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

4 files changed, 317 insertions, 0 deletions

diff --git a/drivers/soc/versatile/Kconfig b/drivers/soc/versatile/Kconfig
new file mode 100644
index 000000000..7bbf54a8d
--- /dev/null
+++ b/drivers/soc/versatile/Kconfig
@@ -0,0 +1,20 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# ARM Versatile SoC drivers
+#
+config SOC_INTEGRATOR_CM
+	bool "SoC bus device for the ARM Integrator platform core modules"
+	depends on ARCH_INTEGRATOR || COMPILE_TEST
+	select SOC_BUS
+	help
+	  Include support for the SoC bus on the ARM Integrator platform
+	  core modules providing some sysfs information about the ASIC
+	  variant.
+
+config SOC_REALVIEW
+	bool "SoC bus device for the ARM RealView platforms"
+	depends on ARCH_REALVIEW || COMPILE_TEST
+	select SOC_BUS
+	help
+	  Include support for the SoC bus on the ARM RealView platforms
+	  providing some sysfs information about the ASIC variant.
diff --git a/drivers/soc/versatile/Makefile b/drivers/soc/versatile/Makefile
new file mode 100644
index 000000000..1e0a37c0a
--- /dev/null
+++ b/drivers/soc/versatile/Makefile
@@ -0,0 +1,3 @@
+# SPDX-License-Identifier: GPL-2.0-only
+obj-$(CONFIG_SOC_INTEGRATOR_CM)	+= soc-integrator.o
+obj-$(CONFIG_SOC_REALVIEW)	+= soc-realview.o
diff --git a/drivers/soc/versatile/soc-integrator.c b/drivers/soc/versatile/soc-integrator.c
new file mode 100644
index 000000000..b67a7fd4a
--- /dev/null
+++ b/drivers/soc/versatile/soc-integrator.c
@@ -0,0 +1,150 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2014 Linaro Ltd.
+ *
+ * Author: Linus Walleij <linus.walleij@linaro.org>
+ */
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/slab.h>
+#include <linux/sys_soc.h>
+#include <linux/platform_device.h>
+#include <linux/mfd/syscon.h>
+#include <linux/regmap.h>
+#include <linux/of.h>
+
+#define INTEGRATOR_HDR_ID_OFFSET	0x00
+
+static u32 integrator_coreid;
+
+static const struct of_device_id integrator_cm_match[] = {
+	{ .compatible = "arm,core-module-integrator", },
+	{ }
+};
+
+static const char *integrator_arch_str(u32 id)
+{
+	switch ((id >> 16) & 0xff) {
+	case 0x00:
+		return "ASB little-endian";
+	case 0x01:
+		return "AHB little-endian";
+	case 0x03:
+		return "AHB-Lite system bus, bi-endian";
+	case 0x04:
+		return "AHB";
+	case 0x08:
+		return "AHB system bus, ASB processor bus";
+	default:
+		return "Unknown";
+	}
+}
+
+static const char *integrator_fpga_str(u32 id)
+{
+	switch ((id >> 12) & 0xf) {
+	case 0x01:
+		return "XC4062";
+	case 0x02:
+		return "XC4085";
+	case 0x03:
+		return "XVC600";
+	case 0x04:
+		return "EPM7256AE (Altera PLD)";
+	default:
+		return "Unknown";
+	}
+}
+
+static ssize_t
+manufacturer_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%02x\n", integrator_coreid >> 24);
+}
+
+static DEVICE_ATTR_RO(manufacturer);
+
+static ssize_t
+arch_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%s\n", integrator_arch_str(integrator_coreid));
+}
+
+static DEVICE_ATTR_RO(arch);
+
+static ssize_t
+fpga_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%s\n", integrator_fpga_str(integrator_coreid));
+}
+
+static DEVICE_ATTR_RO(fpga);
+
+static ssize_t
+build_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%02x\n", (integrator_coreid >> 4) & 0xFF);
+}
+
+static DEVICE_ATTR_RO(build);
+
+static struct attribute *integrator_attrs[] = {
+	&dev_attr_manufacturer.attr,
+	&dev_attr_arch.attr,
+	&dev_attr_fpga.attr,
+	&dev_attr_build.attr,
+	NULL
+};
+
+ATTRIBUTE_GROUPS(integrator);
+
+static int __init integrator_soc_init(void)
+{
+	struct regmap *syscon_regmap;
+	struct soc_device *soc_dev;
+	struct soc_device_attribute *soc_dev_attr;
+	struct device_node *np;
+	struct device *dev;
+	u32 val;
+	int ret;
+
+	np = of_find_matching_node(NULL, integrator_cm_match);
+	if (!np)
+		return -ENODEV;
+
+	syscon_regmap = syscon_node_to_regmap(np);
+	of_node_put(np);
+	if (IS_ERR(syscon_regmap))
+		return PTR_ERR(syscon_regmap);
+
+	ret = regmap_read(syscon_regmap, INTEGRATOR_HDR_ID_OFFSET,
+			  &val);
+	if (ret)
+		return -ENODEV;
+	integrator_coreid = val;
+
+	soc_dev_attr = kzalloc_obj(*soc_dev_attr);
+	if (!soc_dev_attr)
+		return -ENOMEM;
+
+	soc_dev_attr->soc_id = "Integrator";
+	soc_dev_attr->machine = "Integrator";
+	soc_dev_attr->family = "Versatile";
+	soc_dev_attr->custom_attr_group = integrator_groups[0];
+	soc_dev = soc_device_register(soc_dev_attr);
+	if (IS_ERR(soc_dev)) {
+		kfree(soc_dev_attr);
+		return -ENODEV;
+	}
+	dev = soc_device_to_device(soc_dev);
+
+	dev_info(dev, "Detected ARM core module:\n");
+	dev_info(dev, "    Manufacturer: %02x\n", (val >> 24));
+	dev_info(dev, "    Architecture: %s\n", integrator_arch_str(val));
+	dev_info(dev, "    FPGA: %s\n", integrator_fpga_str(val));
+	dev_info(dev, "    Build: %02x\n", (val >> 4) & 0xFF);
+	dev_info(dev, "    Rev: %c\n", ('A' + (val & 0x03)));
+
+	return 0;
+}
+device_initcall(integrator_soc_init);
diff --git a/drivers/soc/versatile/soc-realview.c b/drivers/soc/versatile/soc-realview.c
new file mode 100644
index 000000000..cf91abe07
--- /dev/null
+++ b/drivers/soc/versatile/soc-realview.c
@@ -0,0 +1,144 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2014 Linaro Ltd.
