[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <7a0756a3-527e-482f-987e-dde3f5fc09d5@yandex.ru>
Date: Tue, 12 Nov 2024 08:47:08 +0300
From: Dmitry Antipov <dmantipov@...dex.ru>
To: Miquel Raynal <miquel.raynal@...tlin.com>
Cc: Alexander Aring <alex.aring@...il.com>,
Stefan Schmidt <stefan@...enfreihafen.org>, linux-wpan@...r.kernel.org,
netdev@...r.kernel.org, lvc-project@...uxtesting.org,
syzbot+985f827280dc3a6e7e92@...kaller.appspotmail.com
Subject: Re: [PATCH] mac802154: fix interface deletion
On 11/11/24 10:41 PM, Miquel Raynal wrote:
> Why not just enclose this list_del() within a mutex_lock(iflist_mtx)
> like the others? Would probably make more sense and prevent the use of
> yet another protection mechanism? Is there anything preventing the use
> of this mutex here?
IIUC this will not work because 'ieee802154_if_remove()' may be called for
'sdata' which was previously removed via 'ieee802154_remove_interfaces()'.
After the latter, 'sdata->list' is undefined (or poisoned if CONFIG_DEBUG_LIST
is enabled), so re-entering 'list_del(&sdata->list)' in the former is a bug.
Dmitry
Powered by blists - more mailing lists