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, 363 insertions, 0 deletions

diff --git a/net/atm/resources.c b/net/atm/resources.c
new file mode 100644
index 000000000..d94dc8221
--- /dev/null
+++ b/net/atm/resources.c
@@ -0,0 +1,363 @@
+// SPDX-License-Identifier: GPL-2.0
+/* net/atm/resources.c - Statically allocated resources */
+
+/* Written 1995-2000 by Werner Almesberger, EPFL LRC/ICA */
+
+/* Fixes
+ * Arnaldo Carvalho de Melo <acme@conectiva.com.br>
+ * 2002/01 - don't free the whole struct sock on sk->destruct time,
+ * 	     use the default destruct function initialized by sock_init_data */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ":%s: " fmt, __func__
+
+#include <linux/ctype.h>
+#include <linux/string.h>
+#include <linux/atmdev.h>
+#include <linux/kernel.h> /* for barrier */
+#include <linux/module.h>
+#include <linux/bitops.h>
+#include <linux/capability.h>
+#include <linux/delay.h>
+#include <linux/mutex.h>
+#include <linux/slab.h>
+
+#include <net/sock.h>	 /* for struct sock */
+
+#include "common.h"
+#include "resources.h"
+
+
+LIST_HEAD(atm_devs);
+DEFINE_MUTEX(atm_dev_mutex);
+
+static struct atm_dev *__alloc_atm_dev(const char *type)
+{
+	struct atm_dev *dev;
+
+	dev = kzalloc_obj(*dev);
+	if (!dev)
+		return NULL;
+	dev->type = type;
+	dev->signal = ATM_PHY_SIG_UNKNOWN;
+	dev->link_rate = ATM_OC3_PCR;
+
+	return dev;
+}
+
+static struct atm_dev *__atm_dev_lookup(int number)
+{
+	struct atm_dev *dev;
+
+	list_for_each_entry(dev, &atm_devs, dev_list) {
+		if (dev->number == number) {
+			atm_dev_hold(dev);
+			return dev;
+		}
+	}
+	return NULL;
+}
+
+struct atm_dev *atm_dev_lookup(int number)
+{
+	struct atm_dev *dev;
+
+	mutex_lock(&atm_dev_mutex);
+	dev = __atm_dev_lookup(number);
+	mutex_unlock(&atm_dev_mutex);
+	return dev;
+}
+EXPORT_SYMBOL(atm_dev_lookup);
+
+struct atm_dev *atm_dev_register(const char *type, struct device *parent,
+				 const struct atmdev_ops *ops, int number,
+				 unsigned long *flags)
+{
+	struct atm_dev *dev, *inuse;
+
+	dev = __alloc_atm_dev(type);
+	if (!dev) {
+		pr_err("no space for dev %s\n", type);
+		return NULL;
+	}
+	mutex_lock(&atm_dev_mutex);
+	if (number != -1) {
+		inuse = __atm_dev_lookup(number);
+		if (inuse) {
+			atm_dev_put(inuse);
+			mutex_unlock(&atm_dev_mutex);
+			kfree(dev);
+			return NULL;
+		}
+		dev->number = number;
+	} else {
+		dev->number = 0;
+		while ((inuse = __atm_dev_lookup(dev->number))) {
+			atm_dev_put(inuse);
+			dev->number++;
+		}
+	}
+
+	dev->ops = ops;
+	if (flags)
+		dev->flags = *flags;
+	else
+		memset(&dev->flags, 0, sizeof(dev->flags));
+	memset(&dev->stats, 0, sizeof(dev->stats));
+	refcount_set(&dev->refcnt, 1);
+
+	if (atm_proc_dev_register(dev) < 0) {
+		pr_err("atm_proc_dev_register failed for dev %s\n", type);
+		mutex_unlock(&atm_dev_mutex);
