[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Sat, 9 Sep 2023 00:05:20 -0700
From: Kuniyuki Iwashima <kuniyu@...zon.com>
To: <avagin@...il.com>
CC: <davem@...emloft.net>, <edumazet@...gle.com>, <glider@...gle.com>,
<joannelkoong@...il.com>, <kafai@...com>, <kernel-team@...com>,
<kuba@...nel.org>, <martin.lau@...nel.org>, <netdev@...r.kernel.org>,
<pabeni@...hat.com>, <kuniyu@...zon.com>
Subject: Re: [PATCH net-next] net: Fix incorrect address comparison when searching for a bind2 bucket
From: Andrei Vagin <avagin@...il.com>
Date: Fri, 8 Sep 2023 14:54:12 -0700
> On Mon, Sep 26, 2022 at 05:25:44PM -0700, Martin KaFai Lau wrote:
> > From: Martin KaFai Lau <martin.lau@...nel.org>
> >
> > The v6_rcv_saddr and rcv_saddr are inside a union in the
> > 'struct inet_bind2_bucket'. When searching a bucket by following the
> > bhash2 hashtable chain, eg. inet_bind2_bucket_match, it is only using
> > the sk->sk_family and there is no way to check if the inet_bind2_bucket
> > has a v6 or v4 address in the union. This leads to an uninit-value
> > KMSAN report in [0] and also potentially incorrect matches.
> >
> > This patch fixes it by adding a family member to the inet_bind2_bucket
> > and then tests 'sk->sk_family != tb->family' before matching
> > the sk's address to the tb's address.
>
> It seems this patch doesn't handle v4mapped addresses properly. One of
> gVisor test started failing with this change:
>
> socket(AF_INET6, SOCK_STREAM, IPPROTO_IP) = 3
> bind(3, {sa_family=AF_INET6, sin6_port=htons(0), sin6_flowinfo=htonl(0), inet_pton(AF_INET6, "::ffff:0.0.0.0", &sin6_addr), sin6_scope_id=0}, 28) = 0
> getsockname(3, {sa_family=AF_INET6, sin6_port=htons(33789), sin6_flowinfo=htonl(0), inet_pton(AF_INET6, "::ffff:0.0.0.0", &sin6_addr), sin6_scope_id=0}, [28]) = 0
> socket(AF_INET6, SOCK_STREAM, IPPROTO_IP) = 4
> bind(4, {sa_family=AF_INET6, sin6_port=htons(33789), sin6_flowinfo=htonl(0), inet_pton(AF_INET6, "::1", &sin6_addr), sin6_scope_id=0}, 28) = 0
> socket(AF_INET, SOCK_STREAM, IPPROTO_IP) = 5
> bind(5, {sa_family=AF_INET, sin_port=htons(33789), sin_addr=inet_addr("127.0.0.1")}, 16) = 0
>
> The test expects that the second bind returns EADDRINUSE.
Thanks for the report.
inet_bind2_bucket_match_addr_any() forgot to take care of
IPV6_ADDR_MAPPED inaddr_any case.
This change fixes the regression. I'll post a patch after
checking other two functions that the commit touched.
---8<---
diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index 7876b7d703cb..6f2a8dba24fe 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -837,7 +837,9 @@ bool inet_bind2_bucket_match_addr_any(const struct inet_bind2_bucket *tb, const
if (sk->sk_family == AF_INET)
return net_eq(ib2_net(tb), net) && tb->port == port &&
tb->l3mdev == l3mdev &&
- ipv6_addr_any(&tb->v6_rcv_saddr);
+ (ipv6_addr_any(&tb->v6_rcv_saddr) ||
+ (ipv6_addr_type(&tb->v6_rcv_saddr) == IPV6_ADDR_MAPPED &&
+ tb->v6_rcv_saddr.s6_addr32[3] == 0));
return false;
}
---8<---
---8<---
[root@...alhost ~]# python3
>>> from socket import *
>>>
>>> s1 = socket(AF_INET6, SOCK_STREAM)
>>> s1.bind(('::ffff:0.0.0.0', 0))
>>> port = s1.getsockname()[1]
>>>
>>> s2 = socket(AF_INET, SOCK_STREAM)
>>> s2.bind(('127.0.0.1', port))
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
OSError: [Errno 98] Address already in use
---8<---
Powered by blists - more mailing lists