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, 205 insertions, 0 deletions

diff --git a/drivers/clk/qcom/clk-ref.c b/drivers/clk/qcom/clk-ref.c
new file mode 100644
index 000000000..3efdf6749
--- /dev/null
+++ b/drivers/clk/qcom/clk-ref.c
@@ -0,0 +1,205 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2026, Qualcomm Technologies, Inc. and/or its subsidiaries.
+ */
+
+#include <linux/clk-provider.h>
+#include <linux/clk/qcom.h>
+#include <linux/delay.h>
+#include <linux/err.h>
+#include <linux/export.h>
+#include <linux/platform_device.h>
+#include <linux/regmap.h>
+#include <linux/regulator/consumer.h>
+#include <linux/slab.h>
+
+#define QCOM_CLK_REF_EN_MASK BIT(0)
+
+struct qcom_clk_ref_provider {
+	size_t num_refs;
+	struct qcom_clk_ref refs[] __counted_by(num_refs);
+};
+
+static inline struct qcom_clk_ref *to_qcom_clk_ref(struct clk_hw *hw)
+{
+	return container_of(hw, struct qcom_clk_ref, hw);
+}
+
+static const struct clk_parent_data qcom_clk_ref_parent_data = {
+	.index = 0,
+};
+
+static int qcom_clk_ref_prepare(struct clk_hw *hw)
+{
+	struct qcom_clk_ref *rclk = to_qcom_clk_ref(hw);
+	int ret;
+
+	if (!rclk->desc.num_regulators)
+		return 0;
+
+	ret = regulator_bulk_enable(rclk->desc.num_regulators, rclk->regulators);
+	if (ret)
+		pr_err("Failed to enable regulators for %s: %d\n",
+		       clk_hw_get_name(hw), ret);
+
+	return ret;
+}
+
+static void qcom_clk_ref_unprepare(struct clk_hw *hw)
+{
+	struct qcom_clk_ref *rclk = to_qcom_clk_ref(hw);
+
+	if (rclk->desc.num_regulators)
+		regulator_bulk_disable(rclk->desc.num_regulators, rclk->regulators);
+}
+
+static int qcom_clk_ref_enable(struct clk_hw *hw)
+{
+	struct qcom_clk_ref *rclk = to_qcom_clk_ref(hw);
+	int ret;
+
+	ret = regmap_set_bits(rclk->regmap, rclk->desc.offset, QCOM_CLK_REF_EN_MASK);
+	if (ret)
+		return ret;
+
+	udelay(10);
+
+	return 0;
+}
+
+static void qcom_clk_ref_disable(struct clk_hw *hw)
+{
+	struct qcom_clk_ref *rclk = to_qcom_clk_ref(hw);
+
+	regmap_clear_bits(rclk->regmap, rclk->desc.offset, QCOM_CLK_REF_EN_MASK);
+	udelay(10);
+}
+
+static int qcom_clk_ref_is_enabled(struct clk_hw *hw)
+{
+	struct qcom_clk_ref *rclk = to_qcom_clk_ref(hw);
+	u32 val;
+	int ret;
+
+	ret = regmap_read(rclk->regmap, rclk->desc.offset, &val);
+	if (ret)
+		return 0;
+
+	return !!(val & QCOM_CLK_REF_EN_MASK);
+}
+
+static const struct clk_ops qcom_clk_ref_ops = {
+	.prepare = qcom_clk_ref_prepare,
+	.unprepare = qcom_clk_ref_unprepare,
+	.enable = qcom_clk_ref_enable,
+	.disable = qcom_clk_ref_disable,
+	.is_enabled = qcom_clk_ref_is_enabled,
+};
+
+static int qcom_clk_ref_register(struct device *dev, struct regmap *regmap,
+				 struct qcom_clk_ref *clk_refs,
+				 const struct qcom_clk_ref_desc * const *descs,
+				 size_t num_clk_refs)
+{
+	const struct qcom_clk_ref_desc *desc;
+	struct clk_init_data init_data = {};
+	struct qcom_clk_ref *clk_ref;
+	size_t clk_idx;
+	unsigned int i;
+	int ret;
+
+	for (clk_idx = 0; clk_idx < num_clk_refs; clk_idx++) {
+		clk_ref = &clk_refs[clk_idx];
+		desc = descs[clk_idx];
+
+		/* Skip unpopulated indices; the array is indexed by clock ID. */
+		if (!desc)
+			continue;
+
+		if (WARN_ON(!desc->name))
+			continue;
+
+		clk_ref->regmap = regmap;
+		clk_ref->desc = *desc;
+
+		if (clk_ref->desc.num_regulators) {
+			clk_ref->regulators = devm_kcalloc(dev, clk_ref->desc.num_regulators,
+							   sizeof(*clk_ref->regulators),
+							   GFP_KERNEL);
+			if (!clk_ref->regulators)
+				return -ENOMEM;
+
+			for (i = 0; i < clk_ref->desc.num_regulators; i++)
+				clk_ref->regulators[i].supply =
+					clk_ref->desc.regulator_names[i];
+
+			ret = devm_regulator_bulk_get(dev, clk_ref->desc.num_regulators,
+						      clk_ref->regulators);
+			if (ret)
+				return dev_err_probe(dev, ret,
+						     "Failed to get regulators for %s\n",
+						     clk_ref->desc.name);
+		}
+
+		init_data.name = clk_ref->desc.name;
+		init_data.parent_data = &qcom_clk_ref_parent_data;
+		init_data.num_parents = 1;
+		init_data.ops = &qcom_clk_ref_ops;
+		clk_ref->hw.init = &init_data;
+
+		ret = devm_clk_hw_register(dev, &clk_ref->hw);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static struct clk_hw *qcom_clk_ref_provider_get(struct of_phandle_args *clkspec, void *data)
+{
+	struct qcom_clk_ref_provider *provider = data;
+	unsigned int idx = clkspec->args[0];
+
+	if (idx >= provider->num_refs)
+		return ERR_PTR(-EINVAL);
+
+	if (!provider->refs[idx].regmap)
+		return ERR_PTR(-ENOENT);
+
+	return &provider->refs[idx].hw;
+}
+
+int qcom_clk_ref_probe(struct platform_device *pdev,
+		       const struct regmap_config *config,
+		       const struct qcom_clk_ref_desc * const *descs,
+		       size_t num_clk_refs)
+{
+	struct qcom_clk_ref_provider *provider;
+	struct device *dev = &pdev->dev;
+	struct regmap *regmap;
+	void __iomem *base;
+	int ret;
+
+	base = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(base))
+		return PTR_ERR(base);
+
+	regmap = devm_regmap_init_mmio(dev, base, config);
+	if (IS_ERR(regmap))
+		return PTR_ERR(regmap);
+
+	provider = devm_kzalloc(dev, struct_size(provider, refs, num_clk_refs),
+				GFP_KERNEL);
+	if (!provider)
+		return -ENOMEM;
+
+	provider->num_refs = num_clk_refs;
+
+	ret = qcom_clk_ref_register(dev, regmap, provider->refs, descs,
+				    provider->num_refs);
+	if (ret)
+		return ret;
+
+	return devm_of_clk_add_hw_provider(dev, qcom_clk_ref_provider_get, provider);
+}
+EXPORT_SYMBOL_GPL(qcom_clk_ref_probe);