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, 446 insertions, 0 deletions

diff --git a/drivers/thermal/testing/zone.c b/drivers/thermal/testing/zone.c
new file mode 100644
index 000000000..6db26276d
--- /dev/null
+++ b/drivers/thermal/testing/zone.c
@@ -0,0 +1,446 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2024, Intel Corporation
+ *
+ * Author: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
+ *
+ * Thermal zone tempalates handling for thermal core testing.
+ */
+
+#define pr_fmt(fmt) "thermal-testing: " fmt
+
+#include <linux/debugfs.h>
+#include <linux/idr.h>
+#include <linux/list.h>
+#include <linux/thermal.h>
+
+#include "thermal_testing.h"
+
+#define TT_MAX_FILE_NAME_LENGTH		16
+
+/**
+ * struct tt_thermal_zone - Testing thermal zone template
+ *
+ * Represents a template of a thermal zone that can be used for registering
+ * a test thermal zone with the thermal core.
+ *
+ * @list_node: Node in the list of all testing thermal zone templates.
+ * @trips: List of trip point templates for this thermal zone template.
+ * @d_tt_zone: Directory in debugfs representing this template.
+ * @tz: Test thermal zone based on this template, if present.
+ * @lock: Mutex for synchronizing changes of this template.
+ * @ida: IDA for trip point IDs.
+ * @id: The ID of this template for the debugfs interface.
+ * @temp: Temperature value.
+ * @tz_temp: Current thermal zone temperature (after registration).
+ * @num_trips: Number of trip points in the @trips list.
+ * @refcount: Reference counter for usage and removal synchronization.
+ */
+struct tt_thermal_zone {
+	struct list_head list_node;
+	struct list_head trips;
+	struct dentry *d_tt_zone;
+	struct thermal_zone_device *tz;
+	struct mutex lock;
+	struct ida ida;
+	int id;
+	int temp;
+	int tz_temp;
+	unsigned int num_trips;
+	unsigned int refcount;
+};
+
+DEFINE_GUARD(tt_zone, struct tt_thermal_zone *, mutex_lock(&_T->lock), mutex_unlock(&_T->lock))
+
+/**
+ * struct tt_trip - Testing trip point template
+ *
+ * Represents a template of a trip point to be used for populating a trip point
+ * during the registration of a thermal zone based on a given zone template.
+ *
+ * @list_node: Node in the list of all trip templates in the zone template.
+ * @trip: Trip point data to use for thernal zone registration.
+ * @id: The ID of this trip template for the debugfs interface.
+ */
+struct tt_trip {
+	struct list_head list_node;
+	struct thermal_trip trip;
+	int id;
+};
+
+/*
+ * It is both questionable and potentially problematic from the sychnronization
+ * perspective to attempt to manipulate debugfs from within a debugfs file
+ * "write" operation, so auxiliary work items are used for that.  The majority
+ * of zone-related command functions have a part that runs from a workqueue and
+ * make changes in debugs, among other things.
