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, 276 insertions, 0 deletions

diff --git a/drivers/dibs/dibs_main.c b/drivers/dibs/dibs_main.c
new file mode 100644
index 000000000..20c50997a
--- /dev/null
+++ b/drivers/dibs/dibs_main.c
@@ -0,0 +1,276 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ *  DIBS - Direct Internal Buffer Sharing
+ *
+ *  Implementation of the DIBS class module
+ *
+ *  Copyright IBM Corp. 2025
+ */
+#define pr_fmt(fmt) "dibs: " fmt
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/err.h>
+#include <linux/dibs.h>
+
+#include "dibs_loopback.h"
+
+MODULE_DESCRIPTION("Direct Internal Buffer Sharing class");
+MODULE_LICENSE("GPL");
+
+static const struct class dibs_class = {
+	.name		= "dibs",
+};
+
+/* use an array rather a list for fast mapping: */
+static struct dibs_client *clients[MAX_DIBS_CLIENTS];
+static u8 max_client;
+static DEFINE_MUTEX(clients_lock);
+struct dibs_dev_list {
+	struct list_head list;
+	struct mutex mutex; /* protects dibs device list */
+};
+
+static struct dibs_dev_list dibs_dev_list = {
+	.list = LIST_HEAD_INIT(dibs_dev_list.list),
+	.mutex = __MUTEX_INITIALIZER(dibs_dev_list.mutex),
+};
+
+static void dibs_setup_forwarding(struct dibs_client *client,
+				  struct dibs_dev *dibs)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&dibs->lock, flags);
+	dibs->subs[client->id] = client;
+	spin_unlock_irqrestore(&dibs->lock, flags);
+}
+
+int dibs_register_client(struct dibs_client *client)
+{
+	struct dibs_dev *dibs;
+	int i, rc = -ENOSPC;
+
+	mutex_lock(&dibs_dev_list.mutex);
+	mutex_lock(&clients_lock);
+	for (i = 0; i < MAX_DIBS_CLIENTS; ++i) {
+		if (!clients[i]) {
+			clients[i] = client;
+			client->id = i;
+			if (i == max_client)
+				max_client++;
+			rc = 0;
+			break;
+		}
+	}
+	mutex_unlock(&clients_lock);
+
+	if (i < MAX_DIBS_CLIENTS) {
+		/* initialize with all devices that we got so far */
+		list_for_each_entry(dibs, &dibs_dev_list.list, list) {
+			dibs->priv[i] = NULL;
+			client->ops->add_dev(dibs);
+			dibs_setup_forwarding(client, dibs);
+		}
+	}
+	mutex_unlock(&dibs_dev_list.mutex);
+
+	return rc;
+}
+EXPORT_SYMBOL_GPL(dibs_register_client);
+
+int dibs_unregister_client(struct dibs_client *client)
+{
+	struct dibs_dev *dibs;
+	unsigned long flags;
+	int max_dmbs;
+	int rc = 0;
+
+	mutex_lock(&dibs_dev_list.mutex);
+	list_for_each_entry(dibs, &dibs_dev_list.list, list) {
+		spin_lock_irqsave(&dibs->lock, flags);
+		max_dmbs = dibs->ops->max_dmbs();
+		for (int i = 0; i < max_dmbs; ++i) {
+			if (dibs->dmb_clientid_arr[i] == client->id) {
+				WARN(1, "%s: attempt to unregister '%s' with registered dmb(s)\n",
+				     __func__, client->name);
+				rc = -EBUSY;
+				goto err_reg_dmb;
+			}
+		}
+		/* Stop forwarding IRQs and events */
+		dibs->subs[client->id] = NULL;
+		spin_unlock_irqrestore(&dibs->lock, flags);
+		clients[client->id]->ops->del_dev(dibs);
+		dibs->priv[client->id] = NULL;
+	}
+
+	mutex_lock(&clients_lock);
+	clients[client->id] = NULL;
+	if (client->id + 1 == max_client)
+		max_client--;
+	mutex_unlock(&clients_lock);
+
+	mutex_unlock(&dibs_dev_list.mutex);
+	return rc;
+
+err_reg_dmb:
+	spin_unlock_irqrestore(&dibs->lock, flags);
+	mutex_unlock(&dibs_dev_list.mutex);
+	return rc;
+}
+EXPORT_SYMBOL_GPL(dibs_unregister_client);
+
