Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <b11141f6-95b7-883e-d924-b9b2699eb980@kernel.org>
Date:   Sat, 17 Sep 2022 08:46:42 -0600
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:91.0)
 Gecko/20100101 Thunderbird/91.13.0
Subject: Re: [PATCH v1] net/ipv4/nexthop: check the return value of
 nexthop_find_by_id()
Content-Language: en-US
To:     Nikolay Aleksandrov <razor@blackwall.org>,
        Li Zhong <floridsleeves@gmail.com>,
        linux-kernel@vger.kernel.org, netdev@vger.kernel.org
Cc:     pabeni@redhat.com, kuba@kernel.org, edumazet@google.com,
        davem@davemloft.net, yoshfuji@linux-ipv6.org
References: <20220917023020.3845137-1-floridsleeves@gmail.com>
 <9974177e-7067-aacd-1c53-7e82616f3c3f@blackwall.org>
From:   David Ahern <dsahern@kernel.org>
In-Reply-To: <9974177e-7067-aacd-1c53-7e82616f3c3f@blackwall.org>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 9/17/22 2:29 AM, Nikolay Aleksandrov wrote:
> On 17/09/2022 05:30, Li Zhong wrote:
>> Check the return value of nexthop_find_by_id(), which could be NULL on
>> when not found. So we check to avoid null pointer dereference.
>>
>> Signed-off-by: Li Zhong <floridsleeves@gmail.com>
>> ---
>>  net/ipv4/nexthop.c | 4 ++++
>>  1 file changed, 4 insertions(+)
>>
>> diff --git a/net/ipv4/nexthop.c b/net/ipv4/nexthop.c
>> index 853a75a8fbaf..9f91bb78eed5 100644
>> --- a/net/ipv4/nexthop.c
>> +++ b/net/ipv4/nexthop.c
>> @@ -2445,6 +2445,10 @@ static struct nexthop *nexthop_create_group(struct net *net,
>>  		struct nh_info *nhi;
>>  
>>  		nhe = nexthop_find_by_id(net, entry[i].id);
>> +		if (!nhe) {
>> +			err = -EINVAL;
>> +			goto out_no_nh;
>> +		}
>>  		if (!nexthop_get(nhe)) {
>>  			err = -ENOENT;
>>  			goto out_no_nh;
> 
> These are validated in nh_check_attr_group() and should exist at this point.
> Since remove_nexthop() should run under rtnl I don't see a way for a nexthop
> to disappear after nh_check_attr_group() and before nexthop_create_group().
> 

exactly. That lookup can't fail because the ids have been validated and
all of this is under rtnl preventing nexthop removes.