+ */
+struct tt_work {
+	struct work_struct work;
+	struct tt_thermal_zone *tt_zone;
+	struct tt_trip *tt_trip;
+};
+
+static inline struct tt_work *tt_work_of_work(struct work_struct *work)
+{
+	return container_of(work, struct tt_work, work);
+}
+
+static LIST_HEAD(tt_thermal_zones);
+static DEFINE_IDA(tt_thermal_zones_ida);
+static DEFINE_MUTEX(tt_thermal_zones_lock);
+
+static int tt_int_get(void *data, u64 *val)
+{
+	*val = *(int *)data;
+	return 0;
+}
+static int tt_int_set(void *data, u64 val)
+{
+	if ((int)val < THERMAL_TEMP_INVALID)
+		return -EINVAL;
+
+	*(int *)data = val;
+	return 0;
+}
+DEFINE_DEBUGFS_ATTRIBUTE_SIGNED(tt_int_attr, tt_int_get, tt_int_set, "%lld\n");
+DEFINE_DEBUGFS_ATTRIBUTE(tt_unsigned_int_attr, tt_int_get, tt_int_set, "%llu\n");
+
+static int tt_zone_tz_temp_get(void *data, u64 *val)
+{
+	struct tt_thermal_zone *tt_zone = data;
+
+	guard(tt_zone)(tt_zone);
+
+	if (!tt_zone->tz)
+		return -EBUSY;
+
+	*val = tt_zone->tz_temp;
+
+	return 0;
+}
+static int tt_zone_tz_temp_set(void *data, u64 val)
+{
+	struct tt_thermal_zone *tt_zone = data;
+
+	guard(tt_zone)(tt_zone);
+
+	if (!tt_zone->tz)
+		return -EBUSY;
+
+	WRITE_ONCE(tt_zone->tz_temp, val);
+	thermal_zone_device_update(tt_zone->tz, THERMAL_EVENT_TEMP_SAMPLE);
+
+	return 0;
+}
+DEFINE_DEBUGFS_ATTRIBUTE_SIGNED(tt_zone_tz_temp_attr, tt_zone_tz_temp_get,
+				tt_zone_tz_temp_set, "%lld\n");
+
+static void tt_zone_free_trips(struct tt_thermal_zone *tt_zone)
+{
+	struct tt_trip *tt_trip, *aux;
+
+	list_for_each_entry_safe(tt_trip, aux, &tt_zone->trips, list_node) {
+		list_del(&tt_trip->list_node);
+		ida_free(&tt_zone->ida, tt_trip->id);
+		kfree(tt_trip);
+	}
+}
+
+static void tt_zone_free(struct tt_thermal_zone *tt_zone)
+{
+	tt_zone_free_trips(tt_zone);
+	ida_free(&tt_thermal_zones_ida, tt_zone->id);
+	ida_destroy(&tt_zone->ida);
+	kfree(tt_zone);
+}
+
+static void tt_add_tz_work_fn(struct work_struct *work)
+{
+	struct tt_work *tt_work = tt_work_of_work(work);
+	struct tt_thermal_zone *tt_zone = tt_work->tt_zone;
+	char f_name[TT_MAX_FILE_NAME_LENGTH];
+
+	kfree(tt_work);
+
+	snprintf(f_name, TT_MAX_FILE_NAME_LENGTH, "tz%d", tt_zone->id);
+	tt_zone->d_tt_zone = debugfs_create_dir(f_name, d_testing);
+	if (IS_ERR(tt_zone->d_tt_zone)) {
+		tt_zone_free(tt_zone);
+		return;
+	}
+
+	debugfs_create_file_unsafe("temp", 0600, tt_zone->d_tt_zone, tt_zone,
+				   &tt_zone_tz_temp_attr);
+
+	debugfs_create_file_unsafe("init_temp", 0600, tt_zone->d_tt_zone,
+				   &tt_zone->temp, &tt_int_attr);
+
+	guard(mutex)(&tt_thermal_zones_lock);
+
+	list_add_tail(&tt_zone->list_node, &tt_thermal_zones);
+}
+
+int tt_add_tz(void)
+{
+	int ret;
+
+	struct tt_thermal_zone *tt_zone __free(kfree) = kzalloc_obj(*tt_zone);
+	if (!tt_zone)
+		return -ENOMEM;
+
+	struct tt_work *tt_work __free(kfree) = kzalloc_obj(*tt_work);
+	if (!tt_work)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&tt_zone->trips);
+	mutex_init(&tt_zone->lock);
+	ida_init(&tt_zone->ida);
+	tt_zone->temp = THERMAL_TEMP_INVALID;
+
+	ret = ida_alloc(&tt_thermal_zones_ida, GFP_KERNEL);
+	if (ret < 0)
+		return ret;
+
+	tt_zone->id = ret;
+
+	INIT_WORK(&tt_work->work, tt_add_tz_work_fn);
+	tt_work->tt_zone = no_free_ptr(tt_zone);
+	tt_queue_work(&(no_free_ptr(tt_work)->work));
+
+	return 0;
+}
+
+static void tt_del_tz_work_fn(struct work_struct *work)
+{
+	struct tt_work *tt_work = tt_work_of_work(work);
+	struct tt_thermal_zone *tt_zone = tt_work->tt_zone;
+
+	kfree(tt_work);
+
+	debugfs_remove(tt_zone->d_tt_zone);
+	tt_zone_free(tt_zone);
+}
+
+static void tt_zone_unregister_tz(struct tt_thermal_zone *tt_zone)
+{
+	guard(tt_zone)(tt_zone);
+
+	if (tt_zone->tz) {
+		thermal_zone_device_unregister(tt_zone->tz);
+		tt_zone->tz = NULL;
+	}
+}
+
+int tt_del_tz(const char *arg)
+{
+	struct tt_thermal_zone *tt_zone, *aux;
+	int ret;
+	int id;
+
+	ret = kstrtoint(arg, 10, &id);
+	if (ret < 0)
+		return ret;
+
+	struct tt_work *tt_work __free(kfree) = kzalloc_obj(*tt_work);
+	if (!tt_work)
+		return -ENOMEM;
+
+	guard(mutex)(&tt_thermal_zones_lock);
+
+	ret = -EINVAL;
+	list_for_each_entry_safe(tt_zone, aux, &tt_thermal_zones, list_node) {
+		if (tt_zone->id == id) {
+			if (tt_zone->refcount) {
+				ret = -EBUSY;
+			} else {
+				list_del(&tt_zone->list_node);
+				ret = 0;
+			}
+			break;
+		}
+	}
+
+	if (ret)
+		return ret;
+
+	tt_zone_unregister_tz(tt_zone);
+
+	INIT_WORK(&tt_work->work, tt_del_tz_work_fn);
+	tt_work->tt_zone = tt_zone;
+	tt_queue_work(&(no_free_ptr(tt_work)->work));
+
+	return 0;
+}
+
+static struct tt_thermal_zone *tt_get_tt_zone(const char *arg)
+{
+	struct tt_thermal_zone *tt_zone;
+	int ret, id;
+
+	ret = kstrtoint(arg, 10, &id);
+	if (ret < 0)
+		return ERR_PTR(ret);
+
+	guard(mutex)(&tt_thermal_zones_lock);
+
+	list_for_each_entry(tt_zone, &tt_thermal_zones, list_node) {
+		if (tt_zone->id == id) {
+			tt_zone->refcount++;
+			return tt_zone;
+		}
+	}
+
+	return ERR_PTR(-EINVAL);
+}
+
+static void tt_put_tt_zone(struct tt_thermal_zone *tt_zone)
+{
+	guard(mutex)(&tt_thermal_zones_lock);
+
+	tt_zone->refcount--;
+}
+
+DEFINE_FREE(put_tt_zone, struct tt_thermal_zone *,
+	    if (!IS_ERR_OR_NULL(_T)) tt_put_tt_zone(_T))
+
+static void tt_zone_add_trip_work_fn(struct work_struct *work)
+{
+	struct tt_work *tt_work = tt_work_of_work(work);
+	struct tt_thermal_zone *tt_zone = tt_work->tt_zone;
+	struct tt_trip *tt_trip = tt_work->tt_trip;
+	char d_name[TT_MAX_FILE_NAME_LENGTH];
+
+	kfree(tt_work);
+
+	snprintf(d_name, TT_MAX_FILE_NAME_LENGTH, "trip_%d_temp", tt_trip->id);
+	debugfs_create_file_unsafe(d_name, 0600, tt_zone->d_tt_zone,
+				   &tt_trip->trip.temperature, &tt_int_attr);
+
+	snprintf(d_name, TT_MAX_FILE_NAME_LENGTH, "trip_%d_hyst", tt_trip->id);
