[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20240208085617.1fc58a59@hermes.local>
Date: Thu, 8 Feb 2024 08:56:17 -0800
From: Stephen Hemminger <stephen@...workplumber.org>
To: Maks Mishin <maks.mishinfz@...il.com>
Cc: netdev@...r.kernel.org
Subject: Re: [PATCH] ipaddrlabel: Fix descriptor leak in flush_addrlabel()
On Wed, 7 Feb 2024 23:25:42 +0300
Maks Mishin <maks.mishinfz@...il.com> wrote:
> Added closure of descriptor `rth2.fd` created by rtnl_open() when
> returning from function.
>
> Signed-off-by: Maks Mishin <maks.mishinFZ@...il.com>
> ---
> ip/ipaddrlabel.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/ip/ipaddrlabel.c b/ip/ipaddrlabel.c
> index b045827a..3857fcf5 100644
> --- a/ip/ipaddrlabel.c
> +++ b/ip/ipaddrlabel.c
> @@ -201,8 +201,10 @@ static int flush_addrlabel(struct nlmsghdr *n, void *arg)
> n->nlmsg_type = RTM_DELADDRLABEL;
> n->nlmsg_flags = NLM_F_REQUEST;
>
> - if (rtnl_open(&rth2, 0) < 0)
> + if (rtnl_open(&rth2, 0) < 0) {
> + rtnl_close(&rth2);
> return -1;
> + }
>
> if (rtnl_talk(&rth2, n, NULL) < 0)
> return -2;
This doesn't make sense, the error path in rtnl_open cleans up
itself. Do you have a reproducer or is this just some static analyzer?
Powered by blists - more mailing lists