+static void dibs_dev_release(struct device *dev)
+{
+	struct dibs_dev *dibs;
+
+	dibs = container_of(dev, struct dibs_dev, dev);
+
+	kfree(dibs->dmb_clientid_arr);
+	kfree(dibs);
+}
+
+struct dibs_dev *dibs_dev_alloc(void)
+{
+	struct dibs_dev *dibs;
+
+	dibs = kzalloc_obj(*dibs);
+	if (!dibs)
+		return dibs;
+	spin_lock_init(&dibs->lock);
+	dibs->dev.release = dibs_dev_release;
+	dibs->dev.class = &dibs_class;
+	device_initialize(&dibs->dev);
+
+	return dibs;
+}
+EXPORT_SYMBOL_GPL(dibs_dev_alloc);
+
+static ssize_t gid_show(struct device *dev, struct device_attribute *attr,
+			char *buf)
+{
+	struct dibs_dev *dibs;
+
+	dibs = container_of(dev, struct dibs_dev, dev);
+
+	return sysfs_emit(buf, "%pUb\n", &dibs->gid);
+}
+static DEVICE_ATTR_RO(gid);
+
+static ssize_t fabric_id_show(struct device *dev, struct device_attribute *attr,
+			      char *buf)
+{
+	struct dibs_dev *dibs;
+	u16 fabric_id;
+
+	dibs = container_of(dev, struct dibs_dev, dev);
+	fabric_id = dibs->ops->get_fabric_id(dibs);
+
+	return sysfs_emit(buf, "0x%04x\n", fabric_id);
+}
+static DEVICE_ATTR_RO(fabric_id);
+
+static struct attribute *dibs_dev_attrs[] = {
+	&dev_attr_gid.attr,
+	&dev_attr_fabric_id.attr,
+	NULL,
+};
+
+static const struct attribute_group dibs_dev_attr_group = {
+	.attrs = dibs_dev_attrs,
+};
+
+int dibs_dev_add(struct dibs_dev *dibs)
+{
+	int max_dmbs;
+	int i, ret;
+
+	max_dmbs = dibs->ops->max_dmbs();
+	dibs->dmb_clientid_arr = kzalloc(max_dmbs, GFP_KERNEL);
+	if (!dibs->dmb_clientid_arr)
+		return -ENOMEM;
+	memset(dibs->dmb_clientid_arr, NO_DIBS_CLIENT, max_dmbs);
+
+	ret = device_add(&dibs->dev);
+	if (ret)
+		return ret;
+
+	ret = sysfs_create_group(&dibs->dev.kobj, &dibs_dev_attr_group);
+	if (ret) {
+		dev_err(&dibs->dev, "sysfs_create_group failed for dibs_dev\n");
+		device_del(&dibs->dev);
+		return ret;
+	}
+	mutex_lock(&dibs_dev_list.mutex);
+	mutex_lock(&clients_lock);
+	for (i = 0; i < max_client; ++i) {
+		if (clients[i]) {
+			clients[i]->ops->add_dev(dibs);
+			dibs_setup_forwarding(clients[i], dibs);
+		}
+	}
+	mutex_unlock(&clients_lock);
+	list_add(&dibs->list, &dibs_dev_list.list);
+	mutex_unlock(&dibs_dev_list.mutex);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(dibs_dev_add);
+
+void dibs_dev_del(struct dibs_dev *dibs)
+{
+	unsigned long flags;
+	int i;
+
+	sysfs_remove_group(&dibs->dev.kobj, &dibs_dev_attr_group);
+
+	spin_lock_irqsave(&dibs->lock, flags);
+	for (i = 0; i < MAX_DIBS_CLIENTS; ++i)
+		dibs->subs[i] = NULL;
+	spin_unlock_irqrestore(&dibs->lock, flags);
+
+	mutex_lock(&dibs_dev_list.mutex);
+	mutex_lock(&clients_lock);
+	for (i = 0; i < max_client; ++i) {
+		if (clients[i])
+			clients[i]->ops->del_dev(dibs);
+	}
+	mutex_unlock(&clients_lock);
+	list_del_init(&dibs->list);
+	mutex_unlock(&dibs_dev_list.mutex);
+
+	device_del(&dibs->dev);
+}
+EXPORT_SYMBOL_GPL(dibs_dev_del);
+
+static int __init dibs_init(void)
+{
+	int rc;
+
+	rc = class_register(&dibs_class);
+	if (rc)
+		goto err;
+
+	rc = dibs_loopback_init();
+	if (rc)
+		goto err_unregister;
+
+	return rc;
+
+err_unregister:
+	class_unregister(&dibs_class);
+err:
+	pr_err("%s fails with %d\n", __func__, rc);
+	return rc;
+}
+
+static void __exit dibs_exit(void)
+{
+	dibs_loopback_exit();
+	class_unregister(&dibs_class);
+}
+
+subsys_initcall(dibs_init);
+module_exit(dibs_exit);