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, 226 insertions, 0 deletions

diff --git a/drivers/clk/ux500/clk-sysctrl.c b/drivers/clk/ux500/clk-sysctrl.c
new file mode 100644
index 000000000..ba3258c88
--- /dev/null
+++ b/drivers/clk/ux500/clk-sysctrl.c
@@ -0,0 +1,226 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Sysctrl clock implementation for ux500 platform.
+ *
+ * Copyright (C) 2013 ST-Ericsson SA
+ * Author: Ulf Hansson <ulf.hansson@linaro.org>
+ */
+
+#include <linux/clk-provider.h>
+#include <linux/mfd/abx500/ab8500-sysctrl.h>
+#include <linux/device.h>
+#include <linux/slab.h>
+#include <linux/delay.h>
+#include <linux/io.h>
+#include <linux/err.h>
+#include "clk.h"
+
+#define SYSCTRL_MAX_NUM_PARENTS 4
+
+#define to_clk_sysctrl(_hw) container_of(_hw, struct clk_sysctrl, hw)
+
+struct clk_sysctrl {
+	struct clk_hw hw;
+	struct device *dev;
+	u8 parent_index;
+	u16 reg_sel[SYSCTRL_MAX_NUM_PARENTS];
+	u8 reg_mask[SYSCTRL_MAX_NUM_PARENTS];
+	u8 reg_bits[SYSCTRL_MAX_NUM_PARENTS];
+	unsigned long rate;
+	unsigned long enable_delay_us;
+};
+
+/* Sysctrl clock operations. */
+
+static int clk_sysctrl_prepare(struct clk_hw *hw)
+{
+	int ret;
+	struct clk_sysctrl *clk = to_clk_sysctrl(hw);
+
+	ret = ab8500_sysctrl_write(clk->reg_sel[0], clk->reg_mask[0],
+				clk->reg_bits[0]);
+
+	if (!ret && clk->enable_delay_us)
+		usleep_range(clk->enable_delay_us, clk->enable_delay_us +
+			     (clk->enable_delay_us >> 2));
+
+	return ret;
+}
+
+static void clk_sysctrl_unprepare(struct clk_hw *hw)
+{
+	struct clk_sysctrl *clk = to_clk_sysctrl(hw);
+	if (ab8500_sysctrl_clear(clk->reg_sel[0], clk->reg_mask[0]))
+		dev_err(clk->dev, "clk_sysctrl: %s fail to clear %s.\n",
+			__func__, clk_hw_get_name(hw));
+}
+
+static unsigned long clk_sysctrl_recalc_rate(struct clk_hw *hw,
+					unsigned long parent_rate)
+{
+	struct clk_sysctrl *clk = to_clk_sysctrl(hw);
+	return clk->rate;
+}
+
+static int clk_sysctrl_set_parent(struct clk_hw *hw, u8 index)
+{
+	struct clk_sysctrl *clk = to_clk_sysctrl(hw);
+	u8 old_index = clk->parent_index;
+	int ret = 0;
+
+	if (clk->reg_sel[old_index]) {
+		ret = ab8500_sysctrl_clear(clk->reg_sel[old_index],
+					clk->reg_mask[old_index]);
+		if (ret)
+			return ret;
+	}
+
+	if (clk->reg_sel[index]) {
+		ret = ab8500_sysctrl_write(clk->reg_sel[index],
+					clk->reg_mask[index],
+					clk->reg_bits[index]);
+		if (ret) {
+			if (clk->reg_sel[old_index])
+				ab8500_sysctrl_write(clk->reg_sel[old_index],
+						clk->reg_mask[old_index],
+						clk->reg_bits[old_index]);
+			return ret;
+		}
+	}
+	clk->parent_index = index;
+
+	return ret;
+}
+
+static u8 clk_sysctrl_get_parent(struct clk_hw *hw)
+{
+	struct clk_sysctrl *clk = to_clk_sysctrl(hw);
+	return clk->parent_index;
+}
+
+static const struct clk_ops clk_sysctrl_gate_ops = {
+	.prepare = clk_sysctrl_prepare,
+	.unprepare = clk_sysctrl_unprepare,
+};
+
+static const struct clk_ops clk_sysctrl_gate_fixed_rate_ops = {
+	.prepare = clk_sysctrl_prepare,
+	.unprepare = clk_sysctrl_unprepare,
+	.recalc_rate = clk_sysctrl_recalc_rate,
+};
+
+static const struct clk_ops clk_sysctrl_set_parent_ops = {
