netdev - Re: [PATCH v2] iproute2: prevent memory leak

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

Message-ID: <20231115033248.GC21242@breakpoint.cc>
Date: Wed, 15 Nov 2023 04:32:48 +0100
From: Florian Westphal <fw@...len.de>
To: heminhong <heminhong@...inos.cn>
Cc: stephen@...workplumber.org, netdev@...r.kernel.org
Subject: Re: [PATCH v2] iproute2: prevent memory leak

heminhong <heminhong@...inos.cn> wrote:
> When the return value of rtnl_talk() is less than 0, 'answer' does not
> need to release. When the return value of rtnl_talk() is greater than
> or equal to 0, 'answer' will be allocated, if subsequent processing fails,
> the memory should be free, otherwise it will cause memory leak.

I don't understand this patch.  Care to elaborate where a memory leak is?

rtnl_talk -> __rtnl_talk -> __rtnl_talk_iov

 998 static int __rtnl_talk_iov(struct rtnl_handle *rtnl, struct iovec *iov,
 999                            size_t iovlen, struct nlmsghdr **answer,
1000                            bool show_rtnl_err, nl_ext_ack_fn_t errfn)
[..]
1102                                 if (answer)
1103                                         *answer = (struct nlmsghdr *)buf;
1104                                 else
1105                                         free(buf);
1106                                 return 0;
1107                         }
1108
1109                         if (answer) {
1110                                 *answer = (struct nlmsghdr *)buf;
1111                                 return 0;
1112                         }

I see no other spots where 'answer' is set, i.e. assignment ONLY on
'return 0'.

> Signed-off-by: heminhong <heminhong@...inos.cn>
> ---
>  ip/link_gre.c    | 2 ++
>  ip/link_gre6.c   | 2 ++
>  ip/link_ip6tnl.c | 2 ++
>  ip/link_iptnl.c  | 2 ++
>  ip/link_vti.c    | 2 ++
>  ip/link_vti6.c   | 2 ++
>  6 files changed, 12 insertions(+)
> 
> diff --git a/ip/link_gre.c b/ip/link_gre.c
> index 74a5b5e9..b86ec22d 100644
> --- a/ip/link_gre.c
> +++ b/ip/link_gre.c
> @@ -111,6 +111,8 @@ static int gre_parse_opt(struct link_util *lu, int argc, char **argv,
>  
>  		if (rtnl_talk(&rth, &req.n, &answer) < 0) {
>  get_failed:
> +			if (answer)
> +				free(answer);

This if() is not needed.  free(NULL) is fine.  But in this case,
'answer' can contain stack-garbage, as this variable isn't initialised
to NULL.

Moreover, the placement would need to be ABOVE the label, not below.

But, see above, I don't see a 'answer' related memleak.