+		kfree(dev);
+		return NULL;
+	}
+
+	if (atm_register_sysfs(dev, parent) < 0) {
+		pr_err("atm_register_sysfs failed for dev %s\n", type);
+		atm_proc_dev_deregister(dev);
+		goto out_fail;
+	}
+
+	list_add_tail(&dev->dev_list, &atm_devs);
+
+out:
+	mutex_unlock(&atm_dev_mutex);
+	return dev;
+
+out_fail:
+	put_device(&dev->class_dev);
+	dev = NULL;
+	goto out;
+}
+EXPORT_SYMBOL(atm_dev_register);
+
+void atm_dev_deregister(struct atm_dev *dev)
+{
+	BUG_ON(test_bit(ATM_DF_REMOVED, &dev->flags));
+	set_bit(ATM_DF_REMOVED, &dev->flags);
+
+	/*
+	 * if we remove current device from atm_devs list, new device
+	 * with same number can appear, such we need deregister proc,
+	 * release async all vccs and remove them from vccs list too
+	 */
+	mutex_lock(&atm_dev_mutex);
+	list_del(&dev->dev_list);
+	atm_dev_release_vccs(dev);
+	atm_unregister_sysfs(dev);
+	atm_proc_dev_deregister(dev);
+	mutex_unlock(&atm_dev_mutex);
+
+	atm_dev_put(dev);
+}
+EXPORT_SYMBOL(atm_dev_deregister);
+
+static void copy_aal_stats(struct k_atm_aal_stats *from,
+    struct atm_aal_stats *to)
+{
+#define __HANDLE_ITEM(i) to->i = atomic_read(&from->i)
+	__AAL_STAT_ITEMS
+#undef __HANDLE_ITEM
+}
+
+static void subtract_aal_stats(struct k_atm_aal_stats *from,
+    struct atm_aal_stats *to)
+{
+#define __HANDLE_ITEM(i) atomic_sub(to->i, &from->i)
+	__AAL_STAT_ITEMS
+#undef __HANDLE_ITEM
+}
+
+static int fetch_stats(struct atm_dev *dev, struct atm_dev_stats __user *arg,
+		       int zero)
+{
+	struct atm_dev_stats tmp;
+	int error = 0;
+
+	copy_aal_stats(&dev->stats.aal0, &tmp.aal0);
+	copy_aal_stats(&dev->stats.aal34, &tmp.aal34);
+	copy_aal_stats(&dev->stats.aal5, &tmp.aal5);
+	if (arg)
+		error = copy_to_user(arg, &tmp, sizeof(tmp));
+	if (zero && !error) {
+		subtract_aal_stats(&dev->stats.aal0, &tmp.aal0);
+		subtract_aal_stats(&dev->stats.aal34, &tmp.aal34);
+		subtract_aal_stats(&dev->stats.aal5, &tmp.aal5);
+	}
+	return error ? -EFAULT : 0;
+}
+
+int atm_getnames(void __user *buf, int __user *iobuf_len)
+{
+	int error, len, size = 0;
+	struct atm_dev *dev;
+	struct list_head *p;
+	int *tmp_buf, *tmp_p;
+
+	if (get_user(len, iobuf_len))
+		return -EFAULT;
+	mutex_lock(&atm_dev_mutex);
+	list_for_each(p, &atm_devs)
+		size += sizeof(int);
+	if (size > len) {
+		mutex_unlock(&atm_dev_mutex);
+		return -E2BIG;
+	}
+	tmp_buf = kmalloc(size, GFP_ATOMIC);
+	if (!tmp_buf) {
+		mutex_unlock(&atm_dev_mutex);
+		return -ENOMEM;
+	}
+	tmp_p = tmp_buf;
+	list_for_each_entry(dev, &atm_devs, dev_list) {
+		*tmp_p++ = dev->number;
+	}
+	mutex_unlock(&atm_dev_mutex);
+	error = ((copy_to_user(buf, tmp_buf, size)) ||
+		 put_user(size, iobuf_len))
+		? -EFAULT : 0;
+	kfree(tmp_buf);
+	return error;
+}
+
+int atm_dev_ioctl(unsigned int cmd, void __user *buf, int __user *sioc_len,
+		  int number, int compat)
