]> git.ipfire.org Git - thirdparty/kernel/stable.git/commit
i2c: core: Fix double-free of fwnode in i2c_unregister_device()
authorHans de Goede <hansg@kernel.org>
Sat, 19 Jul 2025 18:01:04 +0000 (20:01 +0200)
committerWolfram Sang <wsa+renesas@sang-engineering.com>
Mon, 28 Jul 2025 08:41:36 +0000 (10:41 +0200)
commit1c24e5fc0c7096e00c202a6a3e0c342c1afb47c2
tree8cafde9f7e7af7a5f3a7dc47649c4a83da81d508
parentf61389a9cd26b424485acade726ccfff96c749de
i2c: core: Fix double-free of fwnode in i2c_unregister_device()

Before commit df6d7277e552 ("i2c: core: Do not dereference fwnode in struct
device"), i2c_unregister_device() only called fwnode_handle_put() on
of_node-s in the form of calling of_node_put(client->dev.of_node).

But after this commit the i2c_client's fwnode now unconditionally gets
fwnode_handle_put() on it.

When the i2c_client has no primary (ACPI / OF) fwnode but it does have
a software fwnode, the software-node will be the primary node and
fwnode_handle_put() will put() it.

But for the software fwnode device_remove_software_node() will also put()
it leading to a double free:

[   82.665598] ------------[ cut here ]------------
[   82.665609] refcount_t: underflow; use-after-free.
[   82.665808] WARNING: CPU: 3 PID: 1502 at lib/refcount.c:28 refcount_warn_saturate+0xba/0x11
...
[   82.666830] RIP: 0010:refcount_warn_saturate+0xba/0x110
...
[   82.666962]  <TASK>
[   82.666971]  i2c_unregister_device+0x60/0x90

Fix this by not calling fwnode_handle_put() when the primary fwnode is
a software-node.

Fixes: df6d7277e552 ("i2c: core: Do not dereference fwnode in struct device")
Cc: stable@vger.kernel.org
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
drivers/i2c/i2c-core-base.c