+	 * don't require IRQ to be populated.
+	 * If IRQ information is not given, then we mask all IRQs and do not
+	 * provide IRQ resources to regulator driver - which then just omits
+	 * the notifiers.
+	 */
+	if (usable_irqs) {
+		struct regmap_irq_chip_data *irq_data;
+		struct mfd_cell *regulators;
+
+		regulators = &bd9576_mfd_cells[BD957X_REGULATOR_CELL];
+		regulators->resources = bd9576_regulator_irqs;
+		regulators->num_resources = ARRAY_SIZE(bd9576_regulator_irqs);
+
+		ret = devm_regmap_add_irq_chip(&i2c->dev, regmap, i2c->irq,
+					       IRQF_ONESHOT, 0,
+					       &bd9576_irq_chip, &irq_data);
+		if (ret)
+			return dev_err_probe(&i2c->dev, ret,
+					     "Failed to add IRQ chip\n");
+
+		domain = regmap_irq_get_domain(irq_data);
+	} else {
+		ret = regmap_update_bits(regmap, BD957X_REG_INT_MAIN_MASK,
+					 BD957X_MASK_INT_ALL,
+					 BD957X_MASK_INT_ALL);
+		if (ret)
+			return ret;
+		domain = NULL;
+	}
+
+	ret = devm_mfd_add_devices(&i2c->dev, PLATFORM_DEVID_AUTO, cells,
+				   num_cells, NULL, 0, domain);
+	if (ret)
+		dev_err_probe(&i2c->dev, ret, "Failed to create subdevices\n");
+
+	return ret;
+}
+
+static const struct of_device_id bd957x_of_match[] = {
+	{ .compatible = "rohm,bd9576", .data = (void *)ROHM_CHIP_TYPE_BD9576, },
+	{ .compatible = "rohm,bd9573", .data = (void *)ROHM_CHIP_TYPE_BD9573, },
+	{ },
+};
+MODULE_DEVICE_TABLE(of, bd957x_of_match);
+
+static struct i2c_driver bd957x_drv = {
+	.driver = {
+		.name = "rohm-bd957x",
+		.of_match_table = bd957x_of_match,
+	},
+	.probe = bd957x_i2c_probe,
+};
+module_i2c_driver(bd957x_drv);
+
+MODULE_AUTHOR("Matti Vaittinen <matti.vaittinen@fi.rohmeurope.com>");
+MODULE_DESCRIPTION("ROHM BD9576MUF and BD9573MUF Power Management IC driver");
+MODULE_LICENSE("GPL");