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, 252 insertions, 0 deletions

diff --git a/drivers/greybus/bundle.c b/drivers/greybus/bundle.c
new file mode 100644
index 000000000..d1831d098
--- /dev/null
+++ b/drivers/greybus/bundle.c
@@ -0,0 +1,252 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Greybus bundles
+ *
+ * Copyright 2014-2015 Google Inc.
+ * Copyright 2014-2015 Linaro Ltd.
+ */
+
+#include <linux/greybus.h>
+#include "greybus_trace.h"
+
+static ssize_t bundle_class_show(struct device *dev,
+				 struct device_attribute *attr, char *buf)
+{
+	struct gb_bundle *bundle = to_gb_bundle(dev);
+
+	return sprintf(buf, "0x%02x\n", bundle->class);
+}
+static DEVICE_ATTR_RO(bundle_class);
+
+static ssize_t bundle_id_show(struct device *dev,
+			      struct device_attribute *attr, char *buf)
+{
+	struct gb_bundle *bundle = to_gb_bundle(dev);
+
+	return sprintf(buf, "%u\n", bundle->id);
+}
+static DEVICE_ATTR_RO(bundle_id);
+
+static ssize_t state_show(struct device *dev, struct device_attribute *attr,
+			  char *buf)
+{
+	struct gb_bundle *bundle = to_gb_bundle(dev);
+
+	if (!bundle->state)
+		return sprintf(buf, "\n");
+
+	return sprintf(buf, "%s\n", bundle->state);
+}
+
+static ssize_t state_store(struct device *dev, struct device_attribute *attr,
+			   const char *buf, size_t size)
+{
+	struct gb_bundle *bundle = to_gb_bundle(dev);
+
+	kfree(bundle->state);
+	bundle->state = kstrdup(buf, GFP_KERNEL);
+	if (!bundle->state)
+		return -ENOMEM;
+
+	/* Tell userspace that the file contents changed */
+	sysfs_notify(&bundle->dev.kobj, NULL, "state");
+
+	return size;
+}
+static DEVICE_ATTR_RW(state);
+
+static struct attribute *bundle_attrs[] = {
+	&dev_attr_bundle_class.attr,
+	&dev_attr_bundle_id.attr,
+	&dev_attr_state.attr,
+	NULL,
+};
+
+ATTRIBUTE_GROUPS(bundle);
+
+static struct gb_bundle *gb_bundle_find(struct gb_interface *intf,
+					u8 bundle_id)
+{
+	struct gb_bundle *bundle;
+
+	list_for_each_entry(bundle, &intf->bundles, links) {
+		if (bundle->id == bundle_id)
+			return bundle;
+	}
+
+	return NULL;
+}
+
+static void gb_bundle_release(struct device *dev)
+{
+	struct gb_bundle *bundle = to_gb_bundle(dev);
+
+	trace_gb_bundle_release(bundle);
+
+	kfree(bundle->state);
+	kfree(bundle->cport_desc);
+	kfree(bundle);
+}
+
+#ifdef CONFIG_PM
+static void gb_bundle_disable_all_connections(struct gb_bundle *bundle)
+{
+	struct gb_connection *connection;
+
+	list_for_each_entry(connection, &bundle->connections, bundle_links)
+		gb_connection_disable(connection);
+}
+
+static void gb_bundle_enable_all_connections(struct gb_bundle *bundle)
+{
+	struct gb_connection *connection;
+
+	list_for_each_entry(connection, &bundle->connections, bundle_links)
+		gb_connection_enable(connection);
+}
+
+static int gb_bundle_suspend(struct device *dev)
+{
+	struct gb_bundle *bundle = to_gb_bundle(dev);
+	const struct dev_pm_ops *pm = dev->driver->pm;
+	int ret;
+
+	if (pm && pm->runtime_suspend) {
+		ret = pm->runtime_suspend(&bundle->dev);
+		if (ret)
+			return ret;
+	} else {
+		gb_bundle_disable_all_connections(bundle);
+	}
+
+	ret = gb_control_bundle_suspend(bundle->intf->control, bundle->id);
+	if (ret) {
+		if (pm && pm->runtime_resume)
+			ret = pm->runtime_resume(dev);
+		else
+			gb_bundle_enable_all_connections(bundle);
+
+		return ret;
+	}
+
+	return 0;
+}
+
+static int gb_bundle_resume(struct device *dev)
+{
+	struct gb_bundle *bundle = to_gb_bundle(dev);
+	const struct dev_pm_ops *pm = dev->driver->pm;
+	int ret;
+
+	ret = gb_control_bundle_resume(bundle->intf->control, bundle->id);
+	if (ret)
+		return ret;
+
+	if (pm && pm->runtime_resume) {
+		ret = pm->runtime_resume(dev);
+		if (ret)
+			return ret;
+	} else {
+		gb_bundle_enable_all_connections(bundle);
+	}
+
+	return 0;
+}
+
+static int gb_bundle_idle(struct device *dev)
+{
+	pm_runtime_mark_last_busy(dev);
+	pm_request_autosuspend(dev);
+
+	return 0;
+}
+#endif
+
+static const struct dev_pm_ops gb_bundle_pm_ops = {
+	SET_RUNTIME_PM_OPS(gb_bundle_suspend, gb_bundle_resume, gb_bundle_idle)
+};
+
+const struct device_type greybus_bundle_type = {
+	.name =		"greybus_bundle",
+	.release =	gb_bundle_release,
+	.pm =		&gb_bundle_pm_ops,
+};
+
+/*
+ * Create a gb_bundle structure to represent a discovered
+ * bundle.  Returns a pointer to the new bundle or a null
+ * pointer if a failure occurs due to memory exhaustion.
+ */
+struct gb_bundle *gb_bundle_create(struct gb_interface *intf, u8 bundle_id,
+				   u8 class)
+{
+	struct gb_bundle *bundle;
+
+	if (bundle_id == BUNDLE_ID_NONE) {
+		dev_err(&intf->dev, "can't use bundle id %u\n", bundle_id);
+		return NULL;
+	}
+
+	/*
+	 * Reject any attempt to reuse a bundle id.  We initialize
+	 * these serially, so there's no need to worry about keeping
+	 * the interface bundle list locked here.
+	 */
+	if (gb_bundle_find(intf, bundle_id)) {
+		dev_err(&intf->dev, "duplicate bundle id %u\n", bundle_id);
+		return NULL;
+	}
+
+	bundle = kzalloc_obj(*bundle);
+	if (!bundle)
+		return NULL;
+
+	bundle->intf = intf;
+	bundle->id = bundle_id;
+	bundle->class = class;
+	INIT_LIST_HEAD(&bundle->connections);
+
+	bundle->dev.parent = &intf->dev;
+	bundle->dev.bus = &greybus_bus_type;
+	bundle->dev.type = &greybus_bundle_type;
+	bundle->dev.groups = bundle_groups;
+	bundle->dev.dma_mask = intf->dev.dma_mask;
+	device_initialize(&bundle->dev);
+	dev_set_name(&bundle->dev, "%s.%d", dev_name(&intf->dev), bundle_id);
+
+	list_add(&bundle->links, &intf->bundles);
+
+	trace_gb_bundle_create(bundle);
+
+	return bundle;
+}
+
+int gb_bundle_add(struct gb_bundle *bundle)
+{
+	int ret;
+
+	ret = device_add(&bundle->dev);
+	if (ret) {
+		dev_err(&bundle->dev, "failed to register bundle: %d\n", ret);
+		return ret;
+	}
+
+	trace_gb_bundle_add(bundle);
+
+	return 0;
+}
+
+/*
+ * Tear down a previously set up bundle.
+ */
+void gb_bundle_destroy(struct gb_bundle *bundle)
+{
+	trace_gb_bundle_destroy(bundle);
+
+	if (device_is_registered(&bundle->dev))
+		device_del(&bundle->dev);
+
+	list_del(&bundle->links);
+
+	put_device(&bundle->dev);
+}