+{
+	struct atm_dev *dev;
+	int error, size = 0;
+
+	dev = try_then_request_module(atm_dev_lookup(number), "atm-device-%d",
+				      number);
+	if (!dev)
+		return -ENODEV;
+
+	switch (cmd) {
+	case ATM_GETTYPE:
+		size = strlen(dev->type) + 1;
+		if (copy_to_user(buf, dev->type, size)) {
+			error = -EFAULT;
+			goto done;
+		}
+		break;
+	case ATM_GETESI:
+		size = ESI_LEN;
+		if (copy_to_user(buf, dev->esi, size)) {
+			error = -EFAULT;
+			goto done;
+		}
+		break;
+	case ATM_SETESI:
+	{
+		int i;
+
+		for (i = 0; i < ESI_LEN; i++)
+			if (dev->esi[i]) {
+				error = -EEXIST;
+				goto done;
+			}
+	}
+		fallthrough;
+	case ATM_SETESIF:
+	{
+		unsigned char esi[ESI_LEN];
+
+		if (!capable(CAP_NET_ADMIN)) {
+			error = -EPERM;
+			goto done;
+		}
+		if (copy_from_user(esi, buf, ESI_LEN)) {
+			error = -EFAULT;
+			goto done;
+		}
+		memcpy(dev->esi, esi, ESI_LEN);
+		error =  ESI_LEN;
+		goto done;
+	}
+	case ATM_GETSTATZ:
+		if (!capable(CAP_NET_ADMIN)) {
+			error = -EPERM;
+			goto done;
+		}
+		fallthrough;
+	case ATM_GETSTAT:
+		size = sizeof(struct atm_dev_stats);
+		error = fetch_stats(dev, buf, cmd == ATM_GETSTATZ);
+		if (error)
+			goto done;
+		break;
+	case ATM_GETCIRANGE:
+		size = sizeof(struct atm_cirange);
+		if (copy_to_user(buf, &dev->ci_range, size)) {
+			error = -EFAULT;
+			goto done;
+		}
+		break;
+	case ATM_GETLINKRATE:
+		size = sizeof(int);
+		if (copy_to_user(buf, &dev->link_rate, size)) {
+			error = -EFAULT;
+			goto done;
+		}
+		break;
+	case ATM_SETLOOP:
+		if (__ATM_LM_XTRMT((int) (unsigned long) buf) &&
+		    __ATM_LM_XTLOC((int) (unsigned long) buf) >
+		    __ATM_LM_XTRMT((int) (unsigned long) buf)) {
+			error = -EINVAL;
+			goto done;
+		}
+		fallthrough;
+	case ATM_SETCIRANGE:
+		if (!capable(CAP_NET_ADMIN)) {
+			error = -EPERM;
+			goto done;
+		}
+		fallthrough;
+	default:
+		if (IS_ENABLED(CONFIG_COMPAT) && compat) {
+#ifdef CONFIG_COMPAT
+			if (!dev->ops->compat_ioctl) {
+				error = -EINVAL;
+				goto done;
+			}
+			size = dev->ops->compat_ioctl(dev, cmd, buf);
+#endif
+		} else {
+			if (!dev->ops->ioctl) {
+				error = -EINVAL;
+				goto done;
+			}
+			size = dev->ops->ioctl(dev, cmd, buf);
+		}
+		if (size < 0) {
+			error = (size == -ENOIOCTLCMD ? -ENOTTY : size);
+			goto done;
+		}
+	}
+
+	if (size)
+		error = put_user(size, sioc_len) ? -EFAULT : 0;
+	else
+		error = 0;
+done:
+	atm_dev_put(dev);
+	return error;
+}
+
+#ifdef CONFIG_PROC_FS
+void *atm_dev_seq_start(struct seq_file *seq, loff_t *pos)
+{
+	mutex_lock(&atm_dev_mutex);
+	return seq_list_start_head(&atm_devs, *pos);
+}
+
+void atm_dev_seq_stop(struct seq_file *seq, void *v)
+{
+	mutex_unlock(&atm_dev_mutex);
+}
+
+void *atm_dev_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+	return seq_list_next(v, &atm_devs, pos);
+}
+#endif