+	.determine_rate = clk_hw_determine_rate_no_reparent,
+	.set_parent = clk_sysctrl_set_parent,
+	.get_parent = clk_sysctrl_get_parent,
+};
+
+static struct clk *clk_reg_sysctrl(struct device *dev,
+				const char *name,
+				const char **parent_names,
+				u8 num_parents,
+				u16 *reg_sel,
+				u8 *reg_mask,
+				u8 *reg_bits,
+				unsigned long rate,
+				unsigned long enable_delay_us,
+				unsigned long flags,
+				const struct clk_ops *clk_sysctrl_ops)
+{
+	struct clk_sysctrl *clk;
+	struct clk_init_data clk_sysctrl_init;
+	struct clk *clk_reg;
+	int i;
+
+	if (!dev)
+		return ERR_PTR(-EINVAL);
+
+	if (!name || (num_parents > SYSCTRL_MAX_NUM_PARENTS)) {
+		dev_err(dev, "clk_sysctrl: invalid arguments passed\n");
+		return ERR_PTR(-EINVAL);
+	}
+
+	clk = devm_kzalloc(dev, sizeof(*clk), GFP_KERNEL);
+	if (!clk)
+		return ERR_PTR(-ENOMEM);
+
+	/* set main clock registers */
+	clk->reg_sel[0] = reg_sel[0];
+	clk->reg_bits[0] = reg_bits[0];
+	clk->reg_mask[0] = reg_mask[0];
+
+	/* handle clocks with more than one parent */
+	for (i = 1; i < num_parents; i++) {
+		clk->reg_sel[i] = reg_sel[i];
+		clk->reg_bits[i] = reg_bits[i];
+		clk->reg_mask[i] = reg_mask[i];
+	}
+
+	clk->parent_index = 0;
+	clk->rate = rate;
+	clk->enable_delay_us = enable_delay_us;
+	clk->dev = dev;
+
+	clk_sysctrl_init.name = name;
+	clk_sysctrl_init.ops = clk_sysctrl_ops;
+	clk_sysctrl_init.flags = flags;
+	clk_sysctrl_init.parent_names = parent_names;
+	clk_sysctrl_init.num_parents = num_parents;
+	clk->hw.init = &clk_sysctrl_init;
+
+	clk_reg = devm_clk_register(clk->dev, &clk->hw);
+	if (IS_ERR(clk_reg))
+		dev_err(dev, "clk_sysctrl: clk_register failed\n");
+
+	return clk_reg;
+}
+
+struct clk *clk_reg_sysctrl_gate(struct device *dev,
+				const char *name,
+				const char *parent_name,
+				u16 reg_sel,
+				u8 reg_mask,
+				u8 reg_bits,
+				unsigned long enable_delay_us,
+				unsigned long flags)
+{
+	const char **parent_names = (parent_name ? &parent_name : NULL);
+	u8 num_parents = (parent_name ? 1 : 0);
+
+	return clk_reg_sysctrl(dev, name, parent_names, num_parents,
+			&reg_sel, &reg_mask, &reg_bits, 0, enable_delay_us,
+			flags, &clk_sysctrl_gate_ops);
+}
+
+struct clk *clk_reg_sysctrl_gate_fixed_rate(struct device *dev,
+					const char *name,
+					const char *parent_name,
+					u16 reg_sel,
+					u8 reg_mask,
+					u8 reg_bits,
+					unsigned long rate,
+					unsigned long enable_delay_us,
+					unsigned long flags)
+{
+	const char **parent_names = (parent_name ? &parent_name : NULL);
+	u8 num_parents = (parent_name ? 1 : 0);
+
+	return clk_reg_sysctrl(dev, name, parent_names, num_parents,
+			&reg_sel, &reg_mask, &reg_bits,
+			rate, enable_delay_us, flags,
+			&clk_sysctrl_gate_fixed_rate_ops);
+}
+
+struct clk *clk_reg_sysctrl_set_parent(struct device *dev,
+				const char *name,
+				const char **parent_names,
+				u8 num_parents,
+				u16 *reg_sel,
+				u8 *reg_mask,
+				u8 *reg_bits,
+				unsigned long flags)
+{
+	return clk_reg_sysctrl(dev, name, parent_names, num_parents,
+			reg_sel, reg_mask, reg_bits, 0, 0, flags,
+			&clk_sysctrl_set_parent_ops);
+}