+ *
+ * Author: Linus Walleij <linus.walleij@linaro.org>
+ */
+#include <linux/device.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/slab.h>
+#include <linux/sys_soc.h>
+#include <linux/platform_device.h>
+#include <linux/mfd/syscon.h>
+#include <linux/regmap.h>
+#include <linux/of.h>
+
+/* System ID in syscon */
+#define REALVIEW_SYS_ID_OFFSET 0x00
+
+static const struct of_device_id realview_soc_of_match[] = {
+	{ .compatible = "arm,realview-eb-soc",	},
+	{ .compatible = "arm,realview-pb1176-soc", },
+	{ .compatible = "arm,realview-pb11mp-soc", },
+	{ .compatible = "arm,realview-pba8-soc", },
+	{ .compatible = "arm,realview-pbx-soc", },
+	{ }
+};
+
+static u32 realview_coreid;
+
+static const char *realview_arch_str(u32 id)
+{
+	switch ((id >> 8) & 0xf) {
+	case 0x04:
+		return "AHB";
+	case 0x05:
+		return "Multi-layer AXI";
+	default:
+		return "Unknown";
+	}
+}
+
+static ssize_t
+manufacturer_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%02x\n", realview_coreid >> 24);
+}
+
+static DEVICE_ATTR_RO(manufacturer);
+
+static ssize_t
+board_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	return sprintf(buf, "HBI-%03x\n", ((realview_coreid >> 16) & 0xfff));
+}
+
+static DEVICE_ATTR_RO(board);
+
+static ssize_t
+fpga_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%s\n", realview_arch_str(realview_coreid));
+}
+
+static DEVICE_ATTR_RO(fpga);
+
+static ssize_t
+build_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%02x\n", (realview_coreid & 0xFF));
+}
+
+static DEVICE_ATTR_RO(build);
+
+static struct attribute *realview_attrs[] = {
+	&dev_attr_manufacturer.attr,
+	&dev_attr_board.attr,
+	&dev_attr_fpga.attr,
+	&dev_attr_build.attr,
+	NULL
+};
+
+ATTRIBUTE_GROUPS(realview);
+
+static void realview_soc_socdev_release(void *data)
+{
+	struct soc_device *soc_dev = data;
+
+	soc_device_unregister(soc_dev);
+}
+
+static int realview_soc_probe(struct platform_device *pdev)
+{
+	struct regmap *syscon_regmap;
+	struct soc_device *soc_dev;
+	struct soc_device_attribute *soc_dev_attr;
+	struct device_node *np = pdev->dev.of_node;
+	int ret;
+
+	syscon_regmap = syscon_regmap_lookup_by_phandle(np, "regmap");
+	if (IS_ERR(syscon_regmap))
+		return PTR_ERR(syscon_regmap);
+
+	soc_dev_attr = devm_kzalloc(&pdev->dev, sizeof(*soc_dev_attr), GFP_KERNEL);
+	if (!soc_dev_attr)
+		return -ENOMEM;
+
+	ret = of_property_read_string(np, "compatible",
+				      &soc_dev_attr->soc_id);
+	if (ret)
+		return -EINVAL;
+
+	soc_dev_attr->machine = "RealView";
+	soc_dev_attr->family = "Versatile";
+	soc_dev_attr->custom_attr_group = realview_groups[0];
+	soc_dev = soc_device_register(soc_dev_attr);
+	if (IS_ERR(soc_dev))
+		return -ENODEV;
+
+	ret = devm_add_action_or_reset(&pdev->dev, realview_soc_socdev_release,
+				       soc_dev);
+	if (ret)
+		return ret;
+
+	ret = regmap_read(syscon_regmap, REALVIEW_SYS_ID_OFFSET,
+			  &realview_coreid);
+	if (ret)
+		return -ENODEV;
+
+	dev_info(&pdev->dev, "RealView Syscon Core ID: 0x%08x, HBI-%03x\n",
+		 realview_coreid,
+		 ((realview_coreid >> 16) & 0xfff));
+	/* FIXME: add attributes for SoC to sysfs */
+	return 0;
+}
+
+static struct platform_driver realview_soc_driver = {
+	.probe = realview_soc_probe,
+	.driver = {
+		.name = "realview-soc",
+		.of_match_table = realview_soc_of_match,
+	},
+};
+builtin_platform_driver(realview_soc_driver);