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, 192 insertions, 0 deletions

diff --git a/drivers/reset/sti/reset-syscfg.c b/drivers/reset/sti/reset-syscfg.c
new file mode 100644
index 000000000..38f78d78f
--- /dev/null
+++ b/drivers/reset/sti/reset-syscfg.c
@@ -0,0 +1,192 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2013 STMicroelectronics Limited
+ * Author: Stephen Gallimore <stephen.gallimore@st.com>
+ *
+ * Inspired by mach-imx/src.c
+ */
+#include <linux/kernel.h>
+#include <linux/platform_device.h>
+#include <linux/property.h>
+#include <linux/module.h>
+#include <linux/err.h>
+#include <linux/types.h>
+#include <linux/of.h>
+#include <linux/regmap.h>
+#include <linux/mfd/syscon.h>
+
+#include "reset-syscfg.h"
+
+/**
+ * struct syscfg_reset_channel - Reset channel regmap configuration
+ *
+ * @reset: regmap field for the channel's reset bit.
+ * @ack: regmap field for the channel's ack bit (optional).
+ */
+struct syscfg_reset_channel {
+	struct regmap_field *reset;
+	struct regmap_field *ack;
+};
+
+/**
+ * struct syscfg_reset_controller - A reset controller which groups together
+ * a set of related reset bits, which may be located in different system
+ * configuration registers.
+ *
+ * @rst: base reset controller structure.
+ * @active_low: are the resets in this controller active low, i.e. clearing
+ *              the reset bit puts the hardware into reset.
+ * @channels: An array of reset channels for this controller.
+ */
+struct syscfg_reset_controller {
+	struct reset_controller_dev rst;
+	bool active_low;
+	struct syscfg_reset_channel channels[];
+};
+
+#define to_syscfg_reset_controller(_rst) \
+	container_of(_rst, struct syscfg_reset_controller, rst)
+
+static int syscfg_reset_program_hw(struct reset_controller_dev *rcdev,
+				   unsigned long idx, int assert)
+{
+	struct syscfg_reset_controller *rst = to_syscfg_reset_controller(rcdev);
+	const struct syscfg_reset_channel *ch;
+	u32 ctrl_val = rst->active_low ? !assert : !!assert;
+	int err;
+
+	if (idx >= rcdev->nr_resets)
+		return -EINVAL;
+
+	ch = &rst->channels[idx];
+
+	err = regmap_field_write(ch->reset, ctrl_val);
+	if (err)
+		return err;
+
+	if (ch->ack) {
+		u32 ack_val;
+
+		err = regmap_field_read_poll_timeout(ch->ack, ack_val, (ack_val == ctrl_val),
+						     100, USEC_PER_SEC);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int syscfg_reset_assert(struct reset_controller_dev *rcdev,
+			       unsigned long idx)
+{
+	return syscfg_reset_program_hw(rcdev, idx, true);
+}
+
+static int syscfg_reset_deassert(struct reset_controller_dev *rcdev,
+				 unsigned long idx)
+{
+	return syscfg_reset_program_hw(rcdev, idx, false);
+}
+
+static int syscfg_reset_dev(struct reset_controller_dev *rcdev,
+			    unsigned long idx)
+{
+	int err;
+
+	err = syscfg_reset_assert(rcdev, idx);
+	if (err)
+		return err;
+
+	return syscfg_reset_deassert(rcdev, idx);
+}
+
+static int syscfg_reset_status(struct reset_controller_dev *rcdev,
+			       unsigned long idx)
+{
+	struct syscfg_reset_controller *rst = to_syscfg_reset_controller(rcdev);
+	const struct syscfg_reset_channel *ch;
+	u32 ret_val = 0;
+	int err;
+
+	if (idx >= rcdev->nr_resets)
+		return -EINVAL;
+
+	ch = &rst->channels[idx];
+	if (ch->ack)
+		err = regmap_field_read(ch->ack, &ret_val);
+	else
+		err = regmap_field_read(ch->reset, &ret_val);
+	if (err)
+		return err;
+
+	return rst->active_low ? !ret_val : !!ret_val;
+}
+
+static const struct reset_control_ops syscfg_reset_ops = {
+	.reset    = syscfg_reset_dev,
+	.assert   = syscfg_reset_assert,
+	.deassert = syscfg_reset_deassert,
+	.status   = syscfg_reset_status,
+};
+
+static int syscfg_reset_controller_register(struct device *dev,
+				const struct syscfg_reset_controller_data *data)
+{
+	struct syscfg_reset_controller *rc;
+	int i, err;
+
+	rc = devm_kzalloc(dev, struct_size(rc, channels, data->nr_channels), GFP_KERNEL);
+	if (!rc)
+		return -ENOMEM;
+
+	rc->rst.ops = &syscfg_reset_ops;
+	rc->rst.of_node = dev->of_node;
+	rc->rst.nr_resets = data->nr_channels;
+	rc->active_low = data->active_low;
+
+	for (i = 0; i < data->nr_channels; i++) {
+		struct regmap *map;
+		struct regmap_field *f;
+		const char *compatible = data->channels[i].compatible;
+
+		map = syscon_regmap_lookup_by_compatible(compatible);
+		if (IS_ERR(map))
+			return PTR_ERR(map);
+
+		f = devm_regmap_field_alloc(dev, map, data->channels[i].reset);
+		if (IS_ERR(f))
+			return PTR_ERR(f);
+
+		rc->channels[i].reset = f;
+
+		if (!data->wait_for_ack)
+			continue;
+
+		f = devm_regmap_field_alloc(dev, map, data->channels[i].ack);
+		if (IS_ERR(f))
+			return PTR_ERR(f);
+
+		rc->channels[i].ack = f;
+	}
+
+	err = reset_controller_register(&rc->rst);
+	if (!err)
+		dev_info(dev, "registered\n");
+
+	return err;
+}
+
+int syscfg_reset_probe(struct platform_device *pdev)
+{
+	struct device *dev = pdev ? &pdev->dev : NULL;
+	const void *data;
+
+	if (!dev || !dev->driver)
+		return -ENODEV;
+
+	data = device_get_match_data(&pdev->dev);
+	if (!data)
+		return -EINVAL;
+
+	return syscfg_reset_controller_register(dev, data);
+}