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, 206 insertions, 0 deletions

diff --git a/drivers/md/dm-uevent.c b/drivers/md/dm-uevent.c
new file mode 100644
index 000000000..491738263
--- /dev/null
+++ b/drivers/md/dm-uevent.c
@@ -0,0 +1,206 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Device Mapper Uevent Support (dm-uevent)
+ *
+ * Copyright IBM Corporation, 2007
+ *	Author: Mike Anderson <andmike@linux.vnet.ibm.com>
+ */
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <linux/kobject.h>
+#include <linux/dm-ioctl.h>
+#include <linux/export.h>
+
+#include "dm.h"
+#include "dm-uevent.h"
+
+#define DM_MSG_PREFIX "uevent"
+
+static const struct {
+	enum dm_uevent_type type;
+	enum kobject_action action;
+	char *name;
+} _dm_uevent_type_names[] = {
+	{DM_UEVENT_PATH_FAILED, KOBJ_CHANGE, "PATH_FAILED"},
+	{DM_UEVENT_PATH_REINSTATED, KOBJ_CHANGE, "PATH_REINSTATED"},
+};
+
+static struct kmem_cache *_dm_event_cache;
+
+struct dm_uevent {
+	struct mapped_device *md;
+	enum kobject_action action;
+	struct kobj_uevent_env ku_env;
+	struct list_head elist;
+	char name[DM_NAME_LEN];
+	char uuid[DM_UUID_LEN];
+};
+
+static void dm_uevent_free(struct dm_uevent *event)
+{
+	kmem_cache_free(_dm_event_cache, event);
+}
+
+static struct dm_uevent *dm_uevent_alloc(struct mapped_device *md)
+{
+	struct dm_uevent *event;
+
+	event = kmem_cache_zalloc(_dm_event_cache, GFP_ATOMIC);
+	if (!event)
+		return NULL;
+
+	INIT_LIST_HEAD(&event->elist);
+	event->md = md;
+
+	return event;
+}
+
+static struct dm_uevent *dm_build_path_uevent(struct mapped_device *md,
+					      struct dm_target *ti,
+					      enum kobject_action action,
+					      const char *dm_action,
+					      const char *path,
+					      unsigned int nr_valid_paths)
+{
+	struct dm_uevent *event;
+
+	event = dm_uevent_alloc(md);
+	if (!event) {
+		DMERR("%s: dm_uevent_alloc() failed", __func__);
+		goto err_nomem;
+	}
+
+	event->action = action;
+
+	if (add_uevent_var(&event->ku_env, "DM_TARGET=%s", ti->type->name)) {
+		DMERR("%s: add_uevent_var() for DM_TARGET failed",
+		      __func__);
+		goto err_add;
+	}
+
+	if (add_uevent_var(&event->ku_env, "DM_ACTION=%s", dm_action)) {
+		DMERR("%s: add_uevent_var() for DM_ACTION failed",
+		      __func__);
+		goto err_add;
+	}
+
+	if (add_uevent_var(&event->ku_env, "DM_SEQNUM=%u",
+			   dm_next_uevent_seq(md))) {
+		DMERR("%s: add_uevent_var() for DM_SEQNUM failed",
+		      __func__);
+		goto err_add;
+	}
+
+	if (add_uevent_var(&event->ku_env, "DM_PATH=%s", path)) {
+		DMERR("%s: add_uevent_var() for DM_PATH failed", __func__);
+		goto err_add;
+	}
+
+	if (add_uevent_var(&event->ku_env, "DM_NR_VALID_PATHS=%d",
+			   nr_valid_paths)) {
+		DMERR("%s: add_uevent_var() for DM_NR_VALID_PATHS failed",
+		      __func__);
+		goto err_add;
+	}
+
+	return event;
+
+err_add:
+	dm_uevent_free(event);
+err_nomem:
+	return ERR_PTR(-ENOMEM);
+}
+
+/**
+ * dm_send_uevents - send uevents for given list
+ *
+ * @events:	list of events to send
+ * @kobj:	kobject generating event
+ *
+ */
+void dm_send_uevents(struct list_head *events, struct kobject *kobj)
+{
+	int r;
+	struct dm_uevent *event, *next;
+
+	list_for_each_entry_safe(event, next, events, elist) {
+		list_del_init(&event->elist);
+
+		/*
+		 * When a device is being removed this copy fails and we
+		 * discard these unsent events.
+		 */
+		if (dm_copy_name_and_uuid(event->md, event->name,
+					  event->uuid)) {
+			DMINFO("%s: skipping sending uevent for lost device",
+			       __func__);
+			goto uevent_free;
+		}
+
+		if (add_uevent_var(&event->ku_env, "DM_NAME=%s", event->name)) {
+			DMERR("%s: add_uevent_var() for DM_NAME failed",
+			      __func__);
+			goto uevent_free;
+		}
+
+		if (add_uevent_var(&event->ku_env, "DM_UUID=%s", event->uuid)) {
+			DMERR("%s: add_uevent_var() for DM_UUID failed",
+			      __func__);
+			goto uevent_free;
+		}
+
+		r = kobject_uevent_env(kobj, event->action, event->ku_env.envp);
+		if (r)
+			DMERR("%s: kobject_uevent_env failed", __func__);
+uevent_free:
+		dm_uevent_free(event);
+	}
+}
+EXPORT_SYMBOL_GPL(dm_send_uevents);
+
+/**
+ * dm_path_uevent - called to create a new path event and queue it
+ *
+ * @event_type:	path event type enum
+ * @ti:			pointer to a dm_target
+ * @path:		string containing pathname
+ * @nr_valid_paths:	number of valid paths remaining
+ *
+ */
+void dm_path_uevent(enum dm_uevent_type event_type, struct dm_target *ti,
+		   const char *path, unsigned int nr_valid_paths)
+{
+	struct mapped_device *md = dm_table_get_md(ti->table);
+	struct dm_uevent *event;
+
+	if (event_type >= ARRAY_SIZE(_dm_uevent_type_names)) {
+		DMERR("%s: Invalid event_type %d", __func__, event_type);
+		return;
+	}
+
+	event = dm_build_path_uevent(md, ti,
+				     _dm_uevent_type_names[event_type].action,
+				     _dm_uevent_type_names[event_type].name,
+				     path, nr_valid_paths);
+	if (IS_ERR(event))
+		return;
+
+	dm_uevent_add(md, &event->elist);
+}
+EXPORT_SYMBOL_GPL(dm_path_uevent);
+
+int dm_uevent_init(void)
+{
+	_dm_event_cache = KMEM_CACHE(dm_uevent, 0);
+	if (!_dm_event_cache)
+		return -ENOMEM;
+
+	DMINFO("version 1.0.3");
+
+	return 0;
+}
+
+void dm_uevent_exit(void)
+{
+	kmem_cache_destroy(_dm_event_cache);
+}