linux-kernel - [PATCH] platform/chrome: cross_ec_type: fix missing fwnode reference decrement

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-Id: <20241009-cross_ec_typec_fwnode_handle_put-v1-1-f17bdb48d780@gmail.com>
Date: Wed, 09 Oct 2024 20:55:44 +0200
From: Javier Carrasco <javier.carrasco.cruz@...il.com>
To: Prashant Malani <pmalani@...omium.org>, 
 Benson Leung <bleung@...omium.org>, Tzung-Bi Shih <tzungbi@...nel.org>, 
 Guenter Roeck <groeck@...omium.org>, 
 Heikki Krogerus <heikki.krogerus@...ux.intel.com>, 
 Enric Balletbo i Serra <eballetbo@...nel.org>
Cc: chrome-platform@...ts.linux.dev, linux-kernel@...r.kernel.org, 
 stable@...r.kernel.org, Javier Carrasco <javier.carrasco.cruz@...il.com>
Subject: [PATCH] platform/chrome: cross_ec_type: fix missing fwnode
 reference decrement

The device_for_each_child_node() macro requires explicit calls to
fwnode_handle_put() upon early exits (return, break, goto) to decrement
the fwnode's refcount, and avoid levaing a node reference behind.

Add the missing fwnode_handle_put() after the common label for all error
paths.

Cc: stable@...r.kernel.org
Fixes: fdc6b21e2444 ("platform/chrome: Add Type C connector class driver")
Signed-off-by: Javier Carrasco <javier.carrasco.cruz@...il.com>
---
I usually switch to the scoped variant of the macro to fix such issues,
but given that the fix is relevant for stable kernels, I have provided
the "classical" approach by adding the missing fwnode_handle_put().

If switching to the scoped variant is desired, please let me know.
This driver and cross_typec_switch could be easily converted.

By the way, I wonder why all error paths are redirected to the same
label to unregister ports, even before registering them (which seems to
be harmless because unregistered ports are ignored, but still). With this
fix, that jump to the label is definitely required, but if the scoped
variant is used, maybe some simple returns would be enough.
---
 drivers/platform/chrome/cros_ec_typec.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/platform/chrome/cros_ec_typec.c b/drivers/platform/chrome/cros_ec_typec.c
index c7781aea0b88..f1324466efac 100644
--- a/drivers/platform/chrome/cros_ec_typec.c
+++ b/drivers/platform/chrome/cros_ec_typec.c
@@ -409,6 +409,7 @@ static int cros_typec_init_ports(struct cros_typec_data *typec)
 	return 0;
 
 unregister_ports:
+	fwnode_handle_put(fwnode);
 	cros_unregister_ports(typec);
 	return ret;
 }

---
base-commit: b6270c3bca987530eafc6a15f9d54ecd0033e0e3
change-id: 20241009-cross_ec_typec_fwnode_handle_put-9f13b4bd467f

Best regards,
-- 
Javier Carrasco <javier.carrasco.cruz@...il.com>