+	debugfs_create_file_unsafe(d_name, 0600, tt_zone->d_tt_zone,
+				   &tt_trip->trip.hysteresis, &tt_unsigned_int_attr);
+
+	tt_put_tt_zone(tt_zone);
+}
+
+int tt_zone_add_trip(const char *arg)
+{
+	int id;
+
+	struct tt_work *tt_work __free(kfree) = kzalloc_obj(*tt_work);
+	if (!tt_work)
+		return -ENOMEM;
+
+	struct tt_trip *tt_trip __free(kfree) = kzalloc_obj(*tt_trip);
+	if (!tt_trip)
+		return -ENOMEM;
+
+	struct tt_thermal_zone *tt_zone __free(put_tt_zone) = tt_get_tt_zone(arg);
+	if (IS_ERR(tt_zone))
+		return PTR_ERR(tt_zone);
+
+	id = ida_alloc(&tt_zone->ida, GFP_KERNEL);
+	if (id < 0)
+		return id;
+
+	tt_trip->trip.type = THERMAL_TRIP_ACTIVE;
+	tt_trip->trip.temperature = THERMAL_TEMP_INVALID;
+	tt_trip->trip.flags = THERMAL_TRIP_FLAG_RW;
+	tt_trip->id = id;
+
+	guard(tt_zone)(tt_zone);
+
+	list_add_tail(&tt_trip->list_node, &tt_zone->trips);
+	tt_zone->num_trips++;
+
+	INIT_WORK(&tt_work->work, tt_zone_add_trip_work_fn);
+	tt_work->tt_zone = no_free_ptr(tt_zone);
+	tt_work->tt_trip = no_free_ptr(tt_trip);
+	tt_queue_work(&(no_free_ptr(tt_work)->work));
+
+	return 0;
+}
+
+static int tt_zone_get_temp(struct thermal_zone_device *tz, int *temp)
+{
+	struct tt_thermal_zone *tt_zone = thermal_zone_device_priv(tz);
+
+	*temp = READ_ONCE(tt_zone->tz_temp);
+
+	if (*temp < THERMAL_TEMP_INVALID)
+		return -ENODATA;
+
+	return 0;
+}
+
+static const struct thermal_zone_device_ops tt_zone_ops = {
+	.get_temp = tt_zone_get_temp,
+};
+
+static int tt_zone_register_tz(struct tt_thermal_zone *tt_zone)
+{
+	struct thermal_zone_device *tz;
+	struct tt_trip *tt_trip;
+	int i;
+
+	guard(tt_zone)(tt_zone);
+
+	if (tt_zone->tz)
+		return -EINVAL;
+
+	struct thermal_trip *trips __free(kfree) = kzalloc_objs(*trips,
+								tt_zone->num_trips);
+	if (!trips)
+		return -ENOMEM;
+
+	i = 0;
+	list_for_each_entry(tt_trip, &tt_zone->trips, list_node)
+		trips[i++] = tt_trip->trip;
+
+	tt_zone->tz_temp = tt_zone->temp;
+
+	tz = thermal_zone_device_register_with_trips("test_tz", trips, i, tt_zone,
+						     &tt_zone_ops, NULL, 0, 0);
+	if (IS_ERR(tz))
+		return PTR_ERR(tz);
+
+	tt_zone->tz = tz;
+
+	thermal_zone_device_enable(tz);
+
+	return 0;
+}
+
+int tt_zone_reg(const char *arg)
+{
+	struct tt_thermal_zone *tt_zone __free(put_tt_zone) = tt_get_tt_zone(arg);
+	if (IS_ERR(tt_zone))
+		return PTR_ERR(tt_zone);
+
+	return tt_zone_register_tz(tt_zone);
+}
+
+int tt_zone_unreg(const char *arg)
+{
+	struct tt_thermal_zone *tt_zone __free(put_tt_zone) = tt_get_tt_zone(arg);
+	if (IS_ERR(tt_zone))
+		return PTR_ERR(tt_zone);
+
+	tt_zone_unregister_tz(tt_zone);
+
+	return 0;
+}
+
+void tt_zone_cleanup(void)
+{
+	struct tt_thermal_zone *tt_zone, *aux;
+
+	list_for_each_entry_safe(tt_zone, aux, &tt_thermal_zones, list_node) {
+		tt_zone_unregister_tz(tt_zone);
+
+		list_del(&tt_zone->list_node);
+
+		tt_zone_free(tt_zone);
+	}
+}