From: Mohamad Raizudeen <raizudeen.kerneldev@gmail.com>
To: dmitry.torokhov@gmail.com
Cc: Mohamad Raizudeen <raizudeen.kerneldev@gmail.com>,
kees@kernel.org, linux-input@vger.kernel.org,
linux-kernel@vger.kernel.org
Subject: [PATCH] Input: serio - fix O(n^2) complexity in serio_unregister_driver()
Date: Wed, 8 Apr 2026 21:58:47 +0530 [thread overview]
Message-ID: <20260408162849.4639-1-raizudeen.kerneldev@gmail.com> (raw)
The current implementation restarts the scan from the beginning after
each disconnection(goto start_over) because serio_disconnect_port() may
delete the current list entry. This results in O(n^2) worst case
behaviour.
Replace with a two-phase approach:
1.Collect only top-level ports bound to the driver(skip those whose parent is also bound to the same driver) into a
temporary list.
2.Process each collected port once, moving it back to serio_list first to maintain invariants expected by serio_destroy_port().
This eliminates the restart loop, reduces complexity from O(n^2) to O(n)
and avoids use-after-free by never collecting child ports separately.
Signed-off-by: Mohamad Raizudeen <raizudeen.kerneldev@gmail.com>
---
drivers/input/serio/serio.c | 21 +++++++++++++--------
1 file changed, 13 insertions(+), 8 deletions(-)
diff --git a/drivers/input/serio/serio.c b/drivers/input/serio/serio.c
index 54dd26249b02..46b8faf1a46c 100644
--- a/drivers/input/serio/serio.c
+++ b/drivers/input/serio/serio.c
@@ -821,22 +821,27 @@ EXPORT_SYMBOL(__serio_register_driver);
void serio_unregister_driver(struct serio_driver *drv)
{
- struct serio *serio;
+ struct serio *serio, *next;
+ LIST_HEAD(disconnect_list);
+
guard(mutex)(&serio_mutex);
drv->manual_bind = true; /* so serio_find_driver ignores it */
serio_remove_pending_events(drv);
-start_over:
- list_for_each_entry(serio, &serio_list, node) {
- if (serio->drv == drv) {
- serio_disconnect_port(serio);
- serio_find_driver(serio);
- /* we could've deleted some ports, restart */
- goto start_over;
+ /*Collect all ports bound to this driver first*/
+ list_for_each_entry_safe(serio, next, &serio_list, node) {
+ if (serio->drv == drv && !(serio->parent && serio->parent->drv == drv)) {
+ list_move_tail(&serio->node, &disconnect_list);
}
}
+
+ list_for_each_entry_safe(serio, next, &disconnect_list, node) {
+ list_move_tail(&serio->node, &serio_list);
+ serio_disconnect_port(serio);
+ serio_find_driver(serio);
+ }
driver_unregister(&drv->driver);
}
--
2.43.0
next reply other threads:[~2026-04-08 16:29 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-04-08 16:28 Mohamad Raizudeen [this message]
2026-04-08 17:07 ` [PATCH] Input: serio - fix O(n^2) complexity in serio_unregister_driver() Dmitry Torokhov
[not found] ` <CABkOv5cU8D8oo8scTOojvJ8hUeq4nxm6GGuA3GaBvebzdMBYpg@mail.gmail.com>
2026-04-08 17:56 ` Dmitry Torokhov
2026-04-08 18:19 ` Mohamad Raizudeen
2026-04-08 18:36 ` Dmitry Torokhov
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260408162849.4639-1-raizudeen.kerneldev@gmail.com \
--to=raizudeen.kerneldev@gmail.com \
--cc=dmitry.torokhov@gmail.com \
--cc=kees@kernel.org \
--cc=